1 CHARACTERIZATION OF THE MOLECULAR DETERMINANTS FOR CLASS A G PROTEIN COUPLED RECEPTOR LIGAND BINDING AND FUNCTION: DRUG DISCOVERY TARGETING THE HISTAMINE H1RECEPTOR By SEAN MICHAEL TRAVERS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 1
2 201 1 Sean Michael Travers
3 ACKNOWLEDGMENTS To all of my family, friends, colleagues, teachers, professors, my committee members, and my advisor, who have helped me reach this point, I would not be here without each and every one of you. In addition, I would like to acknowledge Dr. Raymond Booth as he has guided me through this process collectively known as graduate school. I would also like to thank Doctors Margaret James, Ken Sloan, and JoannaPeris for their input and expertise while serving on my committee; and Dr. Henry Vischer for hi s efforts with our BRET studies as well as all of our post docs past and present. My thanks also go out to the late Dr. Eugene Gooch of Elon University for his unparalleled teaching ability and for instilling a love of chemistry in me. Last and definitely not least, I would like to thank my parents, Bill and Linda, and my brother Garrett for their unwavering support; and I would be remiss if I did not thank my wonderful girlfriend Laura for putting up with me and supporting me throughout this lon g, strange trip known as graduate school
4 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 3 LIST OF TABLES ............................................................................................................ 6 LIST OF FIGURES .......................................................................................................... 7 LIST OF ABBREVIATIONS ............................................................................................. 9 ABSTRACT ................................................................................................................... 12 Chapter 1 G PROTEIN COUPLED RECEPTORS: ENDOGENOUS AND THERAPEUTIC SIGNIFICANCE OF HISTAMINE RECEPTORS ..................................................... 14 Introduction to G Protein Coupled Receptors (GPCRs) .......................................... 14 Ballesteros Numbering System ............................................................................... 17 Usin g Ballesteros Numbering to Identify Critical Aminergic Residues .................... 20 Histamine Receptors ............................................................................................... 24 Histamine Biosynthesis and Catabolism .......................................................... 27 Histamine in the Central Nervous System ........................................................ 28 Therapeutic Use of Histamine Ligands ................................................................... 36 Previously Established Molecular Determinants for H1 Receptors .......................... 38 H1 GPCR Dimerization ............................................................................................ 45 Current GPCR Theory ............................................................................................ 47 Law of Mass Action and Early GPCR Models .................................................. 48 Ternary Complex Model ................................................................................... 49 Extended Ternary Complex Model ................................................................... 49 LITicon .............................................................................................................. 50 Goal of This Thesis ................................................................................................. 56 Aim #1: Char acterization of the Structure Activity Relationships for PAT Analogue Binding at the Wild Type H uman Histamine H1R eceptor (HH1R) .. 57 Aim #2: Characterization of the Role of Amino Acid Residue Y5.48 in Dimerization of the HH1R and its Impact on Ligand Binding and Function. ... 57 Aim #3: Characterization of the Role of HH1R R esidue Y7.53A in Ligand Binding and Activation Using Functionally Selective Phenyl aminotetralin (PAT) Derivatives .......................................................................................... 5 8 2 CHARACTERIZATION OF THE STRUCTURE ACTIVITY RELATIONSHIPS FOR PAT ANALOGUES AT THE WILD TYPE HH1R ............................................. 66 Rationale for Undertaking These Studies ............................................................... 66 Materials and Methods ............................................................................................ 66 Chemicals ......................................................................................................... 66
5 Cell Culture and Transfection ........................................................................... 67 HH1R Binding Assays ....................................................................................... 68 Binding Assay Results and Discussion ................................................................... 69 Unsubstituted Parent PAT Compounds ............................................................ 70 Ortho Substituted PATs .................................................................................... 72 Meta Substituted PATs ..................................................................................... 73 Para Subst ituted PATs ..................................................................................... 75 N substituted PAT Derivatives .......................................................................... 81 Chloro and Hydroxyl Tetrahydronaphthalene Substituted PATs ...................... 85 PAT Like Compounds with Ring Perturbations ................................................ 86 PAT Binding Summary ............................................................................................ 91 3 INVOLVEMENT OF AMINO ACID RESIDUE Y5.48A IN THE DIMERIZATION/FUNCTIONAL STABILIZATION PROCESSES OF THE HUMAN HISTAMINE H1RECEPTOR ................................................................... 104 Literature Review of Ballesteros Position 5.48 ...................................................... 104 Materials and Methods .......................................................................................... 106 Experimental Results and Discussion ................................................................... 108 Saturation Binding Assay Results .................................................................. 108 Competitive Binding Experiments ................................................................... 111 Inositol Phosphate Production Mediated via G ............................................ 114 B ioluminescence R esonanceE nergy T ransfer (BRET) Studies Comparing dimerization of the WT and Y5.48A H1 Receptors ...................................... 118 Summary .............................................................................................................. 120 4 PROBING THE ROLE OF RESIDUE Y7.53A IN THE LIGAND BINDING AND ACTIVATION OF THE HH1R ................................................................................ 126 Rationale for Undertaking These Studies and Literature Review .......................... 126 Materials and Methods .......................................................................................... 127 Experimental Results and Discussion ................................................................... 128 Competitive Binding Studies at Y7.53A .......................................................... 128 Inositol Phosphate Production Mediated via G ............................................ 129 A denylyl C yclase Acti vity/ Cyclic A denosine M ono phosphate Formation Mediated via G S ......................................................................................... 132 Summary .............................................................................................................. 132 5 CONCLUDING REMARKS ................................................................................... 137 LIST OF REFERENCES ............................................................................................. 141 BIOGRAPHICAL SKETCH .......................................................................................... 145
6 LIST OF TABLES Table page 1 1 KI and BMax values for 3H mepyramine and ( ) trans P henylaminotetralin (P AT ) for WT and D3.32A/F6.52A H1 receptors. ................................................ 65 2 1 Affinities of the PAT stereoisomers at the WTHH1R ........................................... 93 2 2 H uman H istamine H1R eceptor (HH1R) binding affinities of (+) and ( ) trans parasubstituted PATs ........................................................................................ 95 2 3 WT HH1R binding affinities of (+) and ( ) trans parasubstituted PATs ............. 96 2 4 Histamine H1 binding affinities of N alkyl substituted PAT analogues ................ 99 2 5 Affinities of diOH and Cl OH PAT analogues at the WT HH1R. ....................... 100 2 6 Affinities of PAT analogues with modified tricyclic ring systems ....................... 103 3 1 Compiled KD and BMax values for 3H mepyramine and ( ) trans PAT at the WT and Y5.48A HH1R. ..................................................................................... 125 3 2 Tabulated KI values for the HH1R ligands examined in this section at both the WT and Y5.48A point mutated receptors. ........................................................ 125 3 3 Displays the correlated functional values for the ligands examined in the phospholipase C (PLC) functional assay at the WT and Y5.48A HH1R. ........... 125 4 1 Compiled affinities of various H1 ligands at the WT and Y7.53A receptors ...... 134 4 2 Tabulates the EC/IC50 and E/Imax values of various ligands at the WT and Y7.53A point mutated receptors ....................................................................... 136
7 LIST OF FIGURES Figure page 1 1 Generic structure of a G P rotein C oupled R eceptor (GPCR) ............................ 59 1 2 GPCR activation demonstrating guanine exchange. .......................................... 59 1 3 Ballesteros numbering proceeds as indicated by the arrowson thegeneric GPCR structure .................................................................................................. 60 1 4 Phylogenic tree for aminergic GPCRs that are closely related to the H1 receptor. ............................................................................................................. 60 1 5 Representation of both signaling pathways for the Human Histamine H1 Receptor (HH1R) ............................................................................................... 61 1 6 Histamine catabolism in the central nervous system ........................................ 62 1 7 The chemical structures of first generation andsecondgeneration (bottom antihistamines. .................................................................................................... 62 1 8 Proposed models of GPCR dimerization ............................................................ 63 1 9 Possible combinations of contact and domainswapped dimers for D3.32A and F6.52A mutations. ....................................................................................... 63 1 10 Structures of the 2adrenergic receptor ligands that were studied using LITicon. ............................................................................................................... 64 1 11 Illustrates the law of mass action ........................................................................ 64 1 12 Indicates the changes that were made to the law of mass action in order to incorporate G proteins. ....................................................................................... 64 1 13 Demonstrates the ternary complex model of G protein activation ...................... 64 1 14 E xtended ternary complex model ....................................................................... 65 2 1 P henylaminotetralin (P AT ) family of stereoisomers demonstrating the stereochemical relationships between each compound. .................................... 93 2 2 Representative competitive binding curves for the PAT family of stereoisomers at the human WT H1 receptor ...................................................... 94 2 3 Enantiomers of ( ) cetirizine that demonstrate high stereoselectivityat the HH1R .................................................................................................................. 94 2 4 Structures of ( ) trans orthoClPAT ( ) trans orthoCH3PAT ........................... 95
8 2 5 Structures of ( ) trans meta halogenatedPAT derivatives ................................. 95 2 6 Structures of the ( ) trans parasubstituted PATs ............................................... 96 2 7 Representative HH1Rcompetitive binding curves for (+) and ( ) trans parasubstituted PATs. ............................................................................................... 97 2 8 Structures of N alkylatedPAT derivatives. ......................................................... 98 2 9 Proposed interaction o f aspartic acid D3.32 with ( ) and (+) trans N,Ndiethyl PAT that may account for higher affinity of the ( ) enantiomer ........................... 99 2 10 Structures of racemic di OH and Cl OH PAT ligands ......................................... 99 2 11 Representative competitive binding curves for di OH and ClOH PAT at the WT HH1R. ......................................................................................................... 100 2 12 Proposed H1 pharmacophore and the ligands that are derived from it.. ........... 101 2 13 Structures for the ring substituted PAT derivatives ........................................... 102 3 1 Saturation binding isotherms for 3H mepyramine at the WT and Y5.48A HH1R ................................................................................................................ 121 3 2 Saturation isotherms for of 3H ( ) trans PAT labeled WT and Y5.48A HH1 receptors .......................................................................................................... 122 3 3 Structures of ligands examined in this section: ................................................. 123 3 4 Competitive binding curves for ( ) trans PAT at the WT and Y5.48A HH1R ..... 124 3 5 Ligand mediated P hospholipase C (PLC) functional activity at WT vs. Y5.48A HH1 receptors. ................................................................................................. 124 4 1 GPCR snake diagram for the HH1R illustrating the location of Y7.53 towards the intracellular side of TMD 7. ......................................................................... 134 4 2 Representative competition binding curves for various H1ligandsat the Y7.53A point mutated receptor: ........................................................................ 135 4 3 Representative PLC functional curves for H1 ligands at the Y7.53A HH1R. ..... 135 4 4 Representative A denylyl C yclase functional curves for histamine stimulationat WT and Y7.53A HH1R. ................................................................ 136
9 LIST OF ABBREVIATION S 5HT serotonin, usually accompanied by a receptor subtype as in 5HT2A AA arachidonic acid AC adenylyl cyclase AD Alzheimers disease AR beta2 adrenergic receptor BRET50 concentration of cDNA, at which 50% of the maximum BRET signal is reached BRET bioluminescence resonance energy tr ansfer BRETMax maximum BRET signal obtained cAMP cyclic 3, 5 adenosinemonophosphate CAT cylclohexylaminotetralin CDNA cloned deoxyribonucleic acid CNS central nervous system COAT cyclooctylaminotetralin CSF cerebrospinal fluid DAG diacylglycerols DOI 2, 5 dimethoxy 4 iodoamphetamine DNA deoxyribonucleic acid EC/IC50 effective concentration or inhibitory concentration of a ligand that produces 50% of the maximal response ECL(s) extracellular loop E/IMax maximum stimulation (E) or inhibition (I) over percent basal G acronym for a particular subunit of a G protein, where X is Q or S GC/MS gas chromatography/mass spectrometry GABA aminobutyric acid
10 GDP guanine diphosphate GI gastrointestinal GPCR G Protein Coupled Receptor GTP guanine tr iphosphate HDC histidine decarboxylase HH1R Human Histamine H1 Receptor HX is used to indicate other histamine receptors, as in H2receptor. HNMT human N methyl transferase HPLC high performance liquid chromatography IP or IP3 Inositol Phosphates or Inositol Triphosphates ICL(s) intracellular loop(s) IL ionic lock KD affinity value of a radioligand for its receptor KI affinity value of a ligand, determined by the displacement of a radiolabeled compound from the receptor MAO B mono amine oxidase B MEM Eagles minimum essential medium PAB cis 5 phenyl 7 dimethylamino5,6,7, 8 tetrahydro9 H benzocycloheptane PA T phenylaminotetralin PD Parkinsons disease PET positron emission tomography PL Phospholipase, usually listed with a subtype as in PLC, equating to phospholipase C PNS peripheral nervous system PTZ pentylenetetrazol R Luc Renillareniformis luciferase (an enzyme)
11 RTS rotamer toggle switch SAR structure activity relationships TH tyrosine hydroxylase S.E.M. standard error of the mean SPA scintillation proximity assay TMD(s ) transmembrane domain(s) TMN tuberomam millary nucleus UV VIS ultraviolet visible spectrometry VLPO ventrolateralpreoptic area WT wild type YFP yellow fluorescent protein
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CHARACTERIZATION OF THE MOLECULAR DETERMINANTS FOR CLASS A G PROTEIN COUPLED RECEPTOR LIGAND BINDING AND FUNCTION: DRUG DISCOVERY TARGETING THE HISTAMINE H1 RECEPTOR By Sean Michael Travers August 2011 Chair: Raymond Booth Major: Pharmaceutical Sci ences This PhD thesis research project focuses on drug discovery targeting the human histamine H1 ( HH1R ) G p roteinc oupled r eceptor (GPCR) Certain novel phenylaminotetralin (PAT) ligands synthesized in our laboratory interact with mammalian brainH1 receptors in vivo to modulate catecholamine (dopamine, norepinephrine) neurotransmitter synthesis Alter a tion of brain catecholamine neurotransmitter levels is linked to several psychiatric and neurological diseases including, schizophrenia, depression, insomnia, epilepsy, and Parkinsons disease. Meanwhile, activation of H1 receptors in the periphery mediates allergic responses. There is evidence that central and peripheral histamine H1 receptors can signal through different second messenger pathways. The functional selectivity hypothesis suggests that the same ligand may act as an agonist or inverse agonist at the same GPCR, depending on differential G protein coupling with the receptor. Thus, it is theoretically possible for the same ligand to modulate brain and peripheral H1 receptor signali ng to result in therapeutically useful central H1 receptor activation without provoking a peripheral H1mediated allergic response. A lipophilic, brain -
13 penetrating ligand, (2 R ,4 S ) ( ) trans N,Ndimethylamino4 phenyl 1,2,3,4tetrahydro2 naphthalenamine ( phenylaminotetralin, PAT) possesses the t herapeutically useful pharmacological properties of inverse agonis mregarding H1 signaling via the H1G and phospholipase (PL) C and inositol phosphate (IP) formation pathway that mediates allergic react ion responses ; while acting as an agonist regarding H1 signaling via the G S adenylyl cyclase (AC) and cyclic 3,5 adenosinemonophosphate (cAMP) pathway that stimulates brain catecholamine neurotransmitter synthesis. The general goal of this dissertation is to characterize the molecular determinants of PAT type ligands with the H1GPCR that govern functionally selective signaling. Results are predicted to characterize H1 molecular structure provide structure activity relationships, and function al determinants to advance H1 drug design for psychiatric and neurological dis orders
14 CHAPTER 1 G PROTEIN COUPLED RECEPTORS: E NDOGENOUS AND THERAP EUTIC SIGNIFICANCE OF HIST AMINE RECEPTORS Introduction to G Protein Coupled Receptors G proteincoupled receptors ( GPCRs ) represent the largest family of membrane proteins in the human genome, and comprise approximately 1 % of our total genetic code ( Johnston & Siderovski, 2007) Knockout models have shown GPCRs that are critical to regulating the signaling processes of the cardiovascular, nervous, endocrine, metabolic, and sensory systems (Kobilka & Rohrer, 1998; Spiegel & Weinstein, 2004) Mutations in the genes encoding for GP CRs have been identified as the root cause of many inherited diseases. For example, a mutation to the gene encoding for G s has been linked to bone abnormalities, hormone resistance (pseudohypoparathyroidism) and hormone hypersecretion (McCuneAlbright syn drome) (Spiegel & Weinstein, 2004) Due to their ubiquitous distribution throughout the body and their critical role in regulating a vast number of signaling cascades GPCRs are an obvious therapeutic target, for both peptide and small molecule ligands. GPCRs share a common structural motif that includes seven transmembrane helices, an extracellular N terminus, an intracellular C terminus, and a unique ability to couple to intracellular signaling modifiers called G proteins, as shown in f igure 11 (Johnston & Siderovski, 2007) Current estimates state there are approximately 800 GPCRs that are encoded within the human genome. However, the majority of these 800 receptors are characterized as orphans, meaning that their endogenous ligand has yet to be uncovered. By comparing their sequence homologies GPCRs can be broadly classified into three separate families: the rhodopsinlike family (A), secretin like (B), glutamate family (C ) ( Kobilka B. 2006) It is worth mentioning that GPCR families can
15 be broken down by the binding location of their endogenous ligand, also called the orthosteric binding site. Class A GPCRs tend to bind their ligands in their transmembrane domains (TMDs). Class B GPCRs tend to bind their endogenous ligand in the extracellular loops, and class C GPCRs bind at their unique N terminal Venus Flytrap moiety. There is currently much debate about the minimal functional unit for class A and B GPCRs. Class C GPCRs (GABAB), are well documented as dimeric receptors, however this has yet to be demonstrated definitively in the remaining two classes (Pin et al., 2004) A recent paper demonstrated that nucleotide exchange could take place 2Adrenergic Receptor is isolated within a reconstituted high density lipoprotein ( Whorton et al., 2007) Contrary to this, several modeling papers indicate G proteins have a more energetically favorable interaction with a dimeric subunit (Hamm, 2001; Liang et al., 2003) The present consensus in the literature is th at class A and B GPCRs can form dimers and oligomers, but the minimal functional unit has yet to be determined. A more detailed discussion on GPCR dimerization will follow later in this document. Despite being arranged into several different classes, all GPCRs couple intracellularly to G proteins. These unique, heterotrimeric proteins allow signals from the extracellular (N terminal) side of the receptor to exert their influence on the intracellular (C terminal) side of a cell. G proteins consist of three These three subunits are crucial to the function of the entire GPCR unit, as they serve as g uanine nucleotide exchange factors The guanine nucleotide exchange takes
16 subunit and is the catalyst for the i ntracellular signaling cascade (Johnston & Siderovski, 2007) It is important to note that many GPCRs have the ability to activate in the absence of an agonist ligand. This phenomenon is caused by the dynamic and flexible nature of GPCRs and is called basal or constitutive activity (Kobilka & Deupi, 2007) Activation of a receptor, with or without an agonist present, causes conformational changes to the receptor that are transmitted to the intracellular loops. These changes result in a conformational change in the G protein, exchanging a guanine diphosphate (GDP) unit that is bound to the G subunit for a guanine triphosph ate (GTP). This nucleotide exchange causes the heterotrimeric G protein to dissociate into Gand G, both of which act as intracellular signaling effectors, acting upon ion channels, adenylyl cyclases, phosphodiesterases, kinases, and phospholipases as shown below in figure 1 2 These are but a few examples from the litany of downstream targets that GPCRs have the abili ty to modulate (Johnston & Siderovski, 2007) The downstream target of a particular GPCR is determined by its location in the body and the G protein to which it is coupled. This is well demonstrated by human Histamine H1 receptor. In most mammalian tissues the H1 receptor signals through G, which activates PLC, increasing the intracellular concentration of IP and diacylglycerols (DAG). In stark contrast to this, H1 receptors found in the mammalian brain demonstrate the ability to function through G s, leading to activation of AC and increased intracellular levels of cAMP as well as intracellular Ca2+ (Moniri et al, 2004) The ability of a GPCR to couple to several G proteins is well established and the phenomenon is called functional selectivity (Moniri et al, 2004; Urban et al., 2007) To date twenty different
17 Gsubunits have been identified in mammalian tissues. These can be further classified into four subfamilies based upon their structure and the functional pathways they modify. These subfamilies ar e G s,G /11,G i/o, and G 12/13 subunits exist and (Offermans, 2003) The ability of GPCRs to affect such a wide variety of downstream mechanisms, in nearly all regions of the body has made them very attractive drug targets. A recent study done by Overington et al. in 2006, suggests that rhodopsinlike (Class A) GPCRs account for over twenty five percent of FDA approved drugs. Class A GPCRs represent only part of the superfamily, so as more information about the other GPCR classes becomes available that number is expected to increase further. Ballesteros Numbering System Given the vast array of GPCRs and their structural differ ences and complexity, it is quite difficult to discuss results across receptors, especially when referring to an amino acid residue as a single number in a chain of 400 plus residues. This issue is compounded by the fact that GPCRs can have differing residue lengths for each of their shared structural features including transmembrane domain, intracellular and extracellular loops, and N and C termini (Hermans, 2003; Kobilka B. 2006) For example, amino acid 400 may be located in the intracellular loops for one receptor, yet be in theTMD region for another class A receptor. These numbers provide no indication as to which TMD the residue belongs, which is critical for GPCRs like the H1 receptor that possess a long third intracellular loop (ICL) of about 200 amino acids. If you used conventional residue numbering, meaning residues from 1400, the long loop would
18 cause the residues in TMD 6 and 7 to have much higher numbers than in related receptors where ICL3 is shorter. T o help address this issue, a new number ing system specifically for GPCRs was proposed by Ballesteros (Ballesteros et al., 2001) This newer classification proved incredibly useful and caught on quite quickly, so an indepth discussion is warranted here. By careful analysis of GPCR sequences, the authors noticed that the origins of both the intracellular and extracellular TMD regions were designated by a positively charged cluster of lysine or arginine residues. These residues associate with the negatively charged phospholipid heads of the cell membrane to create a series of strong ionic bonds that help the receptor associate with the membrane. Arginine and l ysine clusters proved to be consistent markers of the beginning and end of the TMD helices across class A GPCRs; however, the residue number was never the same in the overall sequence. The arginine and lysine observation seemed to beg the question; can a specific position within the GPCR architecture be as important as the amino acid number in the sequence? The answer is an emphatic yes. By using the most conserved residue within each TMD helix as a reference point, and selecting the most conserved residues across the entire GPCR superfamily, a numbering system was created ( Ballesteros et al., 2001) The most conserved residue in each TMD was assigned an arbitrary value of X.50, where X indicates the TMD where the residue is located. As an example W6.48, would correspond to a tryptophan residue in TMD 6 that is two residues before the most conserved residue (6.50). In addition, this nomenclature has been extended to point mutations. The principle is the same, each amino acid is given a number relative to the
19 most conserved residue in that helix ; however the amino acid that the residue is being mutated to is listed in one letter amino acid code immediately following the location. Taking W6.48 again as an example, an alanine mutant in TMD 6, located two residues before the most conserved residue would correspond to W6.48A. Where W is the native residue, 6 is the TMD, .48 is the position, and A, t he one letter code for alanine, is the residue that W is mutated to (Ballesteros et al., 2001) The last nuance to this numbering system is that the numbers increase or decrease as you move intracellularly, depending upon the TMD you are considering. GPCRs are folded proteins that pass through the membrane seven times. It winds down from the extracellular N terminus through the cell membrane into TMD 1, across the cytoplasm in ICL 1, up through the membrane for TMD 2, across extracellular loop 1 ( ECL1 ) down through TMD 3, across ICL 2, up though TMD 4, across ECL 2, down thought TMD 5, across ICL 3, up through TMD 6, across ECL 3, before winding down through the membrane with TMD 7 and the c terminus. Residues always count up toward the most conserved residue, but the starting point varies as you follow the peptide chain through the membrane. For TMD 1, 3, 5, and 7 the numbers increase as you move intracellularly because the sequence is heading in that direction as you follow from the N to C terminus. For TMD 2 4 and 6 the numbers still increase as you move towards X.50, but they now increase as one progress extracellularly. Again, this is due to the origin of these helices being found at the intracellular loops and proceeding upward towards the extracell ular region. An example of this is demonstrated below in f igure 1 3 with notations to indicate how the numbering proceeds ( Ballesteros et al., 2001)
20 Using Ballesteros Numbering to Identify Critical Aminergic Residues The literature is inundated with ot her examples of a position being important across species subtypes. For example, the DRY (AspArg Tyr) motif at the intracellular end of TMD 3 has been implicated in the activation of nearly all rhodopsinlike (class A) receptors. Of the three residues Tyr is the least conserved, yet is still found in 67% of class A GPCRs. Arg, which is strongly basic, and Asp/Glu, which are strongly acidic, form an important ionic lock interaction that is critical to activation. Not surprisingly, Arg and Asp/Glu are cons erved across 87% of class A GPCRs (Mirzadegan et al., 2003) In fact, a GPCR without the DRY motif that was able to function was deemed significant enough to warrant a standalone publication in 2005 (Flanagan, 2005) It is important to note, that even though the residue numbers were changing, all of these GPCRs possess this DRY motif at the same location, the cytoplasmic end of TMD 3. Suggesting that class A GPCRs have conserved structural regions that are critical for receptor function. These regions that are known to be involved in receptor activation are called molecular switches (Kobilka B. 2006) A similar trend is found with a cluster of residues in TMD 6 that are collectively called the rotamer toggle switch. These residues C/T/S6.47, W6.48A, and F 6.52 form a series of stabilizing interactions that modulate the degree of bending caused by a proline residue in TMD 6. This phenomenon is called the proline kink in TMD 6. Ligands 2AR and across many other aminergic GPCRs produce strong interactions with these residues, specifically W6.48 and F6.52 that seem to be linked to receptor activation (Shi et al., 2002) In the off position, there is a stabilizing interaction between the hydroxyl containing residue at position 6.47 and W6.48 that reduces t he bend caused by P6.50. In the on position this stabilizing interaction is lost at the proline kink
21 becomes much more pronounced leading to a movement in the cytoplasmic end of TMD 6 causing receptor activation (Shi et al., 2002) Here again, it is observed that a particular location within a series of GPCRs is just as critical for activation of the receptor as the amino acids that are located at that position. Therefore when discussing GPCRs and attempting to draw conclusions, it is important to use the Ballesteros numbering system rather than the amino acid number in the sequence. Doing so provides significantly more information about that residue as well as its location within the GPCR architecture, while at the same time allowing comparison s to different receptors. When the Ballesteros numbering system is employed, and the molecular determinants for several other aminergic receptors are examined, an unusual trend is noticed. Certain residue numbers seem to be important for the binding of endogenous ligands across several aminergic receptors. Residues at positions 5.42 and 5.46 seem to be important for the binding and function of endogenous compounds at the H1 2adrenergic receptor and at serotonin 2 family ( 5HT2) receptors. A t the H1 receptor it has been proposed that an asparagine at position 5.46 (N5.46) forms a critical hydrogen bond with the Nimidazole nitrogen in histamine. Based upon mutagenesis of this residue it is clear that it has a profound impact upon histamine b inding and function Specifically the affinity (Ki) for histamine is reduced by approximately 40 fold, with PLC activation being similarly impacted; demonstrating a dramatic EC50 reduction of about 300 fold (Smit et al., 1999; Fang, Travers and Booth.,2009 pre publication). Interestingly, the maximum signal (EMax) for PLC activity was not reduced and the potency and efficacy of cAMP accumulation was not impacted.
