Structural and immunological domain analysis of the carboxyterminal tails of the high molecular weight neurofilament sub...


Material Information

Structural and immunological domain analysis of the carboxyterminal tails of the high molecular weight neurofilament subunit proteins NF-M and NF-H
Physical Description:
vii, 129 leaves : ill. ; 29 cm.
Harris, Jeffrey Mark, 1965-
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Neurofilament Proteins -- ultrastructure   ( mesh )
Neurofilament Proteins -- immunology   ( mesh )
Neurofilament Proteins -- isolation & purification   ( mesh )
Intermediate Filaments -- ultrastructure   ( mesh )
Intermediate Filaments -- immunology   ( mesh )
Epitopes   ( mesh )
Immunohistochemistry -- methods   ( mesh )
Parkinson Disease -- pathology   ( mesh )
Rats   ( mesh )
Department of Neuroscience thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Medicine -- Department of Neuroscience -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1992.
Bibliography: leaves 113-128.
Statement of Responsibility:
by Jeffrey Mark Harris.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002328688
oclc - 50910571
notis - ALT2330
System ID:

This item is only available as the following downloads:

Full Text








My ability to complete this dissertation has been

facilitated by the help of several people. Of greatest

assistance has been my mentor, Dr. Gerry Shaw, who has proven

to be not only a superb resource but also an inspirational

example of a creative and hard-working scientist as well as a

truly good person. I will fondly remember the innumerable

hours he has spent with me working at the benchtop while

engaged in spirited discussions. Dr. Gudrun Bennett has also

been quite helpful as a source of insight and direct

experimental assistance. The instructional efforts of the

other members of my committee, Dr. Edward Wakeland and Dr.

Robert Nicholls, are also much appreciated. Throughout my

training several faculty members have been quite helpful in

allowing me to use their facilities; in this regard I thank

Drs. Harry Nick, Tom O'Brien, Floyd Thompson, Greg Erdos, Gino

Van Heeke, Mike King, Marieta Heaton, John MacLennan, Edward

Wakeland, Kyle Rarey, Bill Luttge and Skip Eaker, as well as

Drs. Virginia Lee and John Trojanowski at the University of

Pennsylvania. Many other people have been technically helpful

in completing this work including Scherwin Henry, Chris Browe,

David Lado, Jace Dinehart, Benne Parten, Judy Sallustio,

Jonathon Crone, and Drs. Bill Dougall, Kathy Ketchum, Ken


Horlick, Bill Wong, J.J. Warner and Wouter ten-Cate. In

addition, my fellow graduate student, Laura Errante, has been

a good role model through her diligent working habits and has

contributed technical advice on several occasions.

Outside of the lab my parents, a few close friends and

roommates have been quite supportive through their efforts and

interest. My parents, Margie and Morty Harris, have given me

much love, help and encouragement; I could never thank them

enough for all they have done. Drs. Jennifer Poulakos and

Brian Masters have been invaluable friends and colleagues,

both while they were in Gainesville and since their departure.

Tim Bock and Malcolm Lightner have not only been good friends,

but have also broadened my interests and knowledge. Blossom

Davies has been an invaluable friend and companion, who has

provided considerable emotional and intellectual support.

Financial sponsorship for supplies, travel, and stipend

was provided by N.I.H. grants NS22695 and AG07470 to Gerry

Shaw, as well as by a fellowship by the Center for the

Neurobiology of Aging.




ABSTRACT . . .. vi

Cytoskeletal Networks . 1
Mammalian Neurofilament Subunit Proteins 2
Mammalian NF Protein Structural Features 5
Neurofilament Assembly . 8
Neurofilament Involvement in Disease .. 10
Putative Functions . .. 11
Experimental Goals . .. 13

Introduction . . 15
Methods . . 18
Construction of Fusion Proteins .. 18
Isolation and Verification of Fusion Protein
Clones . . 23
Fusion Protein Expression . .. 24
Chemical Cleavage of Fusion Proteins 25
Results ............... ... 26
Results . . 26
Discussion . . 31