22 Position 5.42 is a threonine in the H H1R and displays a significant reduction in affinity and ef ficacy when compared to the WT receptor. A 10fold reduction in binding affinity and a significant drop in efficacy ( EC50 values ) were observed for this point mutant. Preliminary molecular modeling, with homology to bovine rhodopsin, indicated that this re sidue has an important stabilizing interaction with N5.46, and holds this residue in an ideal location to bind to the imidazole ring of histamine (Fang, Travers and Booth., 2009 prepublication ). 2AR is considered, its endogenous agonist epine phrine forms extremely strong hydrogen bonding contacts in this exact same region. At this receptor, 5.42 and 5.46 are both occupied by serine residues. Upon mutation to nonfunctional alanine residues, affinity for epinephrine was found to decrease markedly. At S5.42A the affinity dropped 25fold and at 5.46A, the affinity was reduced 39 fold. This sharp drop in affinity is attributed to hydrogen bonding between the two serine residues and the hydroxyl groups on the catechol ring (Carmine et al., 2004; Bhattacharya et al., 2ARs (39 fold), is nearly identical to the reduction in affinity for histamine at H1 N5.46Areceptors (40 fold). When the more closely related 5HT2 receptor family is considered, things become slightly more complex as there are three subtypes A, B, and C of this receptor. In the interest of keeping this brief, the focus will be on human 5HT2A receptors. In human 5HT2A receptors position 5.46 is a serine, and has been strongly implicated in ligand binding differences that have been observed between human and rat2A receptors (Miller et al.,2009). In rats, this residue is an alanine, and mutagenesis studies have shown
23 tha t by mutating the rat receptor from an alanine to a serine rescues the binding profile observed in the human receptor. This effect was originally observed in 1992 by Kao et al with the well knownradioligandsketanserin and mesulergine. Briefly, mesulergi ne displayed a low affinity for human 5HT receptors and a high affinity at rat 5HT receptors. By creating a S5.46A mutation to the human receptor, and mimicking the rat sequence, the affinity of 3H mesulergine was dramatically increased and resembled the a ffinity for rat receptors. The researchers concluded that position 5.46 was critical for ligand binding at 5HT receptors (Kao et al., 1992; Johnson et al., 1993). Interestingly, residues 5.46 has also been proposed to interact with the indole nitrogen of s erotonin, which is quite similar to the imidazole interaction seen at this position with histamine (Braden and Nichols, 2007). The same paper concludes that the serine residues at positions 5.43, and possibly 5.42, have hydrogen bond interaction with the hydroxyl moiety of serotonin and other tryptamines. These results fit with the interactions of epinephrine and residues S5.42 2AR. Upon examining the whole picture, it seems that positions 5.42/5.43 are polar residues at these aminergic receptors and produce stabilizing interactions, either directly with a hydrogen bonding moiety on the ligand, or by stabilizing a nearby residue that is critical for ligand binding. The 5.46 position seems to have strong, direct interactions with the endog enous ligands of these aminergic receptors. In 5HT2Aand H1 receptors, 5.46 forms a strong hydrogen bonding interaction with the nitrogen atoms in the heteroaromatic rings in both serotonin and histamine. 2AR is considered, the interaction with t he nitrogen is lost, but a hydrogen bond is formed with the second hydroxyl on the catechol ring of epinephrine. These
24 results, when taken together, suggest that there may be an overall similarity in the way these aminergic GPCRs are binding their endogenous ligands. The fact that these receptors are all derived from a common ancestor fits with this observation and could suggest that each receptor evolved residues at key positions in order to bind its endogenous ligand specifically The conservation of thes e positions across receptor families also suggests that these positions are critical for causing the conformational changes that are responsible for receptor activation, and could suggest that there is a common mechanism of activation among these aminergic GPCRs. This adds significance to probing the molecular determinants of the H1receptor, as the results can be extrapolated across aminergic receptors. Histamine Receptors There are four subtypes of histamine receptors, H1, H2, H3, and H4, each of which is a G PCR. H4 is the most recently discovered and it can be found in bone marrow, leukocytes, and mast cells. The H4 subtype shares approximately 40% of its sequence with H3 and about 18% with H1 and H2. H4 receptors couple to G i /o and inhibit cAMP formation and can also activate PLC/IP signaling (Nakamura et al., 2000). The remaining histamine subtypes (H1H3) are expressed in mammalian brain tissue, and are known to play roles in regulating metabolic states, sleepwake cycles, and behavior patterns (Haas and Panula, 2003). The H3 receptor share approximately 20% of their sequencing with H1 and H2 subtypes, and couple to G i/o/AC/cAMP signaling, inhibit ingcAMP release, and are found in central nervous system (CNS) and peripheral ner vous system( PNS ) cells It can act as either an autoreceptor to inhibit further release of histamine or as a
25 heteroreceptor, which regulates the release of other neurotransmitters like serotonin, norepinephrine, acetylcholine, dopamine and amino butyric acid ( GABA) (Lovenberg et al., 1999; Esbenshade, 2003; Yanai& Toshiro, 2007; Haas et al., 2008) Because of their ability to control the release of a variety of biogenic amines H3 receptors have been proposed as a novel antipsychotic target s(Ito, 2009) The H2receptor shares 21% of its overall homology with H1 and couples mainly to G S/AC/cAMP signaling. It is found in gastric parietal cells, vascular smooth muscle, heart muscle, and in the brain (Hill et al, 1997;Smit et al., 1999) Antagonists for the H2 receptor are used clinically in the treatment of gastrointestinal (GI) ulcers, acid reflux disease, and Zollinger Ellison syndrome. Each disorder is characterized by an overproduction of gastric acid, administering a selective H2 antagonist will decrease a cid secretion from gastric parietal cells and result in an amelioration of the symptoms (Leurs et al, 1995; Smit et al., 1999) This approach has lost favor in recent years, as protonpump inhibitors have shown to be more efficacious (Haas et al., 2008; Malf ertheiner et al, 2009) Interestingly, the H1 receptor has its highest transmembrane homology (44%) with the muscarinic M1 receptor and is more closely related to the m uscarinic M1M5 and the serotonergic 5HT2 family than to the H2 receptor (Leurs et al., 1995; Smit et al., 1999). The phylogenic tree for the receptors that are related to H1can be found in figure 1 4 below. The traditional therapeutic role for H H1R is already well defined. As early as 1910, Dale and Laidlaw observed that injection of histamine into the body produced symptoms that were identical to an allergic response (Brunton, Lazo, & Parker, 2006) In the years
26 since this initial discovery, the H1 receptor has been largely associated with this allergic response. The histamine H1 receptor, however, is a G protein coupled receptor, and as such, it has the ability to couple to several different G proteins. This phenomenon is called functional selectiv ity. (Hermans, 2003; Urban et al., 2007) The G proteins linked to the h uman H1r eceptor include the G/PLC second messenger pathway that is associated with allergic responses in the peripheral nervous system. I n addition to this pathway, it has been well established that the receptor can stimulate the adenylyl cyclase pathway in mammalian brain tissues (Moniri et al, 2004) It is believed to be mediated through G S, but this has yet to be conclusively demonstrated and there is at least some debate in the current literature about G involvement (Maruko et al., 2005). A third signaling pathway is mediated by phospholipase A2 (PLA2), which cleaves the fatty acid from the second carbon of glycerol, resulting in the Ca2+ dependent release of arachidonic acid (AA ) from the cell membrane (Leurs et al, 1995) It has been shown that PLC and DAGs stimulate the release of AA Due to activation by the second messengers, Ca2+ and DAGs, there is significant crosstalk between signaling pathways and AA release has yet to be tied to a specific G subunit. Studies have implicated a direct G protein interaction with PLA2, but this phenomenon has only been conclusively demonstrated in Sporothrixschenkii a pathogenic fungus (Berrios et al., 2009) The pathways for Gand Gsignaling can be found in figure 15 below. H1 receptors have been characterized using the radioligand, 3H mepyramine, in a wide variety of mammalian cells. In 1986, Donaldson et al. demonstrated that H1 receptor activation led to smooth muscle contractio n via the G Q/PLC/IP signaling
27 pathway. This response can present itself as respiratory distress (bronchial constriction), diarrhea (GI contraction), or edema and hypertension (cardiovascular contraction). However, in the adrenal gland and in mammalian brain tissue occupation of H1 by a suitable agonist produces cAMP TH stimulation, catecholamine synthesis, and the release of epinephrine/norepinephrine (Marley and Robotis 1998; Booth & Moniri, 2006) Histamine Biosynthesis and Catabolism Histaminergic neurons possess the ability to synthesize the neurotransmitter histamine. Biosynthesis of histamine is quite efficient, even by the standards of the human body. Histamines precursor, Lhistidine, is taken up by the cerebrospinal fluid (CSF) and by the Lamino acid transporters found directly on histaminergicneurons (Haas et al., 2008; Stahl, 2008) Once taken up, histamine is acted upon by the enzyme histidine decarboxylase (HDC) to synthesize histamine. Contrary to other neurotransmitters, the rate limiti ng step in histamine synthesis is the bioavailability of the precursor (Haas et al., 2008) This is unusual among the biogenic amines and can be explained by the realtively simple biosynthetic pathway for histamine. Histamine is stored in cell somata and is packaged into vessicles by vesicular monoamine transporter 2 by exchanging two protons. Histamine is released when an action potential arrives, and is regulated by the histamine H3receptor (an autoreceptor) through feedback inhibition. Histamine is pred ominantly inactivated in the extracellular space of the CNS by two enzymes, histamine N methyltransferase and monoamine oxidase B (HN M T and MAO B, respectively) (Haas et al., 2008; Stahl, 2008) Initial inactivation occurs by HNMT, using S adenosyl methioni ne (SAM) as a methyl donating substrate, to yield telemethylhistamine. When in the brain, this metabolite is acted upon
28 by MAO B to produce the inactive metabolite t methyl imidazolacetic acid, as depicted in figure 1 6 below. In the periphery, histamine is metabolized directly to imidazolacetic acid by the enzyme diamine oxidase. (Haas et al., 2008) Histamine in the CNS Histaminergic neurons express the enzyme Lhistidine decarboxylase, which confers the ability to synthesize histamine, and are located ex clusively in the tuberomammillary nucleus (TMN) in the posterior hypothalamus. Despite their limited location, histamine neurons project broadly throughout the entire brain and are found in the cerebral cortex, thalamus, basal ganglia, and amygdala(Haas et al., 2008; Stahl, 2008) Ascending pathways innervate the cortex and thalamus and play important physiological roles in arousal and learning. Histamine projections also innervate cholinergic neurons in the basal forebrain, which are postulated to play rol es in learning and arousal. Projections extend into the hippocampus, nucleus accumbens, and amygdala to modulate additional behaviors. Descending pathways innervate brainstem functions and regulate further neurotransmitter pathways capable of modifying neuronal activity and release(Stahl, 2008) Because of their extensive projections from the TMN histaminergic neurons are postulated to play significant roles in a number of neurological pathologies including, but not limited to insomnia, allergic responses, Alzheimers disease, Parkinsons disease, depression, and schizophrenia. Histamine receptors have been associated with sedation since the first antihistamines were produced. In fact, the second generation of antihistamines was created to help alleviate prolonged sedation associated with many of the original antihistamines such as diphenhydramine (Benadryl) ,azatadine, hydroxyzine, triprolidine, and chlorpheniramine(Leurs et al, 1995) The structures for each of these compounds
29 can be found in figure 1 7 below, along with their second generation (nonsedating) counterparts. Many early antipsychotic drugs, such as doxepin and amitriptyline, were originally classified as anti histamines. Both compounds have strong activity as H1 inverse agonists, but exert their antipsychotic activity by blocking serotonin and norepinephrine transporters. Regardless, these compounds also display the strong sedative properties associated with H1 inverse agonism (Stahl, 2008) These observations and many years of research have tightly interwoven histamine H1 inverse ag onism and prolonged sedation. However, recent results have caused this link to come under scrutiny. A study by Stahl in 2008 revealed that many H1 ligands that demonstrate prolonged sedative actions are given in amounts that significantly exceed the necess ary concentrations to effectively target H1 receptors. Amitriptyline and doxepin both have H1 inverse agonism as their most potent property, yet amitriptyline is dosed at 400x its H1KD(Stahl, 2008). At their pharmacological dose, amitriptyline and doxepin should be considered dirty drugs, as they can target serotonin and norepinephrine transporters, muscarinic M1, Alpha1, and serotonin 2A receptors. Of these antihistamines with a mixed pharmacology, doxepin is considered to be the most selective, demonstrat ing H1 selectivity of greater than two orders of magnitude(Stahl, 2008) By giving doxepin in a low dose of 16 mgs, the long term sedation associated with antihistamines has been eliminated, while the soporific qualities associated with H1 inverse agonism remain. This result has been confirmed in two separate clinical trials, which will be discussed later, and seems to suggest that selective H1 inverse agonists do not exhibit the prolonged sedation traditionally associated with antihistamines (Roth et al., 2007; Scharf et al., 2008)
30 The bodys ability to sleep is controlled by a variety of wake and sleep promoting neurotransmitters and hormones that are released in a tightly regulated manner to match circadian rhythms (Stahl, 2008) This balance between wake and sleep promoting compounds can easily be perturbed, leading to excesses in wakepromoting neurotransmitters and causing insomnia. Wakepromoting neurotransmitters include histamine, acetylcholine, norepinephrine, serotonin, and the pept ide orexin. Interestingly, the tuberomammillary nucleus is the location where histaminergic neurons arise and a location where these additional pathways converge. Because of this, the TMN is considered to be one of the most critical brain regions for arous al (Stahl, 2008) The litany of wakepromoting compounds, is balanced primarily by the sleep promoter GABA. A second thalamic region called the ventrolateralpreoptic area (VLPO) has projections into the wakepromoting brain centers, allowing GABA to exert its effects on each wakepromoting neurotransmitter system Effects on histaminergic neurons include 3 types of interaction with varying GABA sensitivities that have been identified s (Haas et al., 2008; Stahl, 2008) Histamine n eurotransmission displays an obvious link to circadian rhythm, as concentrations wax during high activity levels and wane during sleep periods. This has been observed in fish, monkeys, rodents, and humans, suggesting that histamine levels are critical to g overning the sleepwake cycles across species (Haas et al., 2008) Insomnia is characterized by a state of hyperarousal, and in the case of chronic insomnia is hypothesized to occur during the day and night. Perturbations of this nature can be extremely disruptive to a patient as they cannot sleep at night and are unable to make up for lost sleep during the day (Stahl, 2008) Treatments for insomnia must either
31 inhibit wakepromoting neurotransmitters, especially histamine, but also including epinephrine n orepinephrine, acetylcholine, and serotonin, or act by enhancing the inhibitory effects of GABA. Traditional therapeutics for insomnia focuses on the GABA pathways to suppress all wake promoting regions of the brain. These approaches not only produce excessive sedation, but also can be dangerous when used incorrectly; as is the case with barbiturates. Positive a llosteric modulators of GABA, which bind and potentiate the effects of GABA, such as benzodiazepines and zolpidem (Ambien) are the most commonly prescribed current treatments (Stahl, 2008) However, these newer ligands are not perfect. The half life of the compound should not extend past the duration of sleep. If it does, the compound can accumulate in the body and cause severe sedation or present with a hangover effect where sedation is carried into the next day. Neither of these are desired pharmacological properties and some severe cases of side effects have been reported for zolpidem. Side effects range from sleep eating, talking, walk ing, and even driving have been reported. More severe side effects such as temporary amnesia, ataxia, hallucinations, reduced inhibition, and increased sexual libido have also been observed. Prolonged use of Ambien has shown addictive properties and has si gnificant abuse potential (Hoque& Chesson, 2008) By taking the alternate route and suppressing wakepromoting neurotransmitters, it is possible to target one wakepromoting brain region at a time. The selective H1 inverse agonist doxepin is able to fill thi s niche. As mentioned previously, low dose doxepin has been show to be extremely effective in the treatment of insomnia and does not possess the dramatic side effect profile of Ambien. In fact, double blind studies
32 showed that doxepin had a side effect pro file less than placebo and produced no next day sedation( Roth et al., 2007 ; Scharf et al., 2008) .These results strongly suggest that selective H1 inverse agonists are efficacious in the treatment of insomnia and seem to eliminate or at least dramatically l essen the side effect profile of other sleep aids. Recent advances in imaging techniques, particularly positron emission tomography (PET) studies have allowed for real time monitoring of neurotransmission within the human brain. The development of a11C doxepin allowed this technique to be applied to the histaminergic system and comparisons to be drawn between patients with varying behavioral and pathological disorders. A review of these studies was undertaken, comparing histaminergic neurotransmission in c ontrol patient groups against those that were aging, presenting with various stages of Alzheimers disease, depression, schizophrenia, and epilepsy (Yanai & Toshiro, 2007) Each patient group will be addressed briefly here. In the aging brain, it has been well documented that attention, learning, and memory decline in humans and animals. Senescenceaccelerated mice, used as a model for the aging brain, demonstrate significant agerelated deficits in passive avoidance tests. Interestingly, the administration of an H3 antagonist thioperamide markedly improved the results for the same mice in avoiding the constant current and voltage shocks in the passive avoidance tests (Meguro et al., 1995) H3receptors are autoreceptors that provide feedback inhibition to pr event excessive histamine release, blocking these receptors will drastically increase histamine neurotransmission. Thus, increases in histaminergic neurotransmission appear to improve memory and cognition significantly in non human subjects (Meguro et al., 1995; Haas et al., 2008) It has been
33 established that the concentration of histamine metabolites in CSF and histamine levels in the human brain increase with age, suggesting a possible disconnect between histamine release and neurotransmission. In keeping with this, PET studies reveal a pronounced reduction in H1R binding sites in elderly patients (Yanai et al., 1992; Yanai& Toshiro, 2007) It is hypothesized that the excessive release of histamine is able to downregulate the available H H1R population, resulting in a decrease in overall histaminergic neurotransmission and explaining the detrimental effects on memory and cognition observed in murine studies (Haas et al., 2008) The downregulation of H1Rs by histamine is also firmly established, being demonstrated in guinea pigs by Quach et al. in 1991. Alzheimers disease (AD) is traditionally characterized by deficiencies in neurotransmission of the cholinergic neurons of the brain. However, deficiencies in neurotransmission have also been observed in cholinergic receptors, both muscarinic and nicotinic subtypes, and also in acetylcholinesterase levels. Autopsied human brains have also been examined to help unravel the etiology of AD, they have reveal ed the characteristic neurofibrillary tangles in hypothalamus and reported a loss of large histaminergic neurons in the TMN (Nakamura et al., 1993) Thorough analysis by Schneider et al. in 1997 reported that the levels of HDC and choline acetyltransferase, the enzymes responsible for the synthesis of histamine and acetylcholine respectively, were drastically reduced in AD patients. Additional reports suggest that histamine levels in the brain are reduced in patients with AD (Panula et al., 1998) The use of PET imaging in conjunction with 11C doxepin examined the levels of H1R expressed in the brains of mild to severe AD patients and compared them with
34 age matched control groups. A decrease in H1R brain density was observed that was proportional to the sev erity of AD. Despite its limited scope, this study was the first to establish deficits in histamine neurotransmission in the brains of living AD patients. The study concludes that disruption of histamine neurotransmission and the death of histaminergic neurons could play a significant role in the etiology of AD (Higuchi et al., 2000; Haas et al., 2008; Yanai & Toshiro, 2007) Depression is a psychiatric disorder characterized by alterations in the patients normal behavior patterns. This can manifest as dist urbances in sleep, appetite, and/or physical potential. Symptoms include prolonged feelings of helplessness, unnecessary guilt, extreme fatigue, insomnia or excessive sleep, and difficulty concentrating. Given the vast array of symptoms, some of which are contradictory (sleep/wake cycles), it is perhaps unsurprising that an exact mechanism for the etiology of depression has yet to be produced(Yanai& Toshiro, 2007) Regardless, it has long been established that many antipsychotic drugs possess strong H1 rece ptor affinity. In fact, chlorpromazine and amitriptyline, used clinically as first generation treatments for schizophrenia and depression respectively, were originally labeled as antihistamines. Advances in molecular biology, such as receptor cloning and expression, later showed chlorpromazine acted by blocking dopamine D2 receptors and that amitriptyline treated depression by blocking serotonin and norepinephrine transporters (Stahl, 2008) Nevertheless, strong evidence remains for a role of H1 receptors in the etiology of depression. It is interesting to note that many of the symptoms associated with depression, such as disturbances in sleep, appetite, and physical potential can be linked to the
35 functions of the histaminergic neuronal system. Loss of hist amine or histamine receptors has also been established as an animal model for human depression(Haas et al., 2008) Despite this knowledge, there is currently little investigation into the possible role of histamine receptors in depression. As of 2007, only one human PET study examining 11C doxepin binding in the brain of depressed patients has been performed(Yanai& Toshiro, 2007; Haas et al., 2008) Results were compared between patients, who had completed a self rating depression scale to assess the severi ty of their depression, and healthy, agematched patients. Data examined from 10 depressed patients showed a strong correlation between the severity of the depression and a reduction in H1 binding sites found in the frontal, temporal, occipital cortex, and the cingulate gyrus. (Yanai & Toshiro, 2007) Although preliminary, these results seem to suggest a possible role for histamine and histamine receptors in the etiology of depression. Further studies, with anti depressant nave patients, must be undertaken to validate this hypothesis. Histaminergic neurons have also been implicated in schizophrenia. In murine methamphetamine induced schizophrenia models, blockade of H1 receptor neurotransmission has been shown to attenuate changes in animal behavior (Yanai & Toshiro, 2007) Schizophrenic patients demonstrate a pronounced increase in the levels of N tele methyl histamine a histamine metabolite, in their CSF. Upon imaging the schizophrenic human brain with PET, Iwabuchi et al. demonstrated a significant decrease in H1 receptor density in the frontal, temporal, occipital, and cingulate cortices, as well as in the striatum and thalamus (Haas et al., 2008; Yanai & Toshiro, 2007)
36 Therapeutic Use of Histamine Ligands Histamine H1 receptors have been linked with the weight gain associated with atypical antipsychotics, specifically olanzapine and clozapine, in a number of publications (Kirk et al., 2009; Deng et al., 2009; Reynolds and Kirk, 2010). Unfortunately, this literature is inundated with indirect observations based largely upon the affinity of atypical antipsychotics for serotonergic, muscarinic, dopaminergic, and histaminergic receptors. These papers report that H1 antagonism alone is unlikely to explain the weight gain observed with these drugs. As stated in Reynolds and Kirk (2010) These studies (demonstrating circumstantial H1receptor involvement) do not, however, exclude effects on histamine systems being secondary to, or in addition to, 5HT2 c receptor antagonism. This quote demonstrates the consensus in the literature, that H1`receptors may play a role in atypical antipsychotic associated weight gain, but there is scant evidence that H1 receptors are solely responsible for this effect. A thorough study performed by Stephen Stahl (Stahl, 2008) demonstrated that one of the original tricyclic antidepressants, doxepin, possesses a significant H1 receptor selectivity (greater than two orders of magnitude) when compared to serotonin and norepinephrine transporters, muscarinic M1, alpha1, and 5HT2A receptors. However, the dose required to treat depression is large enough to impart a dirty pharmacology; meaning doxepin is able to target serotonin and norepinephrine transporters in addition to H1 receptors. In the same paper, Stahl demonstrates that lower doses (16 mg vs.150300 mg for antidepressant treatment) of doxepin, which specifically target H1 receptors at the aforementioned lower dose, acts as a novel hypnotic that is extremely effective in the treatment of insomnia(Stahl, 2008) Stahl eloq uently demonstrates that
37 many first generation antihistamines that display long term sedation, including diphenhydramine (Benadryl) and doxylamine (the sedating ingredient in Nyquil), are dosed far beyond the amount that would selectively target H1rec eptors. At the administered doses, both compounds have significant anti muscarinic effects that could easily account for the excessive next day sedation observed with these compounds. It is significant to note that next day sedation was not observed during the clinical trials performed with low dose doxepin (Roth et al., 2007; Stahl, 2008) Clinical trials have been undertaken with low dose Doxepin for the treatment of insomnia and the results have been quite positive. In a trial of elderly insomniac pati ents using 16mg of doxepin, all patients showed significant (p < 0.001) improvement in wake time after sleep onset, total sleep time, and overall sleep efficiency (Scharf et al., 2008) A second study examining the effects of similar doses in adults (ages 1864) with primary insomnia demonstrated similar results. 1, 3, and 6 mg doses of doxepin produced improvements in objective and subjective sleep maintenance and duration up until the final hour of the night (Roth et al., 2007) Both studies reported side effects that were comparable or less than the placebo and there were no reported anticholinergic effects, memory impairment, or next day hangover (Roth et al., 2007; Scharf et al., 200 8) .This is an intriguing result, suggesting selective histamine H1antagonism is a novel therapeutic route for the treatment of insomnia, which appears to overcome the limitations of many sleep aids currently on the market. Current research suggests that histamine plays a critical role in the CNS by acting as an all natural anti convulsive agent (Haas et al., 2008; Yanai & Toshiro, 2007; Chen et al., 2003) Alphafluoromethylhistidine, an inhibitor of histidine decarboxylase, has
38 been shown to enhance the s everity of clonic convulsions, and accelerate seizure development in mice where pentylenetetrazol (PTZ) was used to induce a seizure model. This PTZ induced murine seizure model has been recognized as an effective model for human absence epilepsy, as well as myoclonic or generalized tonic clonic seizures (Chen et al., 2003) Interestingly, histamine and selective H1 agonists act as neuroprotective agents, while mepyramine and ketotifen, which are H1 inverse agonists, have been shown to increase focal and ge neral epileptic fits in patients (Yanai& Toshiro, 2007) Diphenhydramine, a similar brainpenetrating H1 antagonist/inverse agonist, has a similar effect when given via an intraperitoneal injection; it exacerbates epileptic fits (Haas et al., 2008) Local hi stamine concentrations have been shown to increase around the foci of complex partial seizures in order to prevent their spread(Chen et al., 2003) These studies strongly emphasize the role of brain histamine as a protective factor against the formation, s pread, intensity, and duration of seizure activity. It is interesting that our lead compound ( ) trans PAT is also a selective, brain penetrating, H1 agonist. The literature failed to mention what selective means from a GPCR standpoint. However, based up on the literature shown above, PLC/IP inverse agonists diphenhydramine, mepyramine, and ketotifen aggravate seizure development, it is reasonable to infer that the selective pathway they are referring to is the adenylyl cyclase/ cAMP pathway. This is the s ame pathway where our lead compound is an agonist and could represent an additional therapeutic role for ( ) trans PAT. Previously Established Molecular Determinants for H1 Receptors When the biogenic amine systems w ere originally characterized by o pthalaldehyde fluorescence chemistry in the 1960s, the brain localization of serotonin
39 and catecholamine systems became the focus of neuroscientists everywhere (Haas et al., 2008; Haas and Panula, 2003) The litany of neuropsychiatric disease states that are characterized by these receptor systems is extensive and intensive research was undertaken on these neurotransmitter systems. However, histamine was not identified as a neurotransmitter at this time, as the widespread amino acid spermidine cross reacted with pthalaldehyde, obscuring the location of histaminergic neurons. Eventually, histamine received its due and was identified as a neurotransmitter, this was verified individually by Panulas group in Washington and Wa das in Osaka in 1984 (Panula et al., 1984; Wada et al., 1984) The belated recognition of histamine as a neurotransmitter, allowed approximately twenty extra years of research to be performed on the catecholinergic and serotonergic systems before work was even begun on histamine. Since then, large strides have been made in characterizing the four subtypes of H1 receptors and the g proteins with which they are associated. Much work has been placed into identifying the amino acid residues that are critical t o the binding and function of histamine ligands. These critical amino acid residues are called molecular determinants and the established literature for the more well known H1 compounds will be discussed here. It is important to discuss the previously est ablished loci that have been identified as important for H1 ligand binding in the literature before delving into novel data. As previously mentioned, the radioligand3H mepyramine has long been used to characterize histamine H1 receptors. Keeping with this, there is a litany of data in the literature about the molecular determinants involved in the binding of mepyramine and the endogenous agonist histamine. Comparing the molecular determinants involved in
40 the binding of mepyramine, an inverse agonist, and hi stamine, a full agonist, will provide a solid framework and a point for comparison in the investigation into the binding of PAT scaffold in TMD 5,6, and 7. It is pertinent to begin the discussion of H1 receptor binding by discussion the endogenous agonist, histamine. Histamine, whose chemical name imidazolethylamine, creates strong contacts with polar amino acid residues in TMD 5, as mentioned briefly above. However, there is a stronger interaction that is conserved across aminergic receptors and is critical to ligand binding, D3.32; this aspartic acid residue forms an ionic bond with the protonated amine moiety of aminergicligands (Smit et al., 1999) D3.32 appears to be the most critical residue involved in ligand binding to the H1 receptor, as both antagonists and agonists require this ionic interaction for proper binding (Smit et al., 1999; Gillard et al., 2002; Jongejan & Leurs, 2005) Radioligand studies using 3H mepyramine and 3H-( ) trans PAT in our own lab indeed confirm this result (Fang, Travers, and Booth., 2009 prepublication) The residues T5.42 and N5.46 play a role in binding histamine to TMD 5, as mentioned briefly earlier, these contacts are important as TMD 5 connects to the ICL3, which is responsible for G protein coupling and subsequent si gnal transduction. Out of the TMD 5 residues, N5.46 possesses the strongest interaction with histamine and demonstrates a 40fold reduction in affinity when an alanine is inserted in its place(Jongejan& Leurs, 2005; Booth et al., 2008; Fang, Travers, and B ooth., 2009 prepublication) Two additional residues in TMD 5, K5.39 and T5.42, are well established as molecular determinants for the binding of histamine. T5.42A is postulated to have a stabilizing hydrogen bond with N5.46A that keeps the receptor in the proper
41 conformation to bind histamine. Several H1 receptor models have demonstrated this hydrogen bond, however a direct interaction with the imidazole ring of histamine has not been ruled out The affinity for histamine is reduced at T5.42A by approximately tenfold (Jongejan & Leurs, 2005; Fang, Travers, and Booth., 2009 prepublication) K5.39A inter acts directly with histamine at the Nof the imidazole ring (Jongejan & Leurs, 2005; Booth et al., 2008) Mutating this residue to a nonfunctional alanine residue resulted in an 8fold reduction in histamine affinity, illustrating that K5.39 is important for maintaining proper histamine binding, but the interaction is not quite as strong as N5.46 or D3.32 (Fang, Travers, and Booth., 2009 prepublication) There is evidence for an auses about a 1000fold reduction of histamine affinity (Wieland et al. 1999; Jongejan & Leurs, 2005) Histamine makes its strongest HH1R interactions with polar residues TMDs 3 and 5 and appears to have a strong aromatic interaction in TMD 6. Interestingly, F6.55A produces a large effect on histamine s potency while nearby aromatic residues associated with the rotamer toggle switch and antagonist binding (F6.52) have little effect on histamines ability to activate the receptor (Wieland et al. 1999; Br uysters et al., 2004; Jongejan & Leurs, 2005) Mepyramine, also known as pyrilamine, is the traditional radiolabel that has been used to identify H1receptors. It is an inverse agonist at H1 receptors, with a high affinity and a strong inverse agonist response. Due to its availability as a radioligand3H mepyramine has been studied at a wide array of point mutations and a great deal is known about its binding. D3.32, the highly conserved aspartic acid residue in TMD 3, is again critical for ligand binding. M utation to an alanine results in the complete
42 abolishment of receptor binding, demonstrating once again that the ionic interaction with the protonated amine is necessary for any ligand, agonist or antagonist, to bind to this aminergicreceptor (Wieland et al 1999; Bruysters et al ., 2004; Jongejan & Leurs, 2005; Fang, Travers, and Booth., 2009 prepublication) By comparing the structure and function of histamine and mepyramine, one can and should conclude that the similarities in binding end there. Histamine is a small, polar, heteroaromatic compound with few degrees of rotational freedom. Mepyramine, while pos sessing the critical protonated amine pharmacophore, is a much larger, bicyclic compound, with many rotatable bonds that provide the ligand with significant flexibility. An early H1 model, with homology to bacterial rhodopsin, predicted that mepyramine would interact with aromatic residues in TMD 4 and 6(Wieland et al. 1999) In contrast to this, histamine is known to form its strongest contacts (D3.32 aside) with a series of polar residues in TMD 5. The residues originally predicted to be significant in the binding of mepyramine were designated Trp167 (W4.56), Phe433 (F6.52), and Phe436 (F6.55). These three predicted interactions turned out to be conserved residues across all known H1 receptor sequences, suggesting they play some integral role in H1receptor binding and/or activation(Wieland et al. 1999) While the Ballesteros numbers assigned here correspond to the same amino acid in the human receptor, the original model was based upon the guinea pig H1sequence so the sequence numbering does not exactl y match the human sequence. In the human receptor, these three residues correspond to Trp158, Phe432, and Phe435, respectively By referring to these positions by the Ballesteros number, rather than their sequence
43 number, it is possible to avoid the confus ion that occurs when cross referencing results across differing species. Invoking saturation binding assays with the alanine mutants of these identified residues allowed Wieland et al. to confirm their hypotheses through experimental results. When W4.56 wa s mutated to an alanine, a complete loss of binding was observed for 3H mepyramine. This strongly suggests that this position and residue are critical to the binding of this H1 inverse agonist and has been confirmed by mutagenesis studies at the HH1R as we ll (Fang, Travers, and Booth., 2009 prepublication) (Jongejan& Leurs, 2005) It is interesting to note there have been no residues identified in TMD 4 that have been directly linked to histamine binding, suggesting there are different binding modes for H1 agonists and inverse agonists. The F6.52 and F6.55 residues also were predicted to interact strongly with H1 inverse agonists and the 3H mepyramine saturation studies revealed similar results. The F6.52 position is located at the rotamer toggle switch in TMD 6, given its role in receptor activation, it is logical to conclude F6.52 would have strong interactions with H1 ligands. F6.52A saturation isotherms revealed that 3H mepyramine was unable to bind appreciably (no difference when compared to mock transfections) to the F6.52A mutated H1 receptor. This result has been confirmed at the HH1R by our lab and several others, indicating F6.52 is critical for the binding of antagonists to the human H1 receptor (Bruysters et al., 2004; Bakker et al., 2004; Fa ng, Travers, and Booth., 2009 pre publication) An intriguing result came from Wielands same paper when F6.52A was examined in the inositol phosphate functional assay. Despite an inability to bind 3H mepyramine, F6.52A produced a functional response that was identical to the WT
44 receptor when histamine was administered as the agonist (Wieland et al. 1999) ; lending further credence to the hypothesis of at least two unique binding pockets, one for agonists and one for inverse agonists, at the H1 receptor. T he F6.55 residue did not produce as dramatic a decrease in binding as the previous two mutations, however in the case of the guinea pig H1 receptor the affinity (KD) was still reduced from approximately 1nM at the WT to > 15 nM at F6.52A (Wieland et al. 1999) When the HH1R is examined, the reduction is slightly lessened, yielding an affinity of 4.76 0.8 nM and a P value of 0.011(Booth Lab unpublished data). F6.55 does seem to have relatively strong interactions with mepyramine, as indicated by the stat istically significant p value, however these aromatic interactions are weaker than W4.56 and F6.52. Having mentioned second generation histamine antagonists briefly, a discussion of their unique binding profile is warranted here. In general, second generation antihistamines are more selective for the H1 receptor and lessen the severity of the sedation commonly associated with first generation drugs (Wieland et al. 1999) The decrease in side effects was due to more selective H1 targeting and a simple structural change that alleviated much of the sedative action possessed by these ligands. The introduction of a polar carboxylic acid moiety into the ring structures of triprolidine and hydroxyzine, resulted in the development of acrivastine and cetirizine, respectively. These ligands were able to decrease the amount of drug that crossed the highly lipophilic blood brain barrier, preventing sedation from occurring. The introduction of the carboxylic acid moiety had a second impact on the overall profile of these new ligands, it created a second strong contact point for anchoring the
45 ligand to the H1 receptor. As predicted by the modeling studies performed by Wieland et al., the carboxylic acid of cetirizine and acrivastine has an ionic interaction with the basic side chain of K5.39. A litany of studies confirmed this result, with mutagenesis data indicating that the affinity of triprolidine (parent compound) is not affected by K5.39A, while the affinity of acrivastine was reduced fifty fold(Wieland et al. 1999; Jongejan & Leurs, 2005) H1 GPCR Dimerization Based upon the current evidence available, it is now relatively well accepted that GPCRs have the ability to exist as monomers, dimers, and even higher order oligomers (Gurevich and Gurevich, 2007). The diverse library of ligands produced by our chemistry lab has provided unique stereochemical probes to carry out a wide array experiments to help characterize the structure and function of the human histamine H1 receptor. The ligand called ( ) trans PAT has been shown to activate the AC/cAMP pathway through G, as well as being an inverse agonist at G. ( ) cis PAB has been shown to activate G and the PLC second messenger pathway (Fang, Travers, and Booth., 2009 prepublication) These two ligands alone provide very interesting tools for probing the molecular determinants involved in activation and signaling of the HH1R. However, previous work from our lab and its collaborators has determined that ( ) trans PAT binds to a specific subpopulation of H1 receptors. Upon further investigation, it was proposed that ( ) trans PAT preferentially labeled dimers while the traditional radioligand3H mepyramine was able to label both monomers and dimers (Booth et al., 2001,2004). Currently, there are two dominant hypotheses about how GPCR dimers and oligomers are formed. These are contact dimers (B) and domainswapped dimers (C)
46 as illustrated in figure 18 ( adapted from Booth et al., 2004) It is proposed that in contact dimers certain transmem brane domains interact, through mainly hydrophobic residues, to form the necessary stabilizing interactions while maintaining their individual binding pockets. In domainswapped dimers, two full transmembranes of each receptor are exchanged forming two com plete chimeric receptors, each of which maintains a binding pocket for the ligand Based upon a clever series of mutagenesis experiments it was determined that the ligand ( ) trans PAT prefers to bind to domain swapped H1 receptors ( Booth et al., 2004) Co expression of two mutations, D3.32A and F6.52A, both of which individually result in a complete loss of radioligand binding, were able to regenerate a binding site for both 3H ( ) trans PAT and 3H mepyramine. This regenerated site possessed an identical KDfor the two radioligands at the WT HH1R however it possessed a significantly lowered BMAX. At the WT receptor, mepyramine possessed a BMAX that was 7x larger than ( ) trans PAT. The differences in BMAX between the two ligands are not a cause for concern and have been previously reported, along with the fact ( ) trans PAT is known to bind to a subset of H1 receptors, while mepyramine is able to label all H1 receptors (Booth et al, 2001). Coexpression of D3.32A and F6.52A significantly lowered both BMaxvalues to 0.33 pMol /mg protein, a 10fold reduction for PAT and nearly 100fold reduction for mepyramine. Actual BMAX and Ki values are listed below in table11 (Booth et al, 2004). When the coexpressed mut ant is considered, it stands out that the Bmax values are identical. This suggests that the two ligands that normally occupy different cross sections of the H1 receptor population, are now sharing the same unique binding site.