Introduction . . 33
Methods .......................34
Methods . . 34
Available NF Antibodies . .. 34
Western Blotting . .. 35
Production of Fusion Protein Antibodies 35
Results . . 36
Discussion . . .. 50

Introduction. . .. .57
Methods . . 58
Results. . .. 61
Discussion . . .. 90

Introduction . . 97
Methods .. . 98
Results . .. .. 100
Discussion . . 106

NF Fusion Proteins . 108
NF Fusion Protein Antibodies .. 109
Immunohistochemical Distribution of NF-M and NF-H 109
NF Distribution in Pathological Human Brain 111
Conclusions . .111



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



Jeffrey Mark Harris

August 1992

Chairperson: Dr. Gerry Shaw
Major Department: Neuroscience

Unlike the smooth 10 nm intermediate filaments (IFs)

found in the cytoskeleton of most cell types, neurofilaments

(NFs), the IFs of neurons, have side-arms projecting out

radially from a 10 nm core filament. The projection arms are

thought to contain the long carboxyterminal tails which are

unique to the middle (NF-M) and high (NF-H) molecular weight

neurofilament subunit proteins. The tails of NF-M and NF-H

appear to contain several domains based upon amino acid

sequence analysis. This study reports on the construction of

cDNA subclones containing these putative domains from rat NF-M

and NF-H, the expression and purification of the corresponding

fusion-proteins, the production and characterization of

polyclonal and monoclonal antibodies against these putative

domains, as well as the histological and pathological

distribution of NF-M and NF-H using these antibodies.


Epitopes for a large number of widely used monoclonal

neurofilament antibodies were mapped to these putative domains

including the commercially available antibodies NN18 and N52.

Several unusual findings were made concerning the

immunohistochemical distribution of NF-M and NF-H in the rat

nervous system. We describe a novel cell type in the

granular layer of the vestibulocerebellum which is strongly

immunoreactive for NF-H, but not for NF-M. We demonstrate NF

reactivity in a subpopulation of cerebellar parallel fibers,

previously thought not to contain NFs. The type I spiral

ganglion neurons are shown to be strongly immunoreactive for

NF-M, but not for NF-H. An immunological masking event on a

specified segment of NF-M is demonstrated to begin at

postnatal day 4 in rat myenteric neurons. In addition, an

immunohistochemical examination of pathological human brains

revealed that the carboxyterminal tails of NF-M and NF-H are

present in the Lewy body inclusions which are pathognomonic

for Parkinson's disease, but are unlikely to be specific

components of the neurofibrillary tangles found in Alzheimer's




Cytoskeletal Networks

Three major cytoskeletal networks are known to exist in

most mammalian cells: the microfilaments composed of

polymerized actin, the microtubules consisting of polymerized

tubulin and the intermediate filaments (IFs), which are

assembled from various cell-type specific subunits. The roles

of the microfilamentous and microtubular networks have been

widely characterized with respect to many cellular processes

including growth, motility, mitosis, intracellular trafficking

and production of tensile strength. The function of

intermediate filaments, however, still remains elusive. For

lack of other evidence, these structures are often assigned a

mechanical role. Because different subunit proteins are

utilized in different cell types, it has been hypothesized

that intermediate filaments are involved in determination or

maintenance of cell shape. Because these cell-type specific

subunits are biochemically and immunologically distinct, they

have proven to be useful cell-type specific markers during

development and oncogenesis when morphological characteristics

are often indiscriminate. Intermediate filament subunits

include: the various acidic and basic keratins found in



epithelia, desmin in muscle, glial fibrillary acidic protein

(GFAP) in astrocytes and Bergmann glia and vimentin in

mesenchymal cells, as well as an increasing number of

neurofilament protein subunits; it is of note that the lamins,

which comprise a major part of the scaffolding found inside

the inner nuclear membrane of all eukaryotic cells, are also

considered members of the intermediate filament family based

on protein sequence (McKeon et al., 1986). A classification

scheme to group different intermediate filament subunit

proteins has been proposed which is based upon sequence

similarity and intron placement (Steinert and Roop, 1988);

Table 1-1 is an updated modification of this classification


Mammalian Neurofilament Subunit Proteins

The 10 nm diameter neurofilaments (NFs) seen in the

electron microscope are now known to correspond to the

neurofibrils seen in the light microscope following the use of

silver stains. Using ultrathin section electron microscopy

neurofilaments can be distinguished from the intermediate

filaments found in other cell types, including glia, due to

the presence of periodic fine projections protruding between

filaments within bundles; these protrusions can be used to

differentiate between processes of neurons and glia.