47 Regeneration of binding with two nonfunctional mutations is quite significant, because it can only take place through dimerization, and more specifical ly domain swapped dimerization. Domain swapped dimers are similar to chimeric receptors in that an individual receptor subunit is com posed of helices from two separate receptors. Because of this unique interaction, it is possible to quarantine the nonfunctional point mutations into a single subunit, leaving the other half of the dimer unperturbed. To demonstrate this point, a schematic from this paper is illustrated below in Figure 1 9 (Bakker et al., 2004) Possible receptor combinations of contact and domain swapped dimers are displayed, with the TMDs contain in g the mutation highlighted in gray and nonbinding receptors marked with an X. Receptors that bind ligand are marked with an L through their center. Contact dimers have no chance of forming a functional receptor, as one of the mutations is always present on each side if the dimer. Because of their unique structure, domain swapped dimers are able to place both mutations in one half of the dimer, allowing the other half to bind radioligand normally. This argument is strengthened by the fact that the BM ax valu es of the comutation are greatly reduced, from wild type (WT), suggesting that extenuating circumstances are necessary to recreate a binding site. In this case, the receptor must assemble as a domainswapped dimer with the caveat that TMDs 3 and 6 must b e from different receptors and be trafficked to the membrane surface. Current GPCR Theory Our understanding of GPC Rs has come a long way since the lock and key, bimodal switch and law of mass action models for GPCR activation were first proposed. Significant discoveries such as constitutive activity, inverse agonism, and
48 allosterismhave quickly forced GPCR models to become increasingly complex. Culminating in the extended ternary complex model and the cubic ternary complex model, proposed by Samama et al in 1993, and Weiss et al in1996, respectively (Christopolous & Kenakin, 2002). The extended terenary complex model was significant because it addressed the possibility of allosterism and accounted for constituitive activity. The cubic terenary complex model is more sophisticated from a thermodynamics standpoint, however its bulky nature significantly hinders its laboratory applications. Recent evidence suggests that the GPCR model may need to be adapted yet again. Modeling results and recent data suggest that there is not one agonist, basal, or inverse agonist conformation of the receptor, instead there are thousa nds. It seems that every ligand induces a unique receptor conformation, with unique functio nal properties (Kobilka & Deupi, 2007; Bhattacharya et al., 2008). Each of these models will be discussed in detail in the coming sections. Law of Mass Action and Early GPCR Models The simplest GPCR models followed enzyme kinetics and ion channels suggest ing that an agonist bound to a receptor causing a physiological response in the process. This concept is called the law of mass action and was defined by the formula listed below. This equation was governed by the equilibrium association constant Ka, whic h governs the association and dissociation of the ligand from the receptor A represented the agonist molecule, R the receptor, and AR the ligand receptor complex as demonstrated below in figure 11 1 T he discovery of intracellular G proteins forced this model, which was originally designed for ion channels, to adapt At this time it was believed that the agonist bound
49 to the receptor, which could isomerize into an active form which was able to function or couple to a G protein. This adaptation is illustrated below in figure 11 2 Ternary Complex Model Further understanding of GPCRs demonstrated that an agonist could bind to a receptor, yet produce no functional effect if it was not attached G protein.Thisobservation led to the original ternary complex model, proposed by DeLean et al in 1980.It is significant to note that the only active species in this model is the ARG complex, indicating that the receptor must be complexed with a ligand and the G protein in order to exert its function (Christopolous & Kenakin, 2002) The ternary complex model is illustrated in figure 11 3 Extended Ternary Complex Model When constitutive activity was first demonstrated conclusively by Costa and Hertz in 1989, it was clear that the pr evious theories would not be able to describe the complex GPCR activation process accurately Around the same time, it was shown that the binding of GTP to the G protein (the critical activating step) resulted in a conformational change reducing the affini ty of the receptor. Thus, receptors demonstrate differing affinities for both ligand and G protein, depending upon which bound to the receptor first. This resulted in the emergence of cooperativity factors ( , and ), which are constants that affect the equilibrium involved in agonist receptor and receptor G protein interactions. The use of cooperativity factors allowed the equilibrium constants to take into account what species was able to bind to the receptor first By incorporating the law of mass act ion and the ternary complex theory into one formula, and allowing for constitutive activation, Samama et al. were able to create the extended ternary complex model in 1993. This model expanded on the ternary complex model
50 with the addition of cooperativity factors and allowed the receptor to adopt an active conformation in a spontaneous, or agonist mediated manner before coupling to the G protein. This expanded model still uses the equilibrium constants KA and KG for the bin ding of the agonist A and G protein G, to the receptor. It adds the equilibrium constant L for the isomerization of the receptor from the inactive R conformation, to the active R* conformation. The cooperativity factors , and constants that modify these equilibrium constants. As mentioned above, the observation that agonist or G protein precoupling was able to alter the binding affinity of the latter necessitated that there be unique equilibrium constants for each transition in the model. For example, the equilibrium between the inactive R conformation and the R* representing basal activity, would not be governed by an identical equilibrium as the AR to AR* transition. The equilibriums may be similar, but the binding of an agonist will alter cooperativity factor. The incorporation of these cooperativity factors made the extended ternary complex model incredibly versatile and represented a signi ficant leap forward in GPCR theory. There are additional complex models like the cubic ternary complex model and the quaternary complex model, which are more thermodynamically correct yet both models are incredibly complex and become unwieldy in a laborat ory setting (Christopolous& Kenakin, 2002) For these reasons, it has been the extended ternary complex model that has been used as the template for GPCR studies. LITicon The extended ternary complex model significantly advanced our understanding of the complex processes that govern GPCR binding and function. However, as
51 technology has continued to progress there have been findings that call into question even the basic tenants of the ternary complex model. The extended ternary complex assumes there are onl y several conformations that a receptor must choose. Making this assumption allowed researchers to create a model that described their experimental results and helped elucidate the major steps in G protein activation. Yet, the variety of functional states available to GPCRs including full, partial, neutral, and inverse agonism, as well as allosteric agonists, positive and negative allosteric modulators, and ago allosteric modulators seems to suggest a litany of unique conformations exist. Describing all of these functional properties in the context of the ternary complex model yields the quaternary complex model that was alluded to earlier (Christopolous & Kenakin, 2002) This model becomes unwieldy rather quickly, with equilibrium constants that can have up to seven unique cooperativity factors. It should be mentioned that despite its unwieldiness, the quaternary ternary complex model is able to describe allosteric receptor interactions accurately, with unique equilibrium constants for each conformation. Despite this incredible accomplishment, a question that needs to be addressed is whether or not agonists all induce an identical receptor conformation. A 2007 paper made use of a novel molecular modeling technique called LITicon, to address this question. LITicon is an acronym for L igand I nduced T ransmembrane Rotat i onal Con formational Changes. Traditional molecular modeling freezes the backbone of the transmembrane domains and allows free movement to the amino acid side chains to interact with the ligand. T his artifact creates a litany of problems for extrapolating modeling to experimental results. First, the binding of a ligand becomes highly
52 dependent upon the crystal structure template that was selected. As defined by the extended ternary complex model, a ligand possesses different affinities for the active and inactive state of the receptor. It follows that the way the ligand binds, especially with frozen TMD backbones that restrict movement, will differ between the active and inactive crystal structures This presents the researcher with an unavoidable choice that will bias their results. Freezing the backbones of the helices would sig nificantly hinder the movement of a receptor that is renowned for its dynamic nature and flexibility. If the backbones of the TMDs were frozen in GPCRs there would likely be no observed constitutive activity, and one of the defining fe atures of GPCRs would not exist; thus freezing the TMD backbones results in an unnatural environment for ligand binding. In addition, f reezing the backbones prevents molecular modelers from observing ligand specific effects on molecular switches (Bhattacharya et al., 2008) Both the ionic lock and rotamer toggle switch require significant movement from TMDs that is difficult to replicate with a rigid helical backbone. The ionic lock between TMDs 3 and 6 is lost during the activation process as the helices pull away from each other (Prioleau et al., 2002; Shi et al., 2002 Kobilka & Deupi, 2007) The rotamer toggle switch is a more subtle movement that is governed by the proline residue 6.50. Upon activation, a series of stabilizing interactions is lost and the proline kink becomes the dominant factor in the TMD structure, causing a pronounced bending of the cytoplasmic end of helix 6 (Prioleau et al., 2002) LITicon addresses these pitfalls associated with traditional molecular modeling and allows a ligand to determine the lowest energy receptor conformation. The process is begun by identifying the helices, or TMDs, that are perturbed during the docking
53 process. Once this process is completed and the ligand is located within the orthosteric binding pocket, each helix is rotated 180 in each direction. The energy of each rotational combination was then compared to the energy of the ligandfree rece ptor. The resulting differences in energy were tabulated for each conformation, and the lowest energy receptor ligand complex was assumed to be the proper ligand binding conformation (Bhattacharya et al., 2008) This technique produced several advantages over traditional molecular modeling the most critical of which is a flexible backbone for each helix. Reducing the constraints on each backbone allows the formation and breaking of new hydrogen bonds, as the ligand and receptor move towards equilibrium. This results in more accurate modeling predictions and molecular determinants. It also allows the ligandreceptor complex to trigger molecular switches, potentially allowing this model to attain the golden fleece for molecular modeling reliably predict ing t he functionality of ligands. An important caveat to this statement is that this model does not take into account G protein coupling and therefore cannot predict functional selectivity. However, when the norepinephrine bound LITicon receptor conformation was used to select a series of 60 2AR agonists from a pool of 10,060 compounds, a 38% increase in success rate was observed when compared to a ligand nave receptor (Bhattacharya et al., 2008) The increased hit rate for LITicon was not the only significant finding in this paper. When a diverse series of 2AR ligands were examined using LITicon, an extremely interesting series of results was noticed. The ligands screened included the full endogenous agonist norepinephrine, the partial agonists dopamine, sal butamol, and catechol, and the inverse agonist ICI118551. Each ligand that was studied using
54 LITicon produced a unique receptor conformation, and induced changes in the receptor that fit its fu nctional profile. Structures for these ligands can be found in figure 11 0 These conformations will be briefly discussed here, and will exclude the redundant D3.32amine interaction and instead focus on the remaining molecular determinants and the ionic lock and rotamer toggle switch. The full endogenous agonist norepinephrine showed strong hydrogen bonding interactions with a series of serine residues in TMD 5. These residues were S203, S204, and S207. When converted into Ballesteros numbers they correspond to 5.42, 5.43, and 5.46, all of which were implicated in the binding of norepinephrine through mutagenesis studies (Carmine et al., 2004; Bhattacharya et al., 2008) Interestingly, S204 and 207 were not initially located within the binding pocket of th e receptor, but after rotating the helices in LITicon these strong hydrogenbonding interactions were observed. There is an additional strong hydrogen bond that is observed with N293 (N6.55). To corroborate these findings both the rotamer toggle switch ( RT S ) and the ionic lock( IL ) are perturbed by norepinephrine binding, suggesting correctly that this ligand should be a full agonist. 2AR, was the next ligand examined.S5.43 and 5.46 again interacted with the p OH group on the catechol ring. Yet, there was no longer an interaction between S5.42 and the m OH on the catechol ring and no hydrogen bond with N6.55. Dopamine was able to disrupt the ionic lock, but only partially disrupted the RTS. The partial activati on of the rotamer toggle switch was determined by an intact hydrogen bond between N7.45 and C6.47, and a lack of a hydrogen bond betweenW6.48 and M5.54(Bhattacharya et al., 2008) Both of these interactions were observed with the norepinephrine bound receptor and further confirm that dopamine
55 and norepinephrine induce different conformations 2AR. With only weak 2AR confirming experimental results. Salbutamol was docked using the LITicon method and was shown to disrupt the ionic lock, but not the RTS. This finding already hinted at a different receptor conformation, but a detailed assessment of the molecular determinants was necessary. Salbutamol made strong contac ts with the three serines in TMD 5, but was unable to interact with N6.55, the residue closest to the RTS. The loss of this interaction prevented the hydrogen bonding pattern mentioned above that is indicative of RTS activation(Bhattacharya et al., 2008) With only the ionic lock disrupted, one would 2AR agonist, just as is observed in vitro. The weak partial agonist catechol, which is simply 1,2di hydroxy benzene, was next to be examined. Since it was lacking the protonated amine moiety, there would be no ionic interaction to hold the ligand in the binding pocket. Instead, catechol interacts with S5.46 and N6.55 by hydrogen bonding to these residues. There was no observed disruption of the IL, and only a weak interaction at the rotamer toggle switch that is unable to replicate the hydrogen bonding patterns caused by norepinephrine(Bhattacharya et al., 2008) With the ionic lock unperturbed and only a weak interaction with N6.55, affecting th e RTS it would be predicted that catechol would 2AR.This prediction mirrors catechols function and demonstrates another unique receptor conformation. The last ligand to be studied was the inverse agonist ICI115881. This compound made no contacts with TMD 5, and caused it to rotate in the opposite direction as the
56 full and partial agonists. It did form a hydrogen bond with N6.55, but it also hydrogen bonded to H6.58, a residue located only one helical turn below that residue. This two pronged hydrogen bond was only observed with this inverse agonist and prevented the rotation of TMD 6(Bhattacharya et al., 2008) With the ionic lock left intact and the RTS switch stabilized by the forked hydrogen bonds in TMD 6, the LITicon conformation predicted a less active receptor conformation for the salbutamol docked receptor. As an 2AR, it reduces the constitutive activity of the receptor. As this paper demonstrates, it seems that each ligand creates a unique drug receptor complex that is defined by the position and types of functional moieties on the ligand itself. This is in stark contrast to the ternary complex models, which propose a single receptor conformation for the inactive, active, and constitutively active states. It is more likely that GPCRs have a limitless number either of conformations that they can adopt at random or in the presence of a ligand. Ligands serve to stabilize specific conformations from t his subset that are able to produce an observable and reproducible physiological effect. These comments are not meant to trivialize the extended ternary complex by any means, as it has been and remains critical to understanding the minutiae of GPCR ligand binding assays. To the contrary, this theory seeks to build on the concepts put forth in ternary models, while painting the most accurate picture of GPCR activation. G oal of T his Thesis It is hypothesized that through a detailed characterization of the molecular determinants involved in ligand binding and function that a better understanding of the molecular events involved in H1 GPCR functional selectivity can be obtained and
57 exploited for drug design purposes. The goal of these thesis studies is to determine which H1 receptor amino acid residues are critical for PAT analog H1ligandbinding, as well as selectively activating the AC/camp vs. PLC/IP signaling pathways Investigating the mechanism of binding and activation for H H1Rs will provide valuable insight into future drug design for the treatment of psychiatric and neurological disorders. I nsight into how ligands influence H1 receptor activation of G proteins is expected to aid design of efficacious drugs with predictable and fewer adverse side effec ts. Information learned about drug discovery targeting the H1 GPCR can be applied to other relatedaminergicGPCRs (e.g., receptors for the neurotransmitters acetylcholine, dopamine, norepinephrine, and serotonin) with implications for both GPCR structure and function and drug discovery. Aim #1: Characterization of th e S tructure A ctivity R elationships for PAT A nalogue B inding at the W ild T ype HH1R Characterizing the structure activity relationships of PAT analogues synthesized in our lab is crucial to guiding future synthe s es and as a tool to guide our molecular modeling program. This aim means to determine the affinity of a variety of PAT derivatives at the WT HH1R receptor with a particular emphasis on stereochemistry Examin ing the binding modes of these varying ligands will provide crucial structural information about the binding pocket of the HH1R that will serve as a guide for future synthesis and molecular modeling experiments Aim #2: Characterization of the R ole of A mino A cid R esidue Y 5.48 in D imerization of the HH1 R and its I mpact o n L igand B inding and F unction. This aim tests the hypothesis that the HH1 GPCR transmembrane domain (TMD) 5 amino acid residue Y5.48 is involved in receptor dimerization that affects ligand binding and/or activation of the receptor. The ( ) trans (2S,4R) PAT compound is known to bind
58 to domainswapped dimers, and this aim addresses the possibility that Y5.48 is involved in the domainswapped dimerization. Aim #3: Characterization of theRole of HH1R esidue Y7.53A in L igand B inding and A cti vation U sing F unctionally S elective PAT D erivatives Studies indicate that TMD 7 amino acid 7.53 plays a critical role in the signaling of the closely related 5HT2C receptor. This aim focuses on elucidating whether or not this effect is specific to 5HT2C receptors, or can be found in the HH1R as well. This aim will determine whether or not a particular signaling pathway and/or a particular ligand are affected by this point mutation A novel addition to this HH1R investigation, is the con cept of functional selectivity. In particular, the focus will be on the efficacy of histamine at Y7.53A, and the effects this mutation has on the ability of the HH1R to couple to G S and G Q.
59 Figure 1 1 Generic structure of a GPCR Figure 1 2 GPCR a ctivation demonstrating guanine exchange. D Y 107 W 158 K 191 T 194 N F Y 198 W 428 F F Y 432 Y 458 Y 468 108 435 199 200 431
60 Figure 1 3 Ballesteros n umbering proceeds as indicated by the arrows onthe generic GPCR structure Figure 1 4 Phylogenic t ree for a minergic GPCRs that are closely related to the H1 receptor.
61 Figure 1 5 Representation of both signaling pathways for the production of A) cAMP mediated via G. B) I nositol triphosphate production mediated via G.
62 Figure 1 6 Histamine catabolism in the central n ervous s ystem Figure 1 7 The chemical structures of first generation (top row) and second generation (bottom row ) antihistamines.
63 Figure 1 8 P roposed m odels of GPCRd imerization .This diagram illustrates GPCR monomers (A), contact dimers (B), and domainswapped dimers (C). Figure 1 9 P ossible combinations of contact ( middle ) and domain swapped dimers (bottom) for D3.32A and F6.52A mutations.
64 Figure 1 1 0 Structure s of the 2adrenergic receptor ligands that were studied using LITicon. Compounds illustrated are: salbutamol (top left) dopamine (top right) catechol (left center), ICI 118551 (right center) and norepinephrine (bottom) Law of mass action: A + R AR Response Figure 1 1 1 Illustrates t he law of mass action A + R AR A + R AR+G Figure 1 1 2 Indicates the changes that were made to the law of mass action in order to incorporate G proteins. Figure 1 1 3 Demonstrates the ternary complex model of G protein activation
65 Figure 1 1 4 When constitutive activity was discovered, the ternary complex model was modified to incorporate this new concept. The result ant extended ternary complex model that is illustrated here. Table 1 1 KIand BMax values for 3H mepyramine and ( ) trans PAT for WT and D3.32A/F6.52A H1 receptors (Bakker et al., 2004) Receptor 3 H mepyramine KD (nM) 3 H mepyramine Bmax ( pMol /mgprotein) 3 H ( ) trans PATKD (nM) 3 H ( ) trans PAT Bmax ( pMol / mgprotein ) H 1 RWT 1.2 0.1 21 4 1.2 0.4 3.4 1.0 H 1 R D3.32A + F6.52A 1.8 0.1 0.34 0.1 3.0 0.6 0.32 0.1
66 CHAPTER 2 CHARACTERIZATION OF THE STRUCTURE ACTIVITY RELATIONSHIPS FOR PAT ANALOGUES AT THE WILD TYPE HH1R Rationale for Undertaking These Studies Our lead compound, ( ) trans PAT possesses therapeutic potential for the treatment of Parkinsons Disease, via activation of the HH1R receptor which leads to increased tyrosine hydroxylase activity and dopamine synthesis in the brain (Moniri and Booth, 2006). HH1Rs also have been implicated in the etiology of insomnia, depression, Alzheimers disease, schizophrenia, and epilepsy (Yanai& Toshiro, 2007; Haas et al., 2008). In addition, ( ) trans PAT has potential as a therapeutic agent for the treatment of psychosis, obesity, and drug addiction via interactions at the 5HT2GPCRfamily (Booth et al., 2009). By conducting a thorough survey of the PAT ligands that have been synthesized in our laboratories i t is hypothesized that functional moieties can be identified that confer selective binding to the H1 receptor. Previous PAT studies examining structure activity relationships ( SAR ) used racemic compounds (Bucholtz et al., 1999). Advances in a symmetric synthesis and HPLC have allowed resolution of PAT enantiomers and an updated SAR including a litany of new PAT derivatives, is warranted. Materials and Methods Chemicals S ynthesis of (+) and ( ) trans PAT, (+) and ( ) cis PAT, (+) and ( ) trans PAB, and (+) and ( ) cis PAB are previously reported (Wyrick et al., 1993; 1995) Histamine, as the dihydrochloride salt, was purchased from SigmaAldrich (St. Louis, MO). The H1 radioligand [pyridinyl 5 -3H] pyrilamine (mepyramine; specific activity 30.0 Ci/m M ol) was purchased from Amersham Biosciences (GE healthcare, Piscataway, NJ)
67 and myo[2 -3H (N)] inositol (specific activity 18.5 Ci/mMol) was purchased from Perkin Elmer (Waltham, MA). Unless otherwise noted, all other compounds were obtained in highest purity from SigmaAldrich. Cell C ulture and T ransfection The cDNA encoding the WT human histamine H1 receptor was purchased from UMR (Rolla, MO). The H1 K5.39A T5.42A, N5.46A Y5.48A, Y7.53A point mutations were prepared using the WT cloned deoxyribonucleic acid ( cDNA ) subcloned in the pAlter plasmid (Promega), according to the manufacturers protocol (Altered Sites II, Promega). Mutations in the cDNA were verified by deoxyribonucleic acid ( DNA ) sequencing using the dideoxy chain termination method. WT and point mutated cDNA s were subcloned into the expression plasmidpcDEF3 (Goldman et al., 1996). Human embryonic kidney 293 cells (HEK 293; from ATCC, number CRL1573) were maintained in Eagle minimum essential medium (MEM) with 10% fetal bovine serum and 2 mM Lglutamine, 0.1 mM nonessential amino acids, 1.5 g/L sodium bica rbonate, 1.0 mM sodium pyruvate. The cDNA of wild type and mutants of human histamine H1 receptors in pcDEF3 expression vectors was prepared with Promega Wizard PlusMidipreps DNA Purification System (Promega A7640). Transfection was carried out with Lipofectamine 2000 (Invitrogen, CA) by following manufactures procedures. Briefly, approximate 13 x 106 cells were seeded onto a 100mm tissue cultur e dish and allow ed to grow to 8595% confluence, then transfected by 12 g of plasmid DNA mixing with 50 l of Lipofectamine 2000 for 24 hours. Expression was allowed for another 24 hours in growth medium without antibiotics. Membranes were harv ested in icecold 50 mM Na2/K
68 phosphate (Na2HPO4/KH2PO4) buffer, pH7.4 ( a ssay b uffer) after total 48 hours transfection and expression. H H1R B inding A ssays HH1R saturation and competition binding assays were performed using membrane homogenates prepared from the transfected HEK cells. [3H] mepyramine was used to label the H1 receptors (Booth et al., 2002; Moniri et al., 2004). Briefly, 48hrs following transfection, cells were harvested and homogenized in 50 mM Na2+/K phosphate buffer, pH7.4. The homogenate was centrifuged at 35,000g for 10 min and the resulting membrane pellet was resuspended in assay buffer. Protein concentration was determined by BioRad protein assays. For saturation binding assays, membrane susp ension containing 5 g total protein was incubated with 0.625 to 10.0 nM [3H] mepyramine in a total assay buffer volume of 250 l. Non specific binding was determined in the presence of 1 M triprolidine. Competition binding assays were conducted under the same conditions using 1.0 nM [3H] mepyramine (~KD concentration). R adioreceptor binding assay mixtures w ere incubated for 30minute s at room temperature, with termination by rapid filtration through Whatman GF/B filters using a 96well cell harvester (Tomt ec, Hamden, CT). The membranebound mepyramine retained on the filter discs was quantified by liquid scintillation spectrometry. Data were analyzed by nonlinear regression using the sigmoidal curvefitting algorithms in Prism 5 .0 2 (GraphPad Software Inc., San Diego, CA). Ligand affinity is expressed as an approximation of Ki values by conversion of the IC50 data to KIvalues using the equation KI= IC50/1 + L/KD where L is the concentration of radioligand having affinity KD (Cheng, 1973). Each experimental condition was performed in triplicate and
69 each experiment was performed a minimum of three times to determine S.E.M. Statistic analysis was carried out with Prism 5 .0 2 using t tests (and nonparametric tests), unpaired t test and ANOVA. P value > 0.05 is considered as not significant, 0.01
70 in Parkinsons disease (PD). In this disorder, there is a pronounced dying off of the dopaminergic neurons within the substantianigra. The loss of these neurons, and the dopamine they produce, leads to a disconnect between the brain and the muscles causing resting tremors, bradykinesia, a pronounced stooping posture, loss of balance, akinesia, and depression (Zhou et al 2008). Since neuronal death permanently pr events further synthesis of dopamine, PD is classified as a neurodegenerative disorder. B y stimulating TH with small molecules, like PAT and PAB, it is believed the dopamine output from the remaining neurons can be increased, resulting in a potential ameli oration of the symptomology of PD. In order to assess the ability of each compound to affect H1 receptor signaling one mu st first determine how well the compounds of interest are binding to their target. This is accomplished through competitive binding as says, where the ligand in question displaces 3H mepyramine from a cloned HH1R transiently expressed in a cell line. The cell lines used in these experiments were either CHO K1 (CHO, ATCC CCL 61) or HEK 293 (HEK, ATCC CRl1573) cell lines. UnsubstitutedP arent PAT C ompounds Significant effort has been placed upon the synthesis and isolation of the enantiomers of cis and trans ()PAT. Enantiomeric separation was originally performed by recrystallization with camphor sulfuric acid, as described in our prev ious papers (Booth et al., 2002; Moniri et al, 2004) As more compound was needed for binding assays, functional, and animal studies it quickly became clear that a more efficient and higher throughput method of enantiomeric purification was necessary. This was made clearly evident by the fact that the resolution of the more potent () trans compound produced a yield of less than 10% under recrystallization conditions. By employing a
71 chiral polysaccharide column (Kromasil, AkzoNobel,Brewster, NY) it was pos sible to increase yields for the more potent trans diastereomers, while reducing the need for labor intensiverecrystallizations ( Booth et al., 2009) It is important to note that some of the data presented under this specific aim consists of racemic compounds. These compounds either had binding data previously reported, were difficult to separate by any means (Cl OH and di OH PATs), or had affinities that were too low to warrant further investigation (NH2PATs). It is important to keep in mind that these compounds are 50% mixtures of two different enantiomers, making any comparison beyond simple structure activity relationships rather difficult. However, some of the data provided by these racemic mixtures has been instrumental in st reamlining the design of future PAT ligands and will be discussed in the following pages. The general trend in HH1affinities for PAT stereoisomers is that the ( ) enantiomer is more potent than (+) and the trans configurationis more potent than the cis configuration T he affinities for the four PAT stereoisomers are 2S 4R ( ) trans > 2 S 4S ( ) cis > 2S 4S (+) trans > 2R 4R (+) ci s. The structures for these compounds are in figure 2 1 The Ki values for each of these compounds and their pvalues in reference to the lead compound ( ) trans PAT are in table 2 1 and representative binding curves are in figure 2 2 The results suggest that stereochemistry is a crucial factor in PAT binding to the HH1 receptor and confirm that the (S) stereochemistry at the amine as the most important structural feature for determining H1 receptor affinity (Bucholtz et al., 1998) T he affinity of these two pairs of enantiomers varies from 1 nM to 100 nM. The results indicate that the ligand binding pocket for PATs at the HH1 receptor is quite
72 selective and can easily differentiate between two enantiomers of the same compound. This effect previously was observed fort h enantiomers of cetirizine, (R)levocetirizine and ( S ) cetirizine at the HH1R (structures in figure 23 ) The Ki values for levocetirizine and its enantiomer at the HH1R, ranged from 3.16 nM for (R)levocetirizine, to 79.4 nM for (S ) cetirizine ( Gillard et al., 2002) These results set a precedent that the binding pocket of the HH1 receptor is highly stereoselective and lend credence to this PAT binding dat a These results fit closely with previously determined Ki values in our lab and with previously published data using both guinea pig and human H1 receptors (Bucholtz et al., 1998; Booth et al., 2002) Ortho S ubstituted PATs A series of o substituted PATs were synthesized t o probe the structure activity relationships of the pendant phenyl ring. These compounds were () trans o Cl and o CH3PAT whose structures can be found in figure 24 As these were older ligands, the binding data consists of the racemic mixture of the ( ) and (+) trans enantiomers. Both compounds exhibit a pronounced reduction in affinity when compared with ( ) trans PAT. Previous modeling studies in our lab reveal that the pendant phenyl moiety of ( ) trans 2S 4R PAT possesses aromatic contacts with W6.48, Y6.51, and Y7.43. The ()trans o ClPAT compound had a Ki value of 24.82 2.7 nM, a tenfold reduction in affinity compared to its unsubstituted parent compound. The ( ) trans o CH3PAT compound has an affinity of 12.7 0.6 nM. Insertion of the methyl group or a chlorine at the ortho position seems to produce a negative steric effect with these residues; resulting in the observed drop in affinity compared to the parent ligand. It can be concluded that the effect is steric, as the opposing electr onegativities of the chlorine
73 atom and the methyl group, would have opposing effects on any nearby polar residue. Since both substituents did the same thing, it follows that the effect is likely steric. Meta S ubstituted PATs Following up on possible pendant phenyl derivatives our chemists synthesized a series of meta substituted PAT compounds trans m F, m Cl, and m Br ( structures of each are found in figure 25 ) The affinity of the m F enantiomers will be discussed first. The ( ) trans m F P AT compound was shown to have a potency similar to that of our lead compound. The Ki value obtained ( ) trans m F PAT for was 3.03 0.26 nM. For (+) trans m F PAT the Ki value was increased approximately thirty fold (compared to the [ ] enantiomer to 93. 56 15.2 nM. The compiled Ki values for binding affinities of the meta halogenated compounds are found in table 22 This ser ies of compounds demonstrates a decrease in stereoselectivity when compared to ( ) and (+)trans PAT, which demonstrate a well documented 10fold difference in enantiomers. The pronounced reduction in affinity for the (+) enantiomer of the metahalogenated series was a n unexpected result, s. The pvalue for the comparison of ( ) and (+)trans m F PAT is P =0.0147, indicating a significant difference in the binding of the two enantiomers. In keeping with the compact size of the fluorine atom, the m F substituent has little effect on the affinity of the ( ) isomer, reducing the affinity from 1.95 nM to 3 nM (P = 0.2926) The (+) trans PAT compound has an affinity of 30 nM, that is reduced to nearly 100 nM with the introduction of the meta fluoro substituent found in (+) m F PAT (P = 0.0262) Given that the affinity of ( ) trans m F PAT was unaffected, the result for the (+) m F PAT enantiomer is rather surprising. T he significant reduction in affinity for the (+)enantiomer, coupled with an unperturbed affinity for the ( ) isomer is suggestive of unique amino acid interactions for