Technically a genuine neurofilament subunit protein is

distinguished from a neurofilament associated protein based



Class Name
I acidic keratins

II basic keratins

III vimentin
glial fibrillary acidic protein (GFAP)


V nuclear lamins

VI nestin

on its ability to actually incorporate into the neurofilament

backbone in vivo. Until recently the neurofilament protein

subunits were referred to as a "triplet" because it was

thought that only three subunits existed, namely NF-L, NF-M

and NF-H; these subunits are so named because of their

relatively low, middle or high molecular weight. The triplet

proteins were originally described as major components of slow

axonal transport with apparent molecular weights by SDS-PAGE

of approximately 68, 145 and 200 kDa (Hoffman and Lasek, 1975;

Liem et al., 1978). With complete sequence information it now

appears that the actual molecular weights should be 60, -95

and -115 kDa;. this discrepancy in calculated and observed

molecular weight is apparently due to secondary structure

influenced to a large degree by multiple phosphorylation sites


in the large carboxyterminal tails (Julien and Mushynski,

1982; Kaufmann et al., 1984).

Although the triplet proteins are still thought to be the

predominant neurofilament subunits of the adult (i.e., post-

mitotic) nervous system (Tapscott et al., 1981), other

neurofilament subunit proteins seem to be more prevalent at

earlier developmental stages. Vimentin is found

developmentally in mesenchymal cells and dividing

neuroepithelial cells and is often coexpressed with

neurofilament triplet subunits during post-mitotic

development. In the adult vimentin is seen in most cells of

mesenchymal origin as well as in reactive microglia and even

in some mature neurons (Drager et al., 1984; Shaw and Weber,

1983; Schwob et al., 1986); therefore, vimentin could be

classified as a genuine neurofilament subunit. Two more

recently described neurofilament subunits named a-internexin

(66kD) (Pachter and Liem, 1985) and peripherin (57kD) (Portier

et al., 1984; Parysek and Goldman, 1987; Leonard et al., 1988)

have somewhat complementary distributions in the developing

central (CNS) and peripheral (PNS) nervous systems,

respectively. Expression of a-internexin appears to precede

that of the triplet subunits in the CNS and can also be found

in neuronal cell types previously thought not to possess

neurofilaments (Kaplan et al., 1990). In addition, peripherin

has now been found to have two additional, though less

abundant, forms derived from differential mRNA splicing


(Landon, et al., 1989). Another newly described neural

intermediate filament protein, nestin, is found in

neuroepithelial stem cells (radial glia and their developing

progeny) and appears to be quite distinct in sequence,

structure, intron pattern and size from other neurofilament

protein subunits (Lendahl et al., 1990).

Mammalian NF Protein Structural Features

Structural characteristics of all intermediate filaments

(IFs) include a globular amino-terminal head, a central a-

helical rod and a carboxyterminal tail. The central rod

region is conserved in all IF subunits. Filament assembly is

thought to involve the rod region, which always contains long

heptad repeats of large hydrophobic amino acids with

relatively rare interruptions. This conserved heptad repeat

structure allows the a-helical regions of two subunits to be

stabilized by intercalation of hydrophobic residues forming an

a-helical coiled-coil (Crick, 1953). Alpha-helical coiled-

coils are also found in dimerized molecules like myosin heavy

chains, tropomyosin and the DNA binding proteins c-fos and

c-jun. The two major heptad breaks, called linkers or

spacers, are usually created by proline residues and define

the borders between coil la, coil lb and coil 2 (see Figure 1-

1). Nestin, NF-M and NF-H do not appear to contain the first

linker such that they may have an unbroken coil 1 (Lees et

al., 1988; Myers et al., 1987). Two heptad "stutters" are



C a


c *4


. 0 E .. 2 (S
e % B iL u QL 8

.> )

L. D ,,
L) ._ a- al)


consistently found near the beginning of and about three-

fifths through coil 2 due to a single residue insertion, thus

shifting the heptad repeat out of register. Further

comparison of the rod domains across IF protein subunits

reveals that the a-helical regions are almost invariant in

length, whereas the linkers are of somewhat variable length.