74 the two stereoisomers. Knowing that the pendant moiety of the ( ) trans PAT scaffold interacts with W6.48, Y6.51, and Y7. 43 it seems that accommodating the small fluorine atom is not problem for this aromatic cluster. In the case of the (+) trans p F enantiomer, the drastic reduction in affinity is suggestive of a strongly negative interaction with a completely different set of residues. These opposing effects on affinity for the enantiomeric pair strongly suggest that the pendant phenyl moiety of each stereoisomer is seeing a distinct region of the HH1R. In order to glean further information about the HH1binding pocket, (+ ) and ( ) trans m ClPAT were examined in a series of binding assays. The resulting Ki values were 10.19 2.4 nM for the ( ) isomer and 79.48 23 for the (+) enantiomer Even though chlorine atomis larger than fluorine, the affinity of (+) trans m F PAT and (+)_trans m ClPAT are similar Comparing the (+) m Cl isomer to its parent enantiomer demonstrates a significant reduction in affinity with a pvalue of 0.0262. The ( ) trans PAT and ( ) trans m Cl PAT compounds also shows a significant reduction in affinity (P = 0. 0057) This trend illustrates that insertion of a meta chlorosubstituent into either the ( ) or (+) trans PAT scaffold results in a reduction in affinity when compared the parent ligand In order to confirm these trends, the affinity of a third meta halogen compound was required. The (+ ) and ( ) trans m Br PAT were synthesized B romine is the next largest halogen atom after chlorine and fluorine, and possesses a reduced electronegativity when compared to fluorine and chlorine. The Ki values for the (+) and ( ) isomers of trans m Br PAT were determined to be 62.52 10.1 and 27.94 1.5 nM, respectively. TheKi values for the (+) parahalogenated PATs range from 94 nM for m F, to 62 nMfor
75 m Br and show no significant differences when compared to each other with ANOVA (P>0.05) When placed in their rank order from high to low affinity the results are (+) trans m Br PAT > (+) trans m Cl PAT > (+) trans m F PAT It is important to recall that the Ki value for (+) trans PAT is 30 nM, thus substituting a halogen atom in the metaposition does not provide HH1 ligands with enhanced affinity. I n contrast to the (+) trans meta halogens, the ( ) trans meta halogens illustrate a clear trend. The rank order of affinity of the ( ) trans compounds is ( ) trans m F PAT > ( ) trans m Cl PAT > ( ) trans m Br PAT As the halogen substituents on the pendant phenyl ring increase in size, the HH1 affinity of the ( ) trans meta halogenatedPATs decreases. Contrasting the effects of meta halogen substitution on the ( ) and (+) enantiomers of PAT, reveals two unique pictures. The (+) enantiomers seem to be unaffected by the size of the halogen substituent (P > 0. 05 between compounds), while the ( ) isomers show a significant reduction in affinity as the halogen size increases (PF Cl:=0.0103 and PCl Br=0.0321). These findings suggest that the ( ) and (+) enantiomers of PAT s are binding differently to the H1receptor and that the pendant phenyl moiety is an important determinant of the binding mode. Despite the intriguing lack of affinity differences with the (+) trans m substituted PATs it is important to keep a sense of scope, as none of the compounds examined improved the affinity of ( ) trans PAT; and therefore would not make suitable H1 ligands. Para S ubstituted PATs To further our understanding of the molecular determinants of PAT at the HH1R it was necessary to look at additional derivatives. In a search for more potent analogues, a series of compounds w ere synthesized with an array of substituents at the paraposition of the pendant phenyl ring. Included among them were a series of halogen-
76 substituted compounds, p F p Cl, p Br PAT, and a methyl substituent, p CH3PAT. The structures of these compounds are found in figure 2 6 The compiled competitive binding data for all of the parasubstituted compounds is located in table 23 Intriguingly, each of these para substituted compounds showed a reduced stereoselectivity between enantiomers. For quite some time, PAT enantiomers have illustrated approximately a ten fold difference in H1 receptor affinity with the ( ) being the more potent of the stereoisomers (Bucholtz et al., 1999; Booth et al., 2002) The p methyl PAT was the first para derivative that was resolved into its respective (+) and ( ) enantiomers and was the first compound investigated. Intriguingly, there is no dif ference in affinity regardingstereoselectivityfor the (+) trans vs. ( ) trans enantiomers (P = 0.19). Binding curves for the p substituted PATs are located in figure 2 7 This result was unique a mong the previous trends that were established for PATs and w as quite unexpected. Based upon the differing affinities for (+) and ( ) trans PAT at the H1 receptor (P = 0.015), and the lack of stereoselectivity between the p CH3PATs (P =0.19) it is clear that the interactions formed by th e pendant phenyl ring with the receptor are being perturbed. This is supported by the fact that the introduction of a small, lipophilic methyl group on the para position of this ring had opposing effects on the binding affinity of the ( ) and (+) isomers. The ( ) trans p CH3 i somer demonstrated a ten fold reduction in affinity, when compared to its parent compound (P=0.023) In contrast, (+) trans p CH3PAT exhibited a threefold increase in affinity from (+) trans PAT (P=0.03 7 ) The dichotomy between the effects of ( ) and (+) p CH3PAT when compared to their parent compounds suggests that the binding pocket of the (+) isomer has more space around the para position of the pendant phenyl ring, with which to
77 accommodate the extra methyl group. Contrary to this, the area surrounding the pendant phenyl of the ( ) isomer seems to be unable to accommodate the same moiety interactions with W6.48, Y6.51, and Y7.43 or by simple steric bulk is difficult to say without modeling results. Regardless, it is clear that the introduction of a p methyl group onto the pendant phenyl ring of the PAT scaffold causes significant changes in H1 receptor binding. The p ara F compound was the next ligands to be resolved into its enantiomers. ( ) and (+) trans p F PAT represent excellent ligands to further probe the binding pocket of the H1 receptor. Methyl and fluorine have approximately the same mass, but quite different steric and electronic properties. Fluorine is sterically smaller than its methyl counterpart is but is much more electronegative and creates a polar C F bond. Methyl has little effect on stereoelectronics when compared to a highly electr onegative fluorine atom that is capable of forming hydrogen bonds (Torrice et al., 2009) It was originally anti cipated that the F substituent w ould create additional hydrogen bonds with the receptor to increase its H1 receptor affinity. When the binding of the p F PAT enantiomers was examined, the results proved similar to the pmethyl derivatives. Again the first observation was the reduction of stereoselectivity between the (+) and ( ) isomers. Rather than the tenfold selectivity between the ( ) and ( +) enantiomers, only a twofold selectivity was observed for the H1 receptor (P=0.0098) In this case, the ( ) was still the more potent of the two isomers having an affinity of 4.065 0.5nM similar to ( ) trans m F PATs 3 nM affinity and not statistica lly different from ( ) 2S,4R trans PATs 2 nM affinity (P=0.0542) K eeping with
78 the trend established by (+) trans p CH3PAT, (+) trans p F PAT was significantly more potent than its parent compound by threefold, having an affinity of 8.268 0.6nM at the WT receptor (P=0.026) The insertion of a fluorine at the para position of the pendant phenyl ring produced nearly identical results to the methyl substituent in the same position. This result was rather surprising as some electronic effects were anti cipated from the highly electronegative fluorine atom. As the affinity for the ( ) p F isomer is nearly as potent as the parent compound ( ) trans PAT it could be gleaned that there was no additional hydrogen bonding taking place. It is intriguing that both (+) p CH3 and F PAT demonstrate a threefold increase in affinity from their parent compound, suggesting that their increase in affinity is not due to hydrogen bonding interactions. Perturbations involving the nearby aromatic residues in TM D 6 can likely be ruled out, due to the fact that methyl and fluorine substituents have opposing effects on electronic density in an aromatic ring. The methyl group would donate additional electron density, while the fluorine would pull electron density fr om the aromatic system weakening the aromatic interactions With these effects ruled out, it is likely that the increase in affinity for the (+)isomer is caused by favorable steric interactions or additional Van der Walls interactions which are only pos sible if the (+) isomer is seeing a different microdomain of the orthosteric binding pocket than its ( ) enantiomer Chlorine was the next substituent o n the paraPAT derivatives produced by our chemistry lab. Upon investigating their binding to the H1 receptor, it quickly became clear that the similar trend of reduced stereoselectivity would continue. The (+) and ( ) p Cl enantiomers showed no stereoselectivity, with Ki values of 6.08 0.75 and 8.38 1.1nM, respectively (P= 0.24 ). When the p ClPATs are compared to their parent
79 enantiomers, a familiar trend emerges. The affinity ( ) trans p ClPAT is reduced by about four fold from the parent (P=0.036 ), and the (+) trans p Cl isomer is five times more potent than its parent (P=0.022) These results sugg est that the (+) isomer is able to better accommodate a variety of structural changes at the para position than its ( ) enantiomer. A gain suggest ing that the ( ) and (+) PATs, and more specifically their pendant phenyl moiet ies are seeing different regions of the H1 binding pocket. (+ and ) trans p Br PAT were synthesized to investigate the steric and electronic effects of para substitutions on the pendant phenyl ring. Br is the largest of the halogen moieties that has been inserted at the para position so far, and should allow information to be gleaned about the amount of space surrounding the para position. The electronic effects should be similar to the chlorosubstituted isomers as there is some bondpolarity, but a lack of strong hydrogen bonding. The (+) and ( ) isomers yielded Ki values of 531.8 62.1 and 1058.2 53.0 nM, respectively and when compared yield a p value of 0.023. T he affinities for both compounds are reduced by approximately 100fold when compared to the other p halogenated PAT molecules Such a drastic reduction in affinity for both stereoisomers, is best explained by steric hindrance. The progression from p F to p Cl yielded only slight differences in affinity between the compounds and slight deviation from the 1 nM Ki o f our lead compound ( ) trans PAT. The lack of affinity differences between the p F / Cl and CH3substituted enantiomersstrongly suggests that hydrogen bonding is not involved in the way these ligands bind to the receptor. If hydrogen bonding were involved, one would expect the F PATs to possess a high er affinity, while the ClPATs would show a slightly reduced affinity, as they possess a reduced ability to hydrogen bond. This is
80 clearly not the case, as the F and Cl PATs are nearly indistinguishable in binding assays, meaningsteric hindrance offers a better explanation of the data As the halogens increase in size, a schism appears between the affinities of p Cl and p Br PAT. The atomic radii for halogens increase as one progress es down group VII in the periodic table. Fluorine the smallest halogen has a n atomic radi us of 0.0709 nM chlorines is 0.0994 nM and bromine is the largest with 0.1145 nM (Visual Elements Group 17: The Halogens Royal Society of Chemistry ) It seems that the binding pocket of the H1 receptor is tolerant of the size increases at the para position until bromine. Once this point is reached, neither the (+) nor ( ) isomers can accommodate the large bromine atom within their binding pockets, resulting in the observed reduction in affinity. It does seem that (+) p Br isomer has slightly more room in its binding pocket to accommodate the bulky bromine moiety, as evidenced by its higher affinity. However, the reduction in Ki is significant enough to rule out p Br as an effective derivative of the PAT scaffold. The stereoelectronics of the PAT derivatives examined varied from strongly electron withdrawing (F and Cl), to electron donating (CH3) and still produced increases in affinity for the (+) stereoisomer; suggest ing that there are no strong el ectronic interactions formed by these compounds If a polar amino acid were interacting with the (+) isomer there would have been a weaker or no gain in affinity for (+) p methyl PAT. The same logic can be used to rule out additional aromatic interactions If an aromatic amino acid was interacting with the pendant phenyl of the (+) isomer diverging effects on affinity would have been observed for (+) trans p Cl/F PAT and p CH3PAT, as their substituents have opposi ng effects on aromatic electron density. C l and F would have
81 pulled electron density away from the ring, while the CH3 would have donated electrons to the aromatic system. With these two interactions ruled out, it seems likely that the differences in binding are occurring due to steric and/or Van der Waals effects. Perhaps the simplest explanation is that there is more room to accommodate a parasubstituted functional moiety at the region of the receptor where the pendant phenyl ring of the (+) trans isomers is binding. However it is important to note that this binding pocket for the (+) trans isomers does have steric limitations, as evidenced by the dramatic reduction in affinity observed for (+) p Br PAT This is in contrast to the ( ) para substituted isomers, which all demonstrate a reduction in affinity when compared to our lead compound ( ) trans PAT. The ( ) trans p F and Cl isomers retain an affinity that is within 4fold of the parent compound. However at the present t ime, it seems that none of the ( ) trans p substitutedPATs was able to improve on the affinity of the parent compound. In summary, it seems that moieties added to the para position of the pendant phenyl ring increase the affinity for the (+) isomer, whil e reducing the affinity for the( ).The result is that stereoselectivity between the ( ) and (+) isomers that has been so well documented, is no longer observed with these para substituted derivatives. The diverging effects on affinity observed for the ( ) and (+) stereoisomerssuggests the enantiomers have unique environments within the orthosteric binding site for the H1 receptor. The p Br PATs follow this trend however the ir larger bromine atom prevents either enantiomer from effectively binding to the H1 receptor N substituted PAT Derivatives To probe the area surrounding D3.32, the residue providing the critical ionic interaction for binding to all aminergic receptors, (+) and ( ) trans N,Ndiethyl PAT were
82 synthesized. These ligands provide important information about the region surrounding residue D.3.32, the most important residue of the H1 receptor binding pocket. The addition of lengthening aliphatic chains around the lone nitrogen of PAT was predicted to have little effect on the strength of the ionic interaction with D3.32A. However, the larger side chains may provide significant steric hindrance and prevent the protonated amine from coming into contact with the desired aspartic acid residue. In the case of ( ) trans N,Ndiethyl PAT this proved not to be the case. It was determined that this compound bound quite similarly to ( ) trans PAT, showing only a 4 fold reduction in affinity with a Ki of 8.59 0.53 nM. Contrary to this, (+) trans N,Ndiethyl PAT show ed a pronounced reduction in affinity with a Ki of 193.76 3.3 nM. T abulated affinity values for all N alkyl PAT derivatives are located in table 24 The structures for the compounds discussed in this section are found in figure 28. These results reitera te how crucial stereochemistry is when binding to the H1 receptor. In order to produce such a drastic decrease in affinity, there must be a significant difference in the binding mode of (+) 2S,4R and ( ) 2S,4R diethyl PATs. Since all aminergic ligands inter act with D3.32, this amino acid can be thought of as an anchor point for ligand binding. Once this interaction has formed, the ligand can form the other Van der Waals, hydrogen bonding, and/or aromatic interactions that allow it bind to the receptor. Altho ugh D3.32 can be considered an anchor point, it by no means restricts the regions of the receptor with which a ligand interacts. This point was eloquently demonstrated in the 2AR using a modeling technique called LITICON, as is discussed in detail in chapter 2. Briefly, it was shown that partial, inverse, and full agonists at the 2AR all stabilized unique receptor conformations with unique residue
83 contacts, despite maintaining the same anchor point at D3.32 (Bhattacharya et al., 2008) It was proposed that after t he initial ionic bond is formed, the ligand and receptor adopt the most energetically favorable conformation, leading to unique conformations for each ligandreceptor complex. Unique molecular determinants and by extension receptor conformations, were shown for the partial agonists dopamine, salbutamol, and catechol the inverse agonist ICI 118551, and the full endogenous agonist norepinephrine (Bhattacharya et al., 2008) By applying a similar logic to the histamine H1 receptor it is possible t o explain the binding differences between (+) and ( )N,N trans diethyl PAT. The most crucial factor in the loss of affinity for the (+) isomer is certainly steric hindrance, as the only difference between the enantiomers is stereochemistry It is possible that the two different binding modes of (+) and ( ) PAT force the protonated amine group into diverging conformations. For example, the aromatic interactions that aid in the binding of the ( ) trans N,Ndiethyl PAT enantiomer may allow this compound to int eract with D3.32 in a favorable forked interaction demonstrated in figure 29 This conformation would allow the protonated amine to interact strongly with D3.32, while minimizing the negative steric effects of the longer diethyl side chains. The inversi on of the stereocenters in the (+) isomer causes a change in the aromatic amino acids that are req uired for binding. This change would alter the most stable conformation of the receptor and the angle with which the amine moiety approaches the crucial D3.32 residue. Even a slight steric clash that weakened the ionic interaction would result in a significant drop in affinity, as the ionic interaction is the strongest factor in determining the affinity of any aminergic ligand.
84 This point is clearly illustrat ed by examining the affinity of older PAT analogs. Our lab has previously synthesized PATs with a litany of N substituted alkyl derivatives ranging from methyl to allyl. The racemic mixtures of cis and trans trimethyl PAT demonstrate that increasing the num ber of N alkyl substituents does not increase H1 receptor affinity. The Ki values for these compounds are 107 4.1 and 120 15 nM, respectively ; w hereas the affinity for () cis PAT is 13.6 nM and () trans PAT is 4.26 nM at the HH1R (Bucholtz et al., 1999) The only structural difference between these ligands and the current generation is the addition of a third methyl group, yet the high affinity and cis/trans stereoselectivity that is so evident with the dimethyl PATs is lost. This result has been pr eviously explained by the loss of a proton at the amine, creating a permanent positive charge and preventing proper hydrogen bonding with D3.32. It seems that a permanent positive charge on the amine is not conducive to binding to the H1 receptor with high affinity (Bucholtz et al., 1999) Removing alkyl groups from the amine moiety also causes pronounced drops in H1 affinity. () trans NH2PAT, a synthetic intermediate in the production of (+) and ( ) trans PAT lacking two N me thyl groups, has little affinity for the H1 receptor with a Ki = 1270 nM. A slight structural analog, () trans NH(CH3) PAT increased H1 receptor affinity approximately twenty fold to 64 nM. The addition of a N allyl group to the free amine created () tra ns NH(C3H5) PAT and again yielded a significant increase in receptor affinity from the free amine (Ki = 45 nM). The addition of a methyl group to this compound created () trans N CH3( C3H5) PAT and further increased the affinity of the ligand to a Ki of 3.4 nM. Substituting a second allyl group for the N methyl yielded () trans N(C3H5)2PAT. This compound was not as potent as its predecessor was (Ki =
85 10.2 nM) and suggested that there were steric limitations for alkyl substitutions on the amine (Buch oltz et al., 1999) As these results clearly indicate, the most potent PAT analogs all contain a N,N di alkyl substitutions. The rank order of H1 affinities for N,N dialkylatedPATs from high to low affinity is dimethyl > diethyl >diallyl. It is clear from this trend and the affinity values shown above that the dialkyl substituents have a profound effect on H1 affinity and that the shorter the alkyl side chain the greater the H1 receptor affinity. The addition or subtraction of any alkyl moiety from the N,N dialkylated scaffold negatively affects the ability of the protonated amine moiety to form its ionic bond with D3.32, and consequently the affinity of the ligand in question for the H1 receptor. Based upon this data, it was concluded that N,N dialkylPATs have the strongest interactions with the receptor and that N,N dimethyl was the most effective alkyl substituent. Chloro and H ydroxyl Tetrahydronaphthalene S ubstituted PATs The synthesis of trans andcis ()6 hydroxy 7 chloroPAT and trans and cis ()6,7 di hydroxy PAT was first reported in 1993, before the H1 receptor was even recognized as their most potent target (Booth et al., 1993) The structures of these compounds are found in figure 210 with representative curves in figure 211 Interesti ngly, the chloro/hydroxyl substituted PATs retained high affinity and the ability to stimulate TH in rodent forebrain while the di hydroxy compounds did not (Bucholtz et al., 1999) It was originally hypothesized that the di hydroxy substituents would mim ic 2 agonists to produce a more efficacious ligand. In hindsight and with the knowledge that differing GPCRs evolved to discriminate between neurotransmitters, this result makes sense. A receptor that had evolved to bind hi stamine specifically would have scant affinity for a catechol ring that is possessed by
86 many catecholamine neurotransmitters. If it did the histaminespecific receptor would likely have high affinity for epinephrine, norepinephrine, and dopamine as these structures contain protonated amines with a catechol ring. As briefly mentioned earlier, the chlorohydroxy derivatives proved to be more potent than the di OH PATs with a rank order of () trans 6 Cl7 OH > () ci s6 Cl 7 OH > () cis di OH > () trans di OH PAT. The affinities listed in table 2 5 were performed using the cloned HH1R and vary slightly from the original publications that used guinea pig brain homogenates (Booth et al., 1993; Bucholtz et al., 1999) The rank order for the ligands remained the same, with the slight exception that () cis and () trans 6 Cl7 OH PAT had very similar affinities in guinea pig brain tissue. Despite their high affinities, the Cl OH PATs have proved difficult to separate into their respective enantiomers. The Cl a nd OH substituents interact very strongly with the polar polysaccharide column used in chiral separation and cannot be resolved in this manner. PAT L ike Compounds with Ring Perturbations A large portion of clinically relevant histamine antagonists, such as triprolidine, hydroxyzine and acrivastine, possess a 1,1diaryl 3 aminopropa(e)ne moiety, found in figure 21 2 (Bucholtz et al., 1999) PATs structure closely resembles this established pharmacophore for the H1 receptor as demonstrated by the overlay in the same figure Two additional successful compounds, chlorphenamine and mepyramine that make use of this pharmacophoreare also depicted. Having already ex plored the majority of this pharmacaphores chemical space, it is necessary to discuss permutations to its tricyclic ring structure in order to complete a thorough SAR analysis. To date several ligands that are PAT like, but possess modifications to their tricyclic ring systems have bee n synthesized. These ligands are cis and trans (+ and ) PAB, (+ and ) trans N,N-
87 dimethylamino4 cyclohexyl 1,2,3,4 tetrahydro2 naphthalenamine, or C yclohexyl A mino T etralin (CAT), (+ and ) trans N,Ndimethylamino4 c yclooctyl 1,2,3,4 tetrahydro2 naphthalenamine or C yclo O ctyl A mino T etralin (COAT), and (+ and ) trans m phenyl PAT. Of these compounds, only the PABs have been previously reported, making the majority of these results novel and providing new SARs to exami ne (Bucholtz et al., 1999; Moniri et al, 2004) The structures of these compounds can be found in figure 21 3 with their KI values tabulated in table 26 As they have been previously reported, the PAB family provides an ideal starting point for the discussion of the tricyclic structure activity relationships. PAB is slightly different from our lead compound ( ) trans PAT, the only difference being an extra methylene group in the fused ring system. This trivial modification, which creates a fused cycl oheptyl ring as opposed to a fused cyclohexyl, causes profound changes in the ability of these ligands to bind to, and activate the H1 receptor. The binding order of the PATs, ( ) trans > ( ) cis> (+) trans > (+)cis, is now completely inaccurate. In fact, the affinity of () trans PAB was so low (> 10 M) that it was never resolved into its enantiomers. () cis PAB did show some promise and its enantiomeric pair was resolved. ( ) cis PAB proved to be the more active stereoisomer with a respectable Ki of 57.5 9.25 nM. (+) cis PAB demonstr ated a Ki of 300 40 nM. Despite its modest affinity, ( ) cis PAB proved to be a n interesting compound, as it was the first G partial agonist within the PAT family. This result is quite remarkable as it demonstrates just how fickle GPCRs can be when selecting an agonist. The four PAT stereoisomers, which differ only in stereochemistry and lack the additional methylene moiety, show no activation of G. In fact, ( ) trans PAT is an inverse agonist at this same pathway. B y
88 introducing a cycloheptyl ring, the conformation of the H1 receptor stabilized by ( ) cis PAB is now unique enough to cause Gto activate instead of dissociate as it would with the inverse agonist ( ) trans PAT. This example illustrates how critical it is to examine each stereoisomer individually when investigating GPCRs and lends credence to this SAR analysis. The pendant phenyl is the next area of the PAT scaffold to consider. A variety of ligands were synthesized to probe the region of the receptor surrounding the pendant phenyl ring. These ligands include m phenyl PAT, CAT, and COAT. The m phenyl PATs investigate the possibility of further aromatic interactions in the region surrounding the pendant phenyl ring. COAT and CAT examine how critical the aromatic portion of PAT is for effective H1 binding. The trans m p henyl PAT s have been synthesized, resolved, and investigated at the WT H1 receptor. A solid Ki was obtained only for ( ) trans m phenyl PAT, which displaced 3H mepyramine with a value of 213.0 20.4 nM. (+) trans m phenyl PAT was unable to displace mepyramine effectively even at concentrations > 10 M. Both compounds are significantly less potent than their parent PATs, but the drastic difference between (+) and ( ) m phenyl PAT does provide a few clues about the environment surrounding the pendant phenyl region. Despite its lowered affinity, the ( ) isomer is able to tolerate the additional phenyl group, confirming that the ( ) pendant phenyl region binds within or close to a cluster of aromatic residues. Based upon previous saturation binding studies it seems that the aromatic residues in TMD 6 are the most likely interaction site. It has been previously reported that the F6.52A mutation abolishes the binding of ( ) trans PAT (Bakker et al., 2004)
89 Similar unpublished saturation experiments revealed the surrounding aromatic residues W6.48,Y6.51, and F6.55 are involved in the binding of 3H trans PAT These residues could stabilize the m phenyl group sufficiently to allow binding of the ( ) isomer to take place. The (+) isomer clearly lacks either the steric space in the region around the pendant phenyl ring or the proper amino acid residues to create a strong enough interaction with the ligand. Either way these results clear ly suggest a differing chemical environment in the regions surrounding the (+) and ( ) enantiomers. It is interesting to note that the p Br compound showed an opposite trend, where the (+) isomer could better tolerate the bulky bromine atom. This can be ex plained by differing amino acid sequences surrounding the pendant phenyl groups of the (+) and ( ) isomers. The (+) isomers demonstrate an increase in affinity for parapendant phenyl substituents like F and Cl, suggesting the presence of polar residues in that region. Aromatic substituents are better tolerated by the ( ) enantiomers that are postulated to interact with the aforementioned aromatic residues in TMD 6. The same ( ) enantiomers show either no change in affinity or a decrease in affinity with nonaromatic pendant phenyl substituents suggesting a lack of suitable polar amino acids in the region surrounding the pendant phenyl moiety of ( ) trans PAT Regardless of the substituent that is present it seems that the bulkier the group that stems from t he pendant phenyl ring the lower the H1 affinity becomes. Future H1 ligand design should focus on pendant phenyl substituents that are no larger than a chlorine atom, as both the p Br and m phenyl substituents drastically reduce binding affinity regardless of stereochemistry. The next two ligands CAT and COAT are radically different from their parent ligands. The parent compound ( ) trans PAT has a pendant phenyl ring at the 4-
90 position of its tetrahydronaphthyl ring. This phenyl has been previously establi shed as a critical region of the PAT scaffold, as its removal decreased the affinity of ( ) trans PAT by 100fold (Bucholtz et al., 1999) These studies confirmed the importance of the aromatic ring, but due to a lack of suitable compounds, no further SAR results were obtained. CAT and COAT replace the pendant phenyl ring with nonaromatic cycloalkyl rings of six and eight carbons respectively, and are perfectly suited to fill this empty niche in the SAR of the PAT family. Both the ( ) and the (+) trans is omers of CAT and COAT have been synthesized, and their binding results are discussed here. The introduction of the more flexible cycloalkyl substituents in place of the pendant phenyl ring explores new regions of the chemical space surrounding the binding pocket for the (+) and ( ) Pats Both the ( ) and (+) CAT isomers demonstrated high binding affinity at the H1 receptor. ( ) trans CAT has an affinity of 1.58 0.5 nM, which is strikingly similar to ( ) trans PAT. (+) CAT demonstrated a nearly identical a ffinity for the H1 receptor of 1.7 0.2 nM These results suggest that the cyclohexyl ring is as effective as the phenyl in producing a high affinity H1 ligand. This result was extremely surprising given the established interactions that ( ) trans PAT has with the aromatic residues in TMD 6. Based on the high affinity, it seems that the additional flexibility of the cyclohexyl ring and the sum of thecyclohexyl Van der Waals forces is able to create an interaction with the receptor that is as strong as the a The potent binding of the (+) isomer is not quite as surprising, given that the binding pocket around the pendant phenyl ring was able to tolerate larger halogen substitutions more effectively
91 than the ( ) isomer. This tolerance suggested a larger and more flexible binding pocket for the pendant phenyl position of the (+) isomer As observed with the p Br compounds, even the most accommodating binding pocket does have its limits. The COAT isome rs appear to push these limits too far, given that both isomers demonstrate a pronounced reduction in affinity when compared to either the PATs or CATs. ( ) trans COAT, which was the only isomer that produced a valid Ki, still possessed a meek affinity of 432 60.1 nM. The enantiomer (+) trans COAT was unable to displace 3H mepyramine from the receptor even at a 10M concentration. As the affinity of both (+ and ) COAT was reduced by at least 100fold from PAT and CAT, this particular analog was not potent enough for further derivatization. PAT Binding Summary Recent advances in GPCR theory have demonstrated that each ligand stabilizes a unique receptor conformation. This result validates and helps explain previous observations of stereoselective binding to the H1 receptor (Bucholtz et al., 1999; Gillard et al., 2002; Booth et al., 2002) Enantiomers are defined as two identical compounds, meaning their physical properties and structures are the same, but t hey are nonsuperimposable mirror images of each o ther. Yet somehow, these receptors are able to discriminate between enantiomers when very sensitive chemical techniques such as I nfrared spectrometry (IR) U ltraviolet V isible spectrometry (UV VIS) Gas chromatography coupled with mass spectrometry( GC/MS ) and most high performance liquid chromatography( HPLC) are unable to do Traditional GPCR theory involved active, inactive, and constitutively active states that interacted with a ligand and a G protein to form the ternary complex model. Unfortunately, these models assumed there
92 would be only one active or inactive conformation of a receptor and had difficulty explaining concepts like allosterism and partial agonism Current GPCR theory assumes that every ligand will produce a different conformation of the receptor and is flexible enough to accommodate these observations. The series of ligands produced by our labs provides an excellent example of modern GPCR theory at work. This chapter details a series of stereospecific ligands that demonstrate functional selectivity and can even exert opposing effects on a signaling pathway with the addition of a simple methylene group. Invoking Accams razor, the simplest explanation for these diverse observations is if each ligand is indeed creating a unique GPCR conformation. This explanation is supported by the stereospecific binding demonstrated by the PAT analogues in the preceding chapter.
93 Figure 2 1 PAT family of stereoisomers demonstrating the stereochemical relationships between each compound. Table 2 1 Affinities of the PAT stereoisomers at the WTHH1R Compound Name Histamine H 1 Affinity in nM (K i S.E.M.) P value compared to ( ) trans PAT ( ) trans PAT 1.95 0.51 N/A ( ) cis PAT 13.7 2.0 0.0307 (+) trans PAT 29.79 3.5 0.0157 (+) cis PAT 177.166 9.4 0.0083 **
94 -11 -10 -9 -8 -7 -6 -5 0 10 20 30 40 50 60 70 80 90 100 110[Ligand][3H]-Mepyramine Binding (% Specific Bound SEM) Figure 2 2 Representative competitive binding curves for the PAT family of stereoisomers at the human WT H1 receptor. The ligands represented are as follows: ( ) trans PAT ( ),() cis PAT ( ,(+) trans PAT ( ) and (+) cis PAT ( ). Figure 2 3 Enantiomers of ( ) cetirizine that demonstrate high stereoselectiv ityat the HH1R
95 Figure 2 4 Structures of ( ) trans o rtho C l PAT (left) and ( ) trans orthoCH3PAT (right) Figure 2 5 Structures of ( ) trans meta halogenatedPAT derivatives where X is F, Cl, or Br Table 2 2 HH1Rb inding affinities of (+) and ( ) trans para substituted PATs Compound Name H H 1 Affinity in nM (K i S.E.M.) and P values for enantiomeric comparison P value compared to parent enantiomer (e.g. ( ) or (+) trans PAT ) ( ) trans m F PAT 3.03 0.26 0.29 (+ ) trans m F PAT 93.56 15.2 (P = 0.014) 0.026 ( ) trans m Cl PAT 10.19 2.4 0.00 60** (+ ) trans m Cl PAT 79.48 23 (P = 0.021) 0.03 9* ( ) trans m Br PAT 27.94 1.5 0.0167 (+ ) trans m Br PAT 62.52 10.1 (P = 0.020 ) 0.0022 **
96 Figure 2 6 Structures of the ( ) trans parasubstituted PAT s where X = F ,Cl,CH3, or Br Table 2 3 WT HH1R b inding affinities of (+) and ( ) trans parasubstituted PATs Compound Name Histamine H 1 Affinity in nM (Ki S.E.M.) and P values for enantiomeric comparison P value compared to parent enantiomer ( e.g. ( ) or (+) trans PAT ) ( ) trans p CH 3 PAT 17.33 2.7 0.0008 ** (+ ) trans p CH 3 PAT 11.30 1.03 (P = 0.0018) 0.0068 ** ( ) trans p F PAT 4.0 7 0.5 0.054 (+ ) trans p F PAT 8.268 0.6 (P = 0.0098) 0.026 ( ) trans p Cl PAT 8.38 1.10 0.036 (+ ) trans p Cl PAT 6.08 0.75 (P =0.2378) 0.022 ( ) trans p Br PAT 1058.2 53.0 0.0025 ** (+ ) trans p Br PAT 531.8 62.1 (P = 0.0232) 0.015
97 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 0 10 20 30 40 50 60 70 80 90 100 110Log [ p -CH3-PAT] (M)[3H]-Mepyramine Binding(% Specific Bound SEM) -10 -9 -8 -7 -6 -5 -4 0 10 20 30 40 50 60 70 80 90 100 110Log [Trans-p-X-PAT] (M)[3H]-Mepyramine Binding (% Specific Bound SEM) Figure 2 7 Representative HH1R competitive binding curves for (+) and ( ) trans parasubstituted PATs A ) ( s ( ) trans p CH3 ; ( represents (+) trans p CH3. B) ( ) trans ( ) & (+)p ara F PAT ( ),() trans ( ) & (+)trans paraClPAT ( ), and ( ) trans ( ) & (+)trans paraBr PAT ( )
98 Figure 2 8 S tructures of N alkylatedPAT derivatives. The names of the compounds are as follows; top row: trans N,Ndiethyl PAT (left), trans trimethyl PAT (center), cis trimethyl PAT (right), center row: trans N methyl N allyl PAT (left), trans N,NdiallylPAT (center), trans N allyl PAT (right), bottom row: NH2PAT(left), and N methyl PAT (right).
99 Table 2 4 Histamine H1 binding affinities of N alkyl substi tuted PAT analogues Compound Name Histamine H 1 Affinity in nM (K i S.E.M.) ( ) trans N,N diethyl PAT 8.589 0.53 (+ ) trans N,N diethyl PAT 193.761 3.3 ( ) trans N + (CH 3 ) 3 PAT 120 15 ( ) ci s N + (CH 3 ) 3 PAT 107 4.1 () trans NCH 3 (C 3 H 5 ) 3.4 0.3 () trans N(C 3 H 5 ) 2 10.2 1.7 () trans NH(C 3 H 5 ) 45 11 () trans NH 2 PAT 1270 92 () trans NH(CH 3 ) PAT 112 31 Figure 2 9 Proposed interaction of aspartic acid D3.32 with ( ) (left) and ( + ) (right) trans N,Ndiethyl PAT that may account for higher affinity of the ( ) enantiomer between the binding of (+) and () trans N,Ndiethyl PAT. Figure 2 10. Structures of racemic di OH and Cl OH PAT ligands
100 Table 2 5 Affinities of di OH and Cl OH PAT analogues at the WT HH1R. Compound Name Histamine H 1 Affinity in nM (K i S.E.M.) ( ) trans di OH PAT ~ 20 ( ) cis di OH PAT 24.5 1.04 ( ) trans 7 OH 8 Cl PAT 2.424 0.44 ( ) cis 7 OH 8 Cl PAT ~ 150 -10 -9 -8 -7 -6 -5 0 10 20 30 40 50 60 70 80 90 100 110[Ligand][3H]-Mepyramine Binding (% Specific Bound SEM) Figure 2 1 1 Representative competitive binding curves for di OH and ClOH PAT at the WT HH1R The ligands represented here are: cis ()ClOH PAT ( and trans ()Cl OH ( cis ()di OH PAT ( and trans ()di OH PAT (
101 Figure 2 1 2 Proposed H1pharmacophore and the ligands that are derived from it Left to right (top row) : the 1,1 diaryl 3 aminopropane moiety ( ) trans PAT structure superimposed on the 1,1diaryl 3 aminopropane moiety and acrivastine. Left to right (bottom row): hydroxyzine, triprolidine, and chlorphenamine.
102 Figure 2 1 3 Structures for the ring substituted PAT derivatives : ( ) cis PAB (top left) ( ) trans PAB (top center) ( ) trans COAT (right), ( ) trans m phenyl PAT (bottom left) and ( ) trans CAT (bottom right)
103 Table 2 6 Affinities of PAT analog ues with modified tricyclic ring systems Compound Name Histamine H 1 Affinity in nM (K i S.E.M.) ( ) trans PAB > 10,000 ( ) cis PAB 57.5 9.3 (+) cis PAB 300 40 ( ) trans CAT 1.58 0.5 (+) trans CAT 1.7 0.2 ( ) trans COAT 432 60 (+) trans COAT > 10,000 ( ) trans m phenyl PAT 213.0 20.4 (+) trans m phenyl PAT > 10,000
104 CHAPTER 3 INVOLVEMENT OF AMINO ACID RESIDUE Y5.48A IN THE DIMERIZATION/FUNCTIONAL STABILIZATION PROCESSES OF THE HUMAN HISTAMINE H1RECEPTOR Literature Review of Ballesteros Position 5.48 Interestingly, there has been little speculation in the literature about the amino acid residues in Ballesteros position 5.48, when compared to its neighboring positions like 5.42, 5.43, 5.46, and 5.47. In 5HT2A receptors, which have about 70% TMD sequence structural homology to HH1Rs, this position (F5.48) is implicated in ligand binding. A mutation to a nonfunctional alanine residue resulted in a decrease in binding affinity efficacy, and potency of 5HT. F 5.48A also reduced t methyl 5 HT, N5 HT and DOI (Shapiro et al. 2000, Roth et al. 2002). Shapiros paper predicted that F5.48 played a role in stabilizing the aromatic residues around the binding pocket indirectly, by allowing neighboring amino acids to achieve their correct helical structure (4 amino acids per helix turn) it is not possible to have residues 5.46 and 5.48 both pointing into the binding pocket (Shapiro et al, 2000) Residue 5.46 is known to be a key postion when differentiating the pharmacology of the rat from the human 5HT2 A receptor as well as in the binding of endogenous ligands at the H1 and 2AR (Smit et al., 1999; Carmine et al., 2004; Bhattachar ya et al., 2008; Fang, Travers and Booth., 2009 prepublication) P olar residues inserted into this position in the rat 5HT 2Areceptor, (S (human endogenous residue), T, and N) shift the pharmacological profile towards the human receptor (Kao et al. 1992, Johnson et al. 1994). Since ligand binding and the pharmacological profile are altered by mutations to this position, it follows that this particluar residue is expected to face into the binding
105 pocket of the receptor. And by extension from the Shapiro 2000 paper, that position 5.48 does not. A second possibility is that postion 5.48 is not directly involved in ligand binding, but is critical in stabilizing the GPCR in the proper conformation to bind its endogenous ligand. This result would manifest itself as a reduction in ligand affinity that could easily be misinterpreted as a direct interaction with the ligand in question. In 2adrenergic receptor, position 5.48 is again occupied by a phenylalanine. In this case, the residue is not predicted to be within the binding pocket of the receptor. Even with the novel modeling technique LIT iCON that allows rotation of the TMD heli cies based upon the most favorable interaction with the ligand, this position remains exterior to the binding pocket (Bhattacharya et al., 2008). These results mesh closely with the HH1R, constructed through Rho homology modeling in our lab, and a second m 2AR crystal structure Both models demonstrate position 5.48as an interface between the critical TMDs 5 and 6, and away from the binding pocket and. To lend further creedence to this hypothesis a series of prot eomics studies were compiled from GPCR sequences by Weinsteins group out of Cornell. They used several different bioinformatics methods to compile the most likely interfaces for the formation of dimers and/or oligomers (Filizola and Weinstein, 2005). By s tudying the evolutionary relationship between the sequence of different GPCRs and eliminating the conserved residues known to be directly involved in ligand binding, the group was able to focus their search to residues that have the potential to be involv ed in dimer formation. Their predictions pointed to residues located in transmembranes 46. Specifically for aminergic receptors, which inclu de the HH1R, the most likely
106 dimerization residues as predicted by lipid exposure, fall withing transmembrane 56 ( Filizola and Weinstein, 2005). When all of the data from these studies was compliled, a histogram of the most frequently selected residues was assembled. The most frequently selected postion turned out to be 5.48. At the time of the publication, there was no known data about the involvement of this particular residue in the dimerization/oligomerization of rhodopsin like GPCRs (Filizola and Weinstein, 2005). In fact, there is very little mutagenesis data that implicates specific amino acids in the dimerizat ion/oligomerization process of class A GPCRs The reported residues as of 2005 are I1.54 and V4.47 in chemokine receptors, C4.58 in the Dopamine 2 (D2) adrenergic receptor (Filizola and Weinstein, 2005). The predictions made i n this paper lend further creedence to our investigation of the Y5.48A mutation to the HH1R, and help confirm this a s a novel discovery. Materials and Methods The techniques and chemicals described in chapter two were used in this chapter as well Any additional techniques that were employed are described below. Measurement of PLC activity/[3H] inositol phosphate formation The methods for measurement of [3H] IP was as described previously (Booth et al., 2002). HEK293 cells were transiently transfected with wild type or mutant receptors in 100mm plates by following manufacturers instructions with 12 g of plasmid DNA and 5 0 l of Lipofectamine 2000 (Invitrogen). Cells were aliquotedi nto wells of 12 or 24well plates and labeled with 1 Ci/ml myo[2 -3H (N)] Inositol in inositol free DMEM medium (Chemicon D 101) with supplements of 5% dialyzed fetal bovine serum (Invitrogen Gibco 26400) and 1% nonessential amino acids overnight (approximately 1224 hrs) Then cells were treated with drugs for 1 hrs with s upplement of 35
107 mMLiCland 10 M of pargyline. Accumulated inositol phosphates were released from cells by 50 mM formic acid incubated for 1 hr at room temperature. [3H] Inositol phosphates were separated by AG1X8 columns (formate form) and eluted by 0.8 M ammonium formate into scintillation vials for counting. Data was analyzed using the nonlinear regression algorithm in Prism 5 .0 2 (GraphPad Software Inc., San Diego, CA) and expressed as mean percentage of basal. Due to variations in constitutive activity of the mutants calculations were based on corresponding mutant basal receptor activity and presented as percentage of basal (%basal), not wild type basal receptor activity Thus allowing comparisons to be made between the point mutations and the WT receptor. Accumulation of [3H] Inositol phosphates by phosphoinositide hydrolysis were also measured by scintillation proximity assay (SPA) (Brandish et al., 2003; Bourdon et al., 2006). Transfected cells were labeled with 10uCi/ml of myo[2 -3H(N)] Inositol in inositol free DMEM medium (Chemicon D 101) with supplements of 5% dialyzed fetal bovine serum (Invitrogen Gibco 26400036), 2 mM glutamine and nonessential amino acids overnight. Then [3H] Inositol phosphates were detected 0.2mg/well of yttrium silicate RNA binding resin (Amersham Biosciences RPNQ0013, Piscataway, NJ). Radioactivity was quantified with reading at MicroBetaTriLux Data was analyzed using the nonlinear regression algorithm in Prism 5 .0 2 (GraphPad Software Inc., San Dieg o, CA) and expressed as mean percentage of basal, EC50 and EMAX.
108 BRET Analysis BRETconstructsweremadebysubcloningY5.48A(obtainedfromBooth)intoH1eYFP/pcDEF3andH1Rluc/pcDEF3 (van Rijn et al., 2006) .Sequenceofallconstructswasverified. HEK 293 cellsweretransientlytransfectedinwhite96wellplateswith50ngGPCR Rlucand01000ng GPCR eYFPDNA/106cellsusingPEI(1:3).TotalDNAwaskeptconstantat2050ng/106cellsbysupple mentingwithempty plasmid. Two daysaftertransfection,theculturemediumwasreplacedwith50uLPBSandeYFPfluoresc encewasmeasuredinaVictor3platereader(ex@485nm;em@535nm). Next,50uLPBSsuppl ementedwith10 M coelenterazineHwasadded,andbioluminescence(em@460nm)and a BRET signal (em @535nm)wasrecordedafter5min ute incubation.CorrectedBRETratiois(BRET/Rluc)x(BRET/Rluc)0.Wherex was cotransfectionofRlucandeYFP,and0transfectionofonlyRluc.YFP/Rluccorrectedis the YFP /RlucratiominusthelowestYFP/Rluc(left shiftedcurve). BRETMaxandBRET50valuesarefittedbyonesitebinding(hyperbola)usingGraphPadPrism4 This technique and its data analysis were performed by Dr. Henry Vischer,a collaborator of ours from Vrijeuniversity in Amsterdam. Experimental Results and Discussion Saturation Binding Assay R esults The following results are discussed based on the finding that that 3H ( ) trans PAT binds preferentially to H1 dimers, while the radioligand 3H mepyramine to binds to H1 momomers and dimers ( Booth et al., 2004) Investigation into the Y5.48A point
109 mutations was begun by quantifying the transiently transfected receptor level with the standard H1 radioligand 3H m epyramine. All data is reported as the mean the standard error of the mean (S.E.M), unless otherwise noted. Results produced a BMAX of 3.085 0.65 pMol /m g protein and a KD value of 1.790 0.18 nM for the Y5.48A receptor When contrasted with the H1 wild type (WT) receptor using a t test, which has a BMAX of 2.9 81 0.5 nM and a KD of 1.024 0.22 nM, the resulting pvalues were P = 0.89 and P = 0.40 respectively. Neither t he difference in BMAX values nor the KD values of 3H mepyramine at the WT and Y5.48A point mutated receptor breaks the bottom significance level of P< 0.05. Taken together, these results indicate that there is no difference in the receptors ability to bind mepyramine effectively, or in its ability to express at the cell surface. Actual saturation binding curves are shown below in f igure 3 1 and BmaxandKDvalues for WT and Y5.48A are tabulated in table 3 1 These results were not in themselves surprising, given that this particular point mutation is oriented away from the ligand binding pocket. However, when our radioligand3H ( ) trans PAT was tested with identical membrane preparations, in the same exact plate as the successful 3H mepyramine saturation assay, a complete loss of ligand binding was observed. This was initially explained away as a failed experiment, or a batch of protein that had denatured. A fter several repetitions with protein concentrations ranging from 0.010.1 mg/mL of transiently transfected protein, the result was confirmed 3H ( ) trans PAT was unable to bind to the Y5.48A point mutated HH1R at concentrations up to 1 0nM. Representative saturation isotherms for ( ) trans PAT at various protein concentrations are also located in figure 32. For a point of comparison,3H ( ) trans PAT is able to bind to the WT H1 receptor with high affinity. It
110 possesses a BMAX of 0.345 0.06 pMol /mg prote in and a KD value of 1.34 0.34 nM in transiently trans fected HEK 293 cells. Data is displayed below in f igure 3 2 The WT BMAX values differ slightly from those published in Baaker et al. 2004, and from Booth et al. 2001. Although, it should be noted that the HH1R was expressed in a different cell line in each of those papers. They expressed the HH1R in COS 7, an African green monkey kidney cell, and in Chinese Hamster Ovary (CHO) cells respectively. In these experiments, the HH1R is expressed in Human Embryonic Kidney cells (HEK 293). The use of various cell lines over the course of our investigation has told us that receptor expression levels can vary significantly between cell lines. KD values remain constant despite the receptor level, as they are dependent upon the sequence of the receptor, which does not vary with the cell line. This makes BMAX comparisons rather difficult, especially with transient transfections, where expression levels can vary based upon: the health of the cell, the passage number, how effectively the cDNA is taken up, and the amount of cDNA that is used for the expression (Dimond, 2007). Fortunately, it has been observed that 3H ( ) trans PAT consistently labels oneseventh of the entire HH1R population that is identified by 3H mepyramine. This has remained true across CHO and COS 7 cells and has served as an internal control to confirm proper 3H ( ) trans PAT binding (Booth et al. 2001. ; Baaker et al. 2004). Based upon the above results, this ratio is confir med in HEK 293 cells, and lends credence to the WT binding results for both 3H Mepyramine and ( ) trans PAT. The Y5.48A saturation results are more difficult to interpret. The previous molecular determinants obtained through mutagenesis studies indicate very similar
111 binding pockets for ( ) trans PAT and mepyramine. In fact, Y5.48A is the first mutation examined by our lab to produce differing effects for the two ligands. Every other mutation examined produced either no effect on ligand binding, or caused th e loss of both ( ) trans PAT and mepyramine binding. The positions that resulted in a complete loss of radioligand binding were D3.32A, Y3.33A, W4.56A, F5.47A, W6.48A, and F6.52A. These interactions have been previously confirmed for 3H mepyramine by Wiela nd et al. in 1999. Based upon these mutagenesis results, it was inferred that ( ) trans PAT and mepyramine are interacting with a similar set of residues. They engage the critical D3.32A residue in TMD 3, possess an interaction with W4.56A, and engage the cluster of aromatic amino acids in TMDs 5 and 6. The loss of ( ) trans PAT binding at Y5.48A caused us to reevaluate that position. As detailed earlier, it was prev iously established that ( ) trans PAT is a domainswapped dimer preferring ligand that binds a subset of the overall H1 receptor population. Therefore, it was postulated that Y5.48A somehow disrupts the formation of these dimers causing the observed loss of 3H ( ) trans PAT binding. Competitive Binding Experiments In order to follow up on these binding results properly it was necessary to further probe the binding of a variety of H1 ligands to this point mutated receptor. Instead of continuing experiments involving high levels of expensive radioligands competitive binding assays using 3H Mepyramine, whose binding affinity is unaffected by this mutation, were performed. The logical starting compounds were ( ) trans PAT and the endogenous agonist histamine. A diverse series of HH1R ligands were included in these experiments including: the inverse agonist mepyramine, the endogenous agonist histamine, an d an enantiomer and a dias tereomer of ( ) trans PAT The PAT
1 12 enantiomer and diastereomer are (+) trans PAT a nd ( ) cis PAT, respectively. They represent the two most potent PAT stereoisomers, after the lead compound ( ) trans PAT. Structures for these compounds are found below in figure 3 3 The binding profile of histamine is altered when compared to the WT HH1R. At the WT histamine has a KI of about 3 M, while at Y5.48A that values is reduced significantly to 15.8 3 M (P = 0.0067**). This reduction in affinity cannot be due to a direct interaction with the ligand as previously detailed, a more lik ely explanation is that the native conformation of TMD 5 is altered by this mutation. The critical binding residues K5.39, T5.42, and N5.46 are all very close to this mutation, and any perturbation to the helical structure is likely to cause a reduction in affinity. Competitive binding of mepyramine was affected as well, but only slightly. The WT KIof 2nM, was reduced to 4.905 1.2 nM (P = 0.026*). As expected for a ligand with minimal interactions with TMD 5, there is only a slight reduction in the abilit y of the Y5.48A HH1R to bind mepyramine. As expected from the saturation binding results, there is a significant reduction in the ability of ( ) trans PAT to bind to the Y5.48A mutant. At the WT HH1R, ( ) trans PAT displaces 3H mepyamine quite effectively, evidenced by its KI value of 1.95 0.51 nM. When Y5.48A is considered, the KI value drops approximately 100fold to 148.4 5.4 nM, when a t test was performed to determine significance the resulting value was P = 0.0014**. Indicating that this reduction in affinity is quite significant when compared to the WT receptor. The next ligands investigated were the two stereoisomers of PAT, ( ) ci sPAT and (+) trans PAT. KI values at the WT rec eptor are 13.7 0.2 nM and 29.8 4.0 nM, respectively. Data from Y5. 48A, indicates that both stereoisomers drop in affinity
113 by about 10fold. ( ) cis PAT drops to 143.6 0.6 nM at Y5.48A, with a P value of 0.0003*** Affinity of (+) trans PAT falls to 245.5 19.2 nM, a t test revealed P = 0.021* E ach of the PAT isomers that were investigated showed a significant drop (P<0.05) in their ability to displace the radioligand. The curves displaying this data are demonstrated below in f igure 3 4 and data is tabulated in table 32 These results indicate that ( ) trans PAT possesses the largest drop in affinity for the Y5.48A point mutation. It is significant to note that if one considered only the P values for each of these ligands, it would erroneously be concluded that ( ) cis PAT is the ligand that is most affected by Y5.48A, since it has the lowest P value for the WT Y5.48A comparison. P values can be considered when evaluating the difference between two sample populations and are useful for determining how different two populati ons are, however they must be used carefully. They should not be used to make conclusions about anything beyond the two populations that are being directly compared. For example, it can be concluded that there are significant differences between the bindi ng of ( ) trans PAT at the WT and Y5.48 HH1R, and that the same is true for ( ) cis PAT. However, the statement that ( ) cis PAT is more strongly affected by Y5.48A than ( ) trans PAT would be incorrect ; despite the fact that ( ) cis PAT has a smaller pva lue (0.0003 vs. 0.0014). In fact, the affinity of ( ) trans PAT is reduced 100fold at Y5.48A, while ( ) cis PAT affinity was lowered by a factor of 10. Therefore, when examining the trends between different compounds at differing receptors, one should consider not only the significance value that the data generates, but also the overall magnitude of affinity gain or reduction as well.
114 The smaller pvalue generated by ( ) cis PAT can be explained by the very small standard deviation values for those experi ments. The t value for a given data set is inversely proportional to the square of the standard deviations ( SX 1 and SX 2 shown below). This gives standard deviation a very strong influence over the value generated by the t test. Since the standard deviation values generated by the ( ) cis PAT data are miniscule, there is less chance of the WT and Y5.48A data overlapping, and validating the null hypothesis. This results in an exaggerated significance value and explains the very low p values observed in these experiments. T test for significance Inositol Phosphate Production Mediated via G Interpreting these results with binding data alone proved quite difficult, so functional experiments were performed. The HH1R displays functional selectivity, like many other GPCRs, signaling through both G/IP3 and G/cAMP. In the periphery of the body the G/IP3 signaling cascade, which has made HH1R receptors synonymous with allergic responses, is the domi nant pathway ( Leurs et al., 1995; Moniri et al., 2004). Our investigation was begun at the G/IP3 pathway, using the endogenous agonist histamine. At the WT HH1R, the function of histamine through G has been well established. From my work at the WT receptor, it has been established that histamine has an E C50 value of 0.159 0.024 M with an EMAX of 401.1 81% basal. This value meshes closely with EC50 values found in the literature for the human H1R (Seifert et al., 2003; Xieet al., 2006). It should be mentioned that histamine is an unusual ligand, since
115 it has a KI value in the M range, yet is able to activate its receptor easily at concentrations 10x lower than the KI ( Xie et al., 2006) This discrepancy highlights an important point that binding affinity (KI) and functional activity (EC50 or IC50) do not always correlate and should be analyzed separately. When Y5.48A is examined in the IP3/PLC/ Gassay with histamine, the receptor is clearly able to function. The EC50 value is reduced at least 50fold to 8.028 0.66 M, result ing in a significant P value of 0.0003 ***. Interestingly, the EMax for the receptor, 340.9 53.5% basal, is unperturbed when compared to the WT (P = 0.54). To ensure that the difference observed between the WT and Y5.48A receptors was indeed significant, an ANOVA analysis was performed in GraphPad prism 5.0to compare the entire curves. Similar to the t test that was performed with the EC50and EM axvalues, the ANOVA analysis revealed a significant difference between the WT and Y5.48A functional curves and resul ted in a P value of 0.0042**.The observed reduction in histamine efficacy indicates that there is some perturbation to the receptor caused by the Y5.48A mutation. This perturbation is unlikely to be caused by a direct residueligand interaction, because hi stamine is a rather small ligand and this mutation does not face into the binding pocket of the receptor. It is more likely the mutation alters the conformation, structure, or the stability of the receptor, to cause the observed reduction in efficacy. It s hould be noted that Y5.48A is located very near a series of residues that are known to be critical for histamine binding.K5.39, T5.42, and N5.46 are all located within two helical turns of Y5.48A. If position 5.48 is involved in stabilizing the helix, its removal could destabilize the helical structure preventing dimerization from occurring and altering the binding of ligands that interact with TMDs 5 and 6.
116 The unchanged EMax value between the WT and Y5.48A HH1Rs suggests that the receptor is still fully functional, but its ability to recognize and efficiently respond to its endogenous agonist is diminished. Compiled EC50 and IC50 for these IP3 experiments can be seen for histamine and ( ) trans PAT in table 3 3 Representative plots are shown below in f igure 3 5 ( ) trans PAT, which is an inverse agonist at the WT HH1R, was the next ligand examined in the IP assay (Travers, Fang, and Booth., 2009 prepublication) Quantifying inverse agonism responses is a much more laborious process than evaluating agonist response. In order to measure a signal, a receptor must first demonstrate a basal activity that is significant enough to be measured and the ligand must be potent enough to bind and reduce the observed basal level of activity. For these experiments, sensitivity is crucial and any methodology that possesses a lower limit of detection (LOD) should be employed. For example, 3H IP /PLC activity can be quantified through anion exchange columns or by binding to Scintillation Proximity Ass ay (SPA) beads. It is possible to evaluate 412 times as many samples with the SPA methodology, but it has a much lower LOD than the columns. After many replicates and inconclusive experiments, IC50 values were obtained for mepyramine and ( ) trans PAT. At the WT receptor, mepyramine yielded an IC50 of 2.51 0.4 nM with an Imax of 61.4 1.3% basal. At Y5.48A, the IC50and Imax values were 3.35 0.3 nM and 63.49 3.2% of basal activity. When IC50 and Imax values were compared between the WT and Y5.48A receptors, neither the potency, nor the efficacy broke the significance limit of P<0.05 giving values of 0.22 and 0.65 respectively. A further analysis using ANOVA to compare the full
117 functional curves between WT and Y5.48A confirmed that there is no significa nt difference for mepyramine, with a P value of 0.63. These data confirm that mepyramine is still able to bind to and activate the Y5.48A mutated HH1R. For () trans PAT an IC50of 251.7 18.2 nM, along with an Imax of 63.8 2.2% basal was obtained at the WT receptor. Similar experiments at Y5.48A, yielded no change from basal levels of activity, regardless of the PAT concentration. ANOVA analysis was then used to confirm that this loss of inverse agonism is significantly different from the WT recept or. The p value resulting from the ANOVA test showed that there was a statistically significant loss of function across the entire curve between the WT and Y5.48A receptors, with a P value of 0.0012**.Suggesting that at Y5.48A ( ) trans PAT appears to los e its ability to function as an inverse agonist. When this finding is contrasted with the ability of histamine to fully activate the Y5.48A receptor and the ability of mepyramine to bind and function, the result is a ligand specific loss of function for () trans PAT at Y5.48A. Given the ligandspecific loss of binding and function of ( ) trans PAT, and the reduction in affinity of histamine at Y5.48A, it is clear that this point mutation has caused some significant structural alterations to the WT HH1R. Wh atever these modifications may be, they are able to generate a fully functioning receptor. As evidenced by histamines ability to produce a similar EMax value as the WT and the unaltered binding profile and functional profile of mepyramine. These facts shi ft the focus away from these ligands and toward the unique properties of ( ) trans PAT. As mentioned above, ( ) trans PAT is a ligand with unusual pharmacology. The functional selectivity at the two main HH1R functional pathways, inverse agonism at
118 Gand a gonism at G S, ha s yet to be duplicated by any other ligand in our lab. Previous studies have shown that ( ) trans PAT labels only oneseventh of the HH1R population that is expressed and that this ligand binds to domainswapped dimers ( Booth et al., 2002; Bakker et al., 2004). These facts coupled with the location of Y5.48A away from the intrahelical ligand binding pocket, and its hypothesized role in receptor dimerization suggest that Y5.48A is somehow preventing and/or hindering the HH1R from creating domain swapped dimers. The loss of these dimers prevents ( ) trans PAT from binding in a natural manner and as a result the subsequent function that ligand exerts upon the receptor is lost B RET S tudies C omparing d imerization of the WT and Y5.48A H1 Re ceptors In order to address the dimerization hypothesis, we contacted some of our colleagues in Amsterdam who are experts in FRET and BRET. Rob Leurs and Henry Vischer investigated dimerization at the WT HH1R using BRET, and compared the results to the poi nt mutated Y5.48A receptor. Before discussing the data, a brief overview of BRET should be detailed BRET is an acronym for B ioluminescenceR esonanceE nergy T ransfer and is a technique that is used to determine the proximity of two proteins. In this case, we are examining the ability of two GPCRs to assemble into a homodimer. BRET depends upon two different proteins that are attached to the N or C terminus of GPCRs. In these experiments a luciferase enzyme, from Renillareniformis is used as the donor. These enzyme catalyses the reaction found in the formula below. R. luciferase catalysis: coelenterazine + O2 CO2 + photon of light
119 The acceptor of the photon of light is yellow fluoresc ent protein (Y FP), which is fused to a second GPCR. When the two tagged GPCRs are cotransfected and assemble into a dimer, a transfer of energy takes place between the two attached proteins. The photon of light emitted by luciferase is transferred via resonance to Y FP, which emits energy pertaining to a unique wavelength. This wavelength is monitored and can be correlated to a proximity between the two tagged receptor, reflecting GPCR dimerization. This effect can be measured as long as the acceptor and donor proteins are within 10 nanometers of each other. When this technique was used to compare the WT and Y5.48A receptors the results showed no change between the WT and point mutated receptor It was hypothesized that domainswapped dimer formation would b e impaired and would be observable through BRET. Unfortunately the data indicated only slight differences in dimerization at WT and Y5.48A H1 receptors. The BRETMaxvalues which are similar to the Bmax in a saturation isotherm, indicated a value of 0.125 0.008 for the WT/WT homodimer and 0.107 0.11 for Y5.48A/Y5.48A. These values were not significantly different running contrary to the original hypothesis. However, upon reexamination of the Bakker 2004 paper that demonstrated domainswapping at the HH1R, it was observed that domainswapped dimers, represented by the coexpressed D3.32A and F6.52A receptors, are a small subset of the H1 receptor population. Mepyramine is able to l ab e l all H1 receptors with a Bmax of 21 4 pMol/mg of protein. In contrast, ( ) trans PAT labels only 3.4 1.0 pMol/mg of protein, about 1/7th of the overall receptor population. Intriguingly, if 1/7th of the reported BRETMaxvalue for the WT receptor is removed, the resultant value is 0.107. This is the exact experimental value for the
120 Y5.48A/Y5.48A homodimer.Despitebeing statistically insignificant, these results, coupled with the loss of saturation binding and function of ( ) trans PAT, suggest that domainswapped dimers are lost, while contact dimers remain at the Y5.48A point mu tated HH1R. Summary When all of the data for these experiments is compiled, the data reveals that Y5.48A is playing a significant role in the functional role of the HH1R. The ligand specific loss of binding up to 10 nM of ( ) trans PAT, coupled w ith its loss of inverse agonism suggested a ligand specific effect. The unaltered binding and functional profile of mepyramine, and the reduced, but measurable accumulation of IP3 produced by histamine incubation confirm that this is indeed a ligandspecific phenomenon. Given the unique nature of this alteration of function, and the previous knowledge that PAT binds preferentially to domainswapped dimers the logical conclusion is that the formation of domainswapped dimers is significantly impaired and/or abolished at the Y5.48A HH1R. BRET results were unable to confirm this result directly but did confirm the existence of contact dimers at Y5.48A; in addition to lending indirect evidence towards this conclusion.
121 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3[3H]-Mepyramine (nM)Specific [3H]-mepyramine binding (pmol/mg protein) Figure 3 1 Saturation binding isotherms for 3H mepyramine at the WT ( ) and Y5.48A HH1R ( ) performed in transiently transfected HEK 293 cells
122 0.0 2.5 5.0 7.5 10.0 12.5 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45[3H-PAT] (nM)[3H]-PAT binding (pmol/mg protein) 0.0 2.5 5.0 7.5 10.0 0.0 0.1 0.2 0.3 0.4[3H]-PAT (nM)[3H]-PAT binding (pmol/mg protein) 0.0 2.5 5.0 7.5 10.0 12.5 0.0 0.1 0.2 0.3 0.4[3H]-PAT (nM) [3H]-PAT Binding (pmol/mg protein) Figure 3 2 Saturation i sotherm s for3H ( ) trans PAT labeled WT ( ) and Y5.48A( ) HH1 receptors .T ransiently expressed in HEK 293 cells with varying protein concentrations. A) 0.01 mg/mL of protein. B) Depicts binding for Y5.48A at protein concentration of 0.02mg/mL. C) Depicts binding for Y5.48A at protein concentration of 0.1mg/ mL
123 Figure 3 3 S tructures of ligands examined in this section: ( ) trans PAT (top left) mepyramine (top right) ( ) cis PAT (left center), (+) trans PAT (right center) and histamine ( bottom )
124 -11 -10 -9 -8 -7 -6 -5 0 10 20 30 40 50 60 70 80 90 100 110Specific [3H]-Mepyramine Binding (%Control)[(-)-(trans )-PAT] Figure 3 4 Competitive binding curves for ( ) trans PAT at the WT ( ) and Y5.48A ( ) HH1R -9 -8 -7 -6 -5 -4 -3 50 100 150 200 250 300 350 400[Ligand]PLC Activity/[3H]-IP formation (% Basal Mean STDEV) -11 -10 -9 -8 -7 -6 -5 -4 50 60 70 80 90 100 110[Ligand]PLC Activity/[3H]-IP formation (% Basal Mean STDEV) Figure 3 5 Ligand mediated PLC/IP 3 functional activity at WT vs. Y5.48A HH1 receptors. A) histamine at WT ( and Y5.48A ( B) ( ) trans PAT at WT ( and Y5.48A ( ), and mepyramine at WT ( (
125 Table 3 1 C ompiled KD and BM ax values for 3H mepyramine and ( ) trans PAT at the WT and Y5.48A HH1R. HH1R Receptor type KD value for 3H mepyramine (nM) Bmax value for 3H mepyramine ( pMol/mg protein) K D value for 3H ( ) trans PAT (nM) B max value for 3H ( ) trans PAT ( pMol/mg protein ) WT 1.024 0.22 2. 98 1 0.5 1.34 0.3 0.345 0.06 Y5.48A 1.790 0.18 (P=0.40) 3.085 0. 6 (P=0. 89) N/A N/A Table 3 2 Tabulated KI values for the HH1R ligands examined in this section at both the WT and Y5.48A point mutated receptors. Ligand KI value for W T HH1R (nM) KI value for Y5.48A HH1R (nM) P value comparing KI at WT and Y5.48A Histamine 3000 280 15800 3000 0.0067** ( ) trans PAT 1.95 0.51 148.4 5.4 0.0014 ** (+) trans PAT 29.8 4.0 245.5 19.2 0.021* Mepyramine 2.14 0.4 4.905 1.2 0.026* ( ) cis PAT 13.7 0.2 143.6 0.6 0.0003*** Table 3 3 Displays the correlated functional values for the ligands examined in the values that are listed with Y5.48A data compare that column to the WT HH1R Ligand WT HH1R EC/IC50 (nM) WT HH1R E/Imax (% Basal response) Y5.48A HH 1 R EC/IC50 (nM) Y5.4 8A HH1R E/Imax (% Basal response) Histamine 159 24 401.1 8 0 8030 660 (P=0.0003***) 340.9 54 (P= 0.54) ( ) trans PAT 251 0.77 63.8 N/A N/A Mepyramine 2.51 0.4 61.4 1.3 3.35 0.4 (P=0.22) 63.49 3.2 (P=0.65)
126 CHAPTER 4 PROBING THE ROLE OF RESIDUE Y7.53A IN THE LIGAND BINDING AND ACTIVATION OF THE HH1R Rationale for U ndertaking T hese S tudies and L iterature R eview Contrary to the previous mutation Y5.48A, there is a litany of literature about position Y7.53A in aminergic GPCR s. In fact, 1B and 2 adrenergic receptors, andM1 muscarinic receptors, all of which are closely related to the histamine H1 receptor (Rosendorff et al., 2000). What makes this particular mutation so intriguing is the vast array of different results this mutation has produced. Everything from no change or slight increase in agonist affinity, to significant (150fold) increase in agonist affinity has been observed when mutating this residue (Prioleau et al., 2002; Rosendorff et al., 2000). This result is rather surprising, given this residue is located at the cytoplasmic end of TMD 7 far away from the orthosteric binding pocket of class A GPCRs this is demonstrated below in figure 41 Due to its location deep within TMD 7, this residue is going to have no direct effect on ligand binding, since it is too far away from the orthosteric binding pocket. Instead, any effects on ligand binding are likely to be due to alterations in the ability of G protein to couple to the receptor or alterations to the structure of TMD 7. If altered G protein coupling were occurring it would manifest itself as an alteration in affinity for agonists, with little to no change in inverse agonist affinity. In addition to the effects on agonist affi nity, Y7.53 has an incredible ability to influence GPCR signaling A quick survey of current GPCR literature reveals Y7.53 mutations have been shown to produce constitutively active receptors with varying activities, a loss of G protein coupling and even receptor internalization (Rosendorff et
127 al., 2000 Prioleau et al., 2002).These results strongly suggest that Y7.53 is a critical residue in the activation processes of GPCRs that are closely related to the HH1R. A very interesting and novel concept with r espect to the Y7.53A mutation is the concept of functional selectivity. Previous research has implicated this position as critical for GPCR function, G protein coupling, and internalization (Rosendorff et al., 2000;Prioleau et al., 2002). To the best of my knowledge the concept of functional selectivity, and Y7.53A in general, has yet to be investigated at the HH1R. Expanding these studies into the H1 receptor will provide further information a bout the activation of class A GPCRs, while investigating a conc ept that is novel to the Y7.53 literature, functional selectivity. By making use of ( ) trans PAT, ( ) cis PAB, and histamine it will be possible to investigate ligands displaying functional selectivity at this residue. By undertaking these studies, it is hypothesized that it will be possible to determine whether Y7.53 is critical to the activation of a single signaling pathway (Gor GS), or is involved in the overall activation process of the human H1 receptor(Gand G). Invoking the use of the library of PAT related compounds, in conjunction with traditional H1 ligands like triprolidine and mepyramine, will allow the undertaking of a thorough Y7.53 study including the effects of inverse agonists full agonists, and functionally selective ligands Materials and Methods The techniques and chemicals described in chapters two and three were used in this chapter as well. PLC measurement and competitive binding assays were also used in the chapter. Any additional techniques that were employed are described below. Measurement of AC activity/cAMP formation HEK293 cells transfected and transiently expressing wild type and Y7.53A H1 receptors were aliquoted into
128 microtubes, and preincubated for 510 min ute s with 5 mMphosphodiesterase inhibitor 3isobutyl 1 methylxanthine followed by treatments with drugs for 30 minute s in serum free media. Cells were lysed by 0.1 M HCl for 20 minute s on ice and supernatants were used in the measurement of cAMP concentration with the Direct Cyclic AMP Enzyme Immunoassay Kit (Assay Designs Inc., Ann Arbor, MI). Data were expressed as mean percentage of control cAMP formation as obtained by linear standard curve extrapolation by GraphPad Prism version 5.02. Experimental Results and Discussion Competitive B inding S tudies at Y7.53A Competitive binding assays have been completed for ( ) and (+) trans PAT triprolidine, mepyramine, ( ) cis PAB and histamine at WT and Y7.53A HH1Rs As was observed with other agonists, Y7.53A produced an increase in affinity for the endogenous agonist histamine. Its affinity was 1.0 0.1 M at Y7 53 A, increased 3fold from a WT affinity of 2.99 0.28 M at WT receptors. This increase in affinity was significant yielding a P= 0.0014**. In contrast, ( ) and (+) trans PAT both demonstrated a slight loss in affinity from WT, yielding KI values of 2.6 0.53 and 64 14 nM, respectively. When the KI values for the PAT ligands were compared to the WT receptor via the T test, it was observed t hat there were no differences (P>0.05) between any of these ligands and the WT receptor. Representative binding curves can be found in figure 41, and the binding affinities for all compounds studied at WT and Y7.53A HH1Rs, as well as their P values can be found in table 41. In addition to the ligands mentioned above, the affinities of the partial agonist ( ) cis PAB and the inverse agonists mepyramine and triprolidine were assessed. Mepyramine and triprolidine yielded KI values of 2.14 0.6 and 1.25 0.2
129 nMrespectively. When compared to the WT values in the literature, these compounds yielded P values greater than 0.05, indicating no significant differences in binding of these ligands between Y 7 53A and the WT HH1R. The last ligand that was investigated was the partial H1 agonist, ( ) cis PAB. At the Y7.53A point mutated receptor PAB produced a reduced affinity of 233.3 18nM, which was significantly different from the WT yielding a pvalue of 0.012* Representative binding curves for these data can be found in figure 42. Taken together these results mesh quite well with previous publications studying Y7.53 at receptors that are closely related to the HH1R. An increase in agonist affinity was previously observed at Y7.53 point mutations in the M1 muscarinic receptor, as well as the 5HT2Creceptor (Rosendorff et al., 2000; Prioleau et al., 2002) Prioleaus paper showed a 2 to 6 fold increase in agonist affinity for 5HT2C receptors, which is consistent with the findings observed for histamine (3fold increase). In addition both papers demonstrated that inverse agonists for 5HT2C receptors showed no change in affinity when Y7.53 is mutated. This result holds true for H1 receptors as well, given that ( ) trans PAT, (+) trans PAT, mepyramine, and triprolidine all show no statistical difference from their WT values. The perturbation to PABs KI is more puzzling, but is likely cause by perturbations to the native structure of TMD 7. Inositol Phosphate Production Mediat ed via G In order to investigate this hypothesis a thorough functional study w as undertaken involving the functional ly selective PAT ligands, tr a ditional H1 ligands, and the endogenous agonist histamine. These IP3/G Q functional studies have been complet ed with histamine traditional H1 inverse agonists mepyramine and triprolidine, and several ligands from the PAT family including ( ) trans PAT .Unfortunately, no response was
130 oberved for the H1partial agonist ( ) cis PAB after six separate experiments whe re histamine produced a reproducable curve. This is likely due to the small window of activation for the Y7.53A receptor. Even the full IP3endogenous agonist histamine, produces only a 40% increase over basal, as apposed to a 400% increase at the WT receptor. Despite the high limit of detection for the column method of measuring inositol phosphates, accurately measuring a partial agonist response that is conservatively estimated at around 20% of basal proved to be too tall a task. The EC50 value for histamine at Y7.53A was 1.515 0.3 M reduced approximately ten fold from the WT value of 0.157 0.012.When the se valuesare compared in a t test the resultant P Value is 0.00 99**, inidicating a significant dichotomy in the ability of histamine to ac t ivate G Q. Comparing the Emax signals between WT andY7.53A revealed an even more interesting trend. As demonstrated in figure42 below, a nearly fivefold reduction in efficacy is observed between Y7.53A and the WT receptor. The Emax at the WT receptor is 401.1 81% b asal, while Y7.53A gives an Emax of 143 8.0% yielding a P<0.0001*** These reults indicate that histamine is less effective at activating the receptor, in terms of both efficacy and maximal response, despite having a higher binding affinity for the receptor. This is strongly suggestive of a reduction in the ability of G Q to couple to the receptor at the Y7.53A point mutation. Given the reduction of both efficacy and affinity at the HH1R, the location of Y7.53 deep within TMD 7, and the previously reported works at closely related class A GPCRs, the Y7.53A point mutation drastically reduces the ability of the HH1R to couple to G Q. To build upon the binding results m epyramine and ( ) trans PAT both of which act as inverse agonist s, were examined in these functional experiments. ( ) trans PAT
131 yielded an IC50value of 225 30 nM at Y7.53A and 251.7 18 nM at the WT receptor, giving a P =0.62. Mepyramine gave an IC50 of 2.2 0.6 nM at Y7.53A, which was nearly identical to the WT value of 2.51 0.8 nM A t test for significance yielded a pvalue of 0.75 A t test of the IM ax values for both ligands revealed the values were unchanged from the WT HH1R. At Y7.53A, mepyramine produced an Imax of 61.4 1.3% basal (P=0.75), while ( ) trans PAT yielded a value of 59.9 6 (P=0.51). Representative functional curves for these data can be found in figure 43. These findings are identical to the established literature for Y7.53 mutations that demonstrate that the affinity of inverse agonists is unchanged when compared to the WT 5HT2C receptor (Rosendorff e t al., 2000; Prioleau et al., 2002) To the best of my knowledge the comparison of the WT IC50 values for inverse agonists, to those at Y7.53A is a novel investigation at the H1 receptor The results of which shed light on the effects that reduced G p rotein coupling has on the efficacy of inverse agonists. Based upon the findings listed above, the reduced G protein coupling at Y7.53A has no effect on the functional response of inverse agoinsts at the HH1R .This result fits well with our understanding of inverse agonists from current GPCR theory. Inverse agonists reduce the basal level of signalling by inducing, selecting, or stabilizing a conformation of the receptor that does not favor g protein coupling. It seems that the reduced ability of Y7.53A to c ouple with G Q does not have an affect on the ability of inverse agonists to reduce basal signalling. It should also be noted that a similar effect was observed for the 5HT2C inverse agonist, SB206553, hinting at strong similarities in the activation relay s between the two closely related receptors (Rosendorff et al., 2000)
132 AC A ctivity/cAMP F ormation Mediated via G S To further investigate the effects of Y7.53A on HH1R function, the ability of histamine to stimulate cAMP formation was assessed. At the WT HH1R histamine has an EC50 of 1.41 0.14 M and an Emax of 250 33 % basal cAMP accumulation. When identical experiments were performed with the Y7.53A HH1R there was no increase over basal singalling observed. Inidcating the ability of G S to couple to the HH1R has been abolished. Representative functional curves for WT and Y7.53A can be found in figure 44. The loss of the ability of histamine mediated cAMP produciton at Y7.53A stands in stark contrast to the reduced, yet measurable accumulation of IP3 with the same ligand. Since the mutation is the same, there are no differences in binding of histamine to the receptor and any change in ligand binding must be due to the binding of the G protein. This strongly implicates the G protein in the lack of cAMP production. Previous literature has suggested that Y7.53 mutations can uncouple the G protein from the receptor and it appears that this is what is occuring with G S. Summary It has previously been suggested that this mutation can uncouple the G protein from the receptor (Rosendorff et al., 2000). These experiments suggest that a similar result is taking place at the HH1R and G S. This is supported by the fact that histamine is able to bind Y7.53A, and is able to function, albeit with reduced efficacy, via G Q toproduce inositol phosphates It is an intriguing finding that G Sand G Q show unique results when examined at Y7.53A in functional assays.Previous literature suggested that G proteins could not couple as effectively to the receptor when the tyrosine residue at Y7.53 was lost, but neglected to examine the concept of two different G proteins coupling to the same receptor (Rosendorff et al., 2000; Prioleau et al., 2002) From
133 these experiments in can be inferred that G Sand G Q have unique affinities for the Y7.53A mutation, and that these affinities differ from the native HH1R. G Q appears to have a reduced affinity for the Y7.53A receptor, inidcated by its weakened ability to stimulate IP3 formation ; while all histamine signaling thru G Sis lost indicating a lack of receptor coupling Since the Y7.53A receptor is going to have identical conformations presented to both G proteins, it can be said that G Sand G Q must have differing affinities. This concept fits well with our understanding of functional selectivity and hints that native receptor conformations preferred by G Sand G Q should be distinct from one another. By extension, it would seem plaus ible that multiple binding modes could exist for the same ligand; and that the conformation selected/or induced by the ligand may be influenced by the G protein to which the receptor is coupled. These distinct conformations would allow ligands to stabilize, induce, and/or select a receptor conformation that preferentially targets one G protein over another, providing a possible molecular explaination for the phenomenon known as functional selectivity.
134 Figure 4 1 GPCR snake diagram for the HH1R illustrating the location of Y7.53 towards the intracellular side of TMD 7 very close to the G protein coupling interfaces of ICL 2 & 3 and the C terminus of the receptor. Table 4 1 Compiled affinities of various H1 ligands at the WT and Y7.53A receptors Ligand KI value for WT HH1R (nM) KI value for Y7.53A HH1R (nM) P value comparing KI at WT and Y5.48A Histamine 3000 280 1000 3 00 0.00 99 ** ( ) trans PAT 1.95 0.51 0 88 5.4 0.1 8 (+) trans PAT 29.8 4.0 14 2 3.7 0.2 2 Mepyramine 2.14 0.4 2.84 0.0 2 0.2 2 Triprolidine 1 25 0.2 2.14 0. 4 0.99 ( ) cis PAB 57.5 9.2 260.9 40 0.012* D Y 107 W 158 K 191 T 194 N F Y 198 W 428 F F Y 432 Y 458 Y 468 108 435 199 200 431
135 -10 -9 -8 -7 -6 -5 -4 -3 0 10 20 30 40 50 60 70 80 90 100 110Log [Ligand] (M)[3H]-Mepyramine Binding (% Specific Bound SEM) Figure 4 2 Representative competition binding curves for various H1ligandsat the Y7.53A point mutated receptor : ( ) trans PAT ( ), (+)trans PAT ( ), histamine ( ), mepyramine ( ), triprolidine ( ), and ( ) cis PAB as ( ). -10 -9 -8 -7 -6 -5 -4 -3 0 50 100 150 200 250 300 350log[Histamine] (M)PLC Activity/[3H]-IP formation (% Basal Mean STDEV) -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 50 60 70 80 90 100 110log[Ligand] (M)PLC Activity/[3H]-IP formation (% Basal Mean STDEV) Figure 4 3 Representative G/PLC functional curves for H1 ligands at the Y7.53A HH1R .A) histamine at WT ( ) and Y7.53A ( ) receptors. B) mepyramine ( )and ( ) trans PAT ( ).
136 Table 4 2 Tabulates the EC/IC50 and E/Imax values of various ligands at the WT and Y7.53A point mutated receptors Ligand W T HH1R EC/IC50 (nM) W T HH1R Emax/Imax (% Basal response) Y7 53 A HH 1 R EC/IC50 (nM) Y 7 53 A HH 1 R Emax/Imax (% Basal response) Histamine 159 24 401.1 81 1510 3 0 0 (P=0.0003***) 143 8.0 (P< 0. 0001*** ) ( ) trans PAT 251 .8 18 63.8 2.2 225 30 (P=0.62) 59.9 6 (P=0.51) Mepyramine 2.51 0.4 61.4 1.3 2.2 0. 46 (P=0. 75 ) 6 1 .4 1.3 (P=0.65) -10 -9 -8 -7 -6 -5 -4 -3 100 150 200 250[Histamine]AC Activity/ cAMP formation (% Basal Mean STDEV) Figure 4 4 RepresentativecAMP/AC functional curves for histamine stimulation at WT ( ) and Y7.53A ( = assay 1) and( =assay 2 ) HH1R.
137 CONCLUDING REMARKS It is my hope that these studies of the HH1R have advanced the understanding of how H1 ligands bind to and activate the receptor. The unique binding and functional profile of ( ) trans PAT represents a new class of ligand and understanding its subtle nuances will provide a basis for studying other ligands possessing fu nctional selectivity at other GPCRs. The ability of this ligand to inhibit the PLC/IP signaling pathway and activate the AC/cAMP at the same receptor is a rare therapeutic quality. In this particular case, the unfavorable PLC/IP pathway that caus es allergic responses is inhibited, while the pathway with therapeutic potential to stimulate neurotransmitter synthesis is activated. This unusual ability to activate the therapeutic pathway at the HH1R and inhibit the pathway that causes side effects represents a significant step in GPCR therapeutics. By understanding the molecular basis for this particular effect it is hoped that additional ligands at differing GPCRs could be designed to exploit the concept of functional selectivity. In doing so, it should be possible to produce more specific drugs that activate only their desired functional property, while drastically reduc ing their nonspecific side effects In my first specific aim I set out to better understand the binding of our lead compound ( ) trans PAT and a series of similar analogues that were based off of this parent scaffold. My studies revealed that there are steric and electronic factors to consider around the amine region, which interacts ionically with D3.32A. This interact ion is critical for the binding of all ligands to the HH1R. When meta and para substituted PAT analogues were examined it was determined that each enantiomer seemed to bind to unique regions of the receptor. More specifically, the pendant phenyl ring of the ligand produces unique interactions with the HH1R that lead the ( ) and (+) enantiomers to bind
138 differently to the receptor. Through these studies it was also learned that larger substituents such as bromine and cyclooctyl rings are too lar ge to be accommodated within the orthosteric binding pocket of the HH1R. By better understanding the substitutions that cause highaffinity binding to the receptor it is hoped that our lab can produce more potent ligands that bi nd more specifically to the HH1R. The goal of my second aim was to understand what made ( ) trans PAT such a unique ligand, meaning what was the root cause of its functional selectivity and its ability to label only a partial subset of theoverall H1receptor population. This investiga t i on began with a single amino acid residue (Y5.48A) that produced a loss of specific 3H ( ) trans PAT binding, but did not prohibit 3H mepyramine from binding to the same mutation. From there, it was observed that PAT was unable to bind or function at Y5.48A, while histamine and mepyramine were able to function and bind to the receptor. This made the effects of Y5.48A on PAT ligand spec ific, and led to the conclusion that PAT exerts its effects through domainswapped dimers. The mutation of this residue to alanine precluded the formation of these dimers and subsequently the loss of PAT binding and function. These results were supported by BRET studies and helped us arrive at the conclusion that PAT functioned exclusively through domainswapped dimers. In my last specific aim the goal was to investigate a particular residue, Y7.53 that was known to be involved in the functional processes of GPCRs that are closely phylogenically related to the HH1R The goal of these studies was to determine whether or not an alanine mutation at this particular position would have differing effects upon the pathways that the HH1R ca n activate. My studies revealed a dichotomy between the
139 ability of the receptor to activate PLC/IP and AC/cAMP at Y7.53A. The overall response for histamine at AC/cAMP was ablated, while the PLC/IP pathway was still able to function, albeit in a significan tly reduced manner .It was also observed that the inverse agonists ( ) trans PAT and mepyramine were still able to function at Y7.53A. From these experiments in can be inferred that G Sand G Q have unique affinities for the Y7.53A mutation, and that these affinities differ from the native HH1R. G Q appears to have a reduced affinity for the Y7.53A receptor, indicated by its weakened ability to stimulate IP3 formation; while all histamine si gnaling thr o u gh G Sis lost indicating a lack of receptor coupling. Since the Y7.53A receptor is going to have identical conformations presented to both G proteins, it can be said that G Sand G Q must have differing affinities. This concept fits well with our understanding of functional selectivity and hints that native receptor conformations preferred by G Sand G Q should be distinct from one another. By extension, it would seem plausible that mul tiple binding modes could exist for the same ligand; and that the conformation selected/or induced by the ligand may be influenced by the G protein to which the receptor is coupled. These distinct conformations would allow ligands to stabilize, induce, and/or select a receptor conformation that preferentially targets one G protein over another, providing a possible molecular explanation for the phenomenon known as functional selectivity. Throughout these studies there were a couple shortcomings that warrant mentioning here. The inability to produce a double mutation for D3.32A/Y5.48A and Y5.48A/F6.52A would have allowed me to directly prove the domain swapped dimer hypothesis for PAT. As things stand I had to use indirect evidence to prove this h ypothesis, which while sufficient was not ideal If I had been able to produce these two
140 double mutations it would have been possible to demonst r a te conclusively that Y5.48A prevents the receptor from forming domainswapped dimers. I also wish I had more time to study Y7.53A, this particular mutation proved to be quite interesting and seemed to have a pronounced effect on the ability of the receptor to couple to the G protein. My studies here illustrate for the first time that Y7.53A has differing effects on the ability of the HH1R to activate two different G proteins. If I had more time I would have loved to see if this phenomenon also occurredin receptors closely related to the HH1R. It is my hope that these studies have contributed to our overall understanding of GPCRs in general and more specifically the concept of functional selectivity. Ligand directed functional selectivity, as demonst rated with ( ) trans PAT,possess significant potential for future drug design. Although this may not be realized in the near future, it is my sincere hope that my studies lay the groundwork for others to follow up on, and will furtherour understanding of these critical transmembrane receptors; leading to the design of more specific and e fficacious drugs in the near future.
141 LIST OF REFERENCES Anthony et al. (2009). Dualsteric GPCR targeting: a novel route to binding and signaling pathway selectivity. FASEB J. 442 450. Bakker et al. (2004). Domain swapping in the h uman h istamine H1r eceptor. The Journal of Pharmacology anmd Experimental Therapeutics 131 138. Ballesteros et al. (2001). Structural m imicry in G p roteincoupled r eceptors: Implications of the h igh r esolution s tructure of r hodopsin for structure f unction a nalysis of r hodopsinl ike r eceptors. Mol ecular Pharmacology 1 19. Bhattacharya et al. (2008). Ligandstabilized conformational states of human beta2 adrenergic receptor: insight into G proteincoupled receptor activation. Biophysical Journal 20272042. Booth et al. (2009). (1R, 3S) ( Tra ns PAT: A novel full efficacy serotonin 5HT2C receptor agonist with 5 HT2A and 5HT2B receptor inverse agonist/antagonist activity. European Journal of Pharmacology 1 9. Booth et al. (2002). A n ovel p henylaminotetralin r adioligand r eveals a subpopulation of h istamine H1r eceptors. The Journal of Pharmacology and Experimental Procedures 328336. Booth et al. (2008). Molecular d eterminants of liganddirected singaling for the histamine H1 receptor. Inflamm. Res. S01 S02. Booth et al. (1993). New sigma like r eceptor r ecognized by n ovel p henylaminotetralins: l igand b inding and f unctional studies. The American Chemical Society for Pharmacology and Experimental Techniques 12321239. Booth, & Moniri. (2006). Role of PKA and PKC in histamine H1 receptor mediated activation of catecholamine neurotransmitter synthesis. Neuroscience Letters 249253. Brunton, Lazo, & Parker, a. (2006). Histamine, b radykinin, and t heir a ntagonists. In The Pharmacological Basis of Therapeutics (pp. 629 652). McGraw Hill. Br uysters et al. (2004). Mutational a nalysis of the h istamine H1r eceptor b inding p ocket of h istaprodifens. European Journal of Pharmacology 55 63. Bucholtz et al. (1999). Synthesis, e valuation, and c omparative m olecular f ield a nalysis of 1 phenyl 3 amino 1,2,3,4tetrahydronaphthalenes as l igands for the Histamine H1r eceptor. J. Med. Chem. 30413054. Chen et al. (2003). Chemical kindling induced by pentylenetetrazol in h istamine H1 receptor gene knockout mice (H1KO), histidine decarboxylase deficient mice (HDC/ ) and mast cell deficient (W/Wv) mice. Brain Research 162166.
142 Christopolous, & Kenakin. (2002). G p rotein c oupled r eceptor a llosterism and complexing. Pharmacological Reviews 323 374. Fang, Travers, and Booth RG (2011) PrePublication Flanagan, C. (2005). A GPCR That Is Not DRY. Molecular Pharmacology 1 3. Gillard et al. (2002). Binding characteristics of cetirizine and l evocetirizine to h uman H1h istamine r eceptors: contribution of Lys (191) and Thr(194). Molecular Pharmacology 39 1 399. Haas and Panula. (2003). The Role of h istamine and the t u be reromamillary n ucleus in the n ervous system. Nature Reviews: Neuroscience 121 130. Haas et al. (2008). Histamine in the CNS. Physiol Rev. 11831241. Hamm, H. (2001). How activated recept ors couple to G proteins. PNAS 48194821. Hermans, E. (2003). Biochemical and pharmacological control of the multiplicity of coupling at G proteincoupled receptors. Pharmacology and Therapeutics 25 44. Hoque, & Chesson. (2008). Zolpidem i nduced sleepwalking, sleep r elated e ating d isorder, and sleepd riving: f luorine18f lourodeoxyglucose Positron Emission Tomography a nalysis, and a l iterature r eview of o ther u nexpected clinical e ffects of z olpidem. Journal of Clinical Sleep Medicene 471 476. Ito, C. (2009). Histamine H3receptor inverse agonists as novel antipsychotics. Cent Nerv Syst Agents Med Chem 132136. Johnston, & Siderovski. (2007). Receptor m ediated a ctivation of h eterotrimeric G p roteins: Current Structural Insights. Molecular Phar macology 219230. Jongejan, & Leurs. (2005). Delineation of r eceptor l igand i nteractions at the h uman h istamine H1r eceptor by a combined a pproach of site d irected m utagensis and computational t echniques or h ow to b ind the H1r eceptor. Arch. Phram. Chem. 248259. Kobilka, & Deupi. (2007). Conformational complexity of G p rotein c oupled r eceptors. TRENDS in Pharmacological Sciences 397 406. Kobilka, & Rohrer. (1998). Insights from in vivo modification of adrenergic receptor gene expression. Annu. Rev. Pharmacol. Toxicol 351373. Kobilka, B. (2006). G protein coupled receptor structure and activation. Biochimica et Biophysica Acta. 794 807. Leurs et al. (1995). Molecular p harmacological a spects of h istamine r eceptors. Pharmac. Ther. 413 463.
143 Liang et al. (2003). Organization of the G p rotein coupled r eceptors r hodopsin and o psin in n ative m embranes. Journal of Biological Chemistry 21655 21662. Malfertheiner et al. (2009). Peptic ulcer disease. Lancet 14491461. Meguro et al. (1995). Effects of t hioperamide, a h istamine H3 Ligand Pharmacology Biochemistry and Behavior 321 325. Mirzadegan et al. (2003). Sequence a nalyses of G p rotein c oupled r eceptors: Similarities to r hodopsin. Biochemistry 27592767. Moniri et al. (2004). Ligandd irected f u nctional h eterogeneity of h istamine H1r eceptors: n ovel d ual f unction l igands selectively a ctivate and block H1mediated phospholipase C and adenylyl cyclase signaling JPET 274 281. Offermans, S. (2003). Gproteins as transducers in transmembrane signalling. Biophysics and Molecular Biology 101 130. Panula et al. (1984). Histaminecontaining neurons in the rat hypothalamus. Proc. Nati. Acad. Sci. 25722576. Pin et al. (2004). Acti vation mechanism of the heterodimeric GABAB receptor. Biochemical Pharmacology 15651572. Prioleau et al. (2002). Conserved h elix 7 t yrosine a cts as a m ultistate conformational switch in the 5HT2Cr eceptor. The Journal of Biological Chemistry 36577 6584. Rosendorff et al. (2000). Conserved helix 7 tyrosine functions as an activation relay at the serotnin 5HT2C receptor. Molecular Brain Research 90 96. Roth et al. (2007). Efficacy and safety of d oxepin 1 mg, 3 mg, and 6 mg in a dults with p rimary i nsomni a. Sleep 1555 1561. Saxena, A., & Saxena, M. (1992). Developments in antihistamines (H1). Progress in Drug Research 35 125. Scharf et al. (2008). Efficacy and safety of d oxepin 1 mg, 3 mg, and 6 mg in e lderly p atients w ith p rimary i nsomnia: ar andomized, d oubleBlind, p lacebocontrolled crossover study. J Clin Psychiatry 155764. Shi et al. (2002). Beta2 Adrenergic r eceptor a ctivation: modulation of the proline kink in transmembrane 6 by a rotamer toggle switch. The Journal of Biological Chemis try 40989 40996. Smit et al. (1999). Molecular Propeties and signaling p athways of the H1r eceptor. Clinical and Experimental Allergy 19 28.
144 Spiegel, & Weinstein. (2004). Inherited diseases involving g proteins and g protein coupled receptors Annu. Rev Med. 27 39. Stahl. (2008). Selective histamine h1 antagonism: novel hypnotic Trends in Psychopharmacology 10271038. Torrice et al. (2009). Probing the role of the cation GPCRs using unnatural amino acids. Proce edings of the National Academy of Sceinces PNAS 11919 11924. Urban et al. (2007). Functional Selectivity and Classical Concepts of Quantitative Pharmacology. JPET 1 13. Van Rijn et al. (2006). Oligomerization of recombinant and endogenously expressed H4receptors. Mol Pharm 604615. Visual Elements Group 17: The Halogens (n.d.). Retrieved June 28, 2010, from Royal Society of Chemistry: http://www.rsc.org/chemsoc/visualelements/pages/data/intro_groupvii_data.html Wada et al. (1984). Distribution of the histaminergic neuron system in the central nervous system of rats; a fluorescent immunohistochemical analysis with histidine decarboxylase as a marker. Brain Res 13 25. Whorton et al. (2007). A monomeric G proteincoupled receptor isolated in a highdensi ty lipoprotein particle efficiently activates its G protein. PNAS 76827687. Wieland et al. (1999). Mutational Analysis of the Antagonist binding Site of the Histamine H1 Receptor Journal of Biological Chemistry 2999430000. Yanai, & Toshiro. (2007). The physiological and pathophysiological roles of neuronal histamine: An insight from human PET Studies. Pharmacology and Thereapeutics 1 15.
145 BIOGRAPHICAL SKETCH The author was born in 1984, to Bill and Linda Travers in Baltimore, MD and was lucky enough to have a younger brother Garrett. He attended John Carroll High School in Bel Air, MD. After graduation, he attened Elon University in Elon, NC. He graduated with an American Chemical Society certified Bachelor of S cience degree in chemistry and a minor in mathematics. Immediately after graduation, he started graduate school at the University of Florida, under the tutelage of Dr. Raymond Booth and completed his Doctor of Philosophy from Universityof Florida in Augus t 2011.