The amino-terminal region of coil la and the carboxyterminal

end of coil 2 are very highly conserved in amino acid

sequence. This high conservation of coil 2 region accounts

for the ability of an antibody (a-IFA) described by Pruss et

al. (1981) to detect all known IF protein subunits (Geisler et

al., 1983).

Across all IFs the head and tail regions are most

variable, though strong similarities in these segments can be

found in members within IF classes and occasionally between IF

classes (see Table 1-1 and Figure 1-1). The globular heads

predominantly contain P-sheets and P-turns which have prolines

and small neutral amino acids; in class III and IV IF proteins

the head also has an abundance of arginine (Geisler and Weber,

1982). The most dramatic feature of the high molecular weight

class IV neurofilament proteins, however, is their large and

apparently multi-domain carboxytails (Geisler et al., 1984,

1985a and 1985c). The rod domains of NF-L, NF-M and NF-H are

assumed to form the central filament core. The NF-M and NF-H

tail extensions are thought to account for the thin fibrous

protrusions visualized by ultrathin section electron


microscopy (Willard and Simon, 1981; Sharp et al., 1982;

Hirokawa et al., 1984; Hisanaga and Hirokawa, 1988; Mulligan

et al., 1991). These carboxy-tail regions are proposed to

contain the domains in which neurofilament-specific functions

would reside; one such domain which contains repetitive

sequences rich in lysine, serine and proline (KSP) is known to

be the predominant site of phosphorylation in these molecules

(Shaw, 1991 for review).

Neurofilament Assembly

Although the details of intermediate filament assembly

are still controversial, the consensus is that a-helical

coiled-coil dimers associate laterally to form tetramers and

octomers. Questions of polarity and alignment, as well as

which oligomeric unit or units might serve as a protofilament

structure capable of incorporation into a filament, are,

however, still topics of debate (Traub et al., 1992; Hisanaga

and Hirokawa, 1990; Hisanaga et al., 1990b; Tokutake, 1990b;

Steinert and Roop, 1988; Geisler et al., 1985b). Attempts to

dissect and recreate the assembly process of neurofilaments

(and other intermediate filaments) have made use of in vitro

reconstitution techniques. Subunit proteins are typically

purified by DE-52 column chromatography in the presence of 6M

urea (Tokutake, 1984) and then dialyzed to remove the

denaturants. When renaturation is complete most 10nm filament

subunits self-polymerize to form filaments. From such


experiments it seems that pure NF-L, but not NF-M or NF-H

alone, is capable of homopolymerization into normal appearing

filaments; NF-M and NF-H can, however, co-assemble with NF-L

(Geisler and Weber, 1981; Liem and Hutchison, 1982; Zackroff

et al., 1982) and have been reported to homopolymerize into

shorter than normal, rough-surfaced filamentous structures

(Tokutake et al., 1984; Tokutake, 1990). Experiments

attempting to study assembly in cultured cells have utilized

DNA transfection techniques to place normal or mutagenized

subunits that are not usually found in the cell type being

used (Chin et al., 1991; Chin and Liem, 1989) These studies

also find that NF-M and NF-H are not capable of

homopolymerization, but can use a native type III IF network

for heteropolymerization. It has been suggested that the long

carboxytails, which contain multiple phosphorylation sites,

interfere with the homopolymerization of NF-M and NF-H (Liem

and Hutchison, 1982; Wong et al., 1984; Wong et al., 1990);

this premise implies that NF-M and NF-H can readily form

heteropolymers with NF-L because NF-L effectively dilutes out

the ratio of tail to rod domains. Alternatively, it has been

suggested that heteropolymerization could be an inherent

property of NF-M and NF-H, analogous to the pattern seen with

the pairing of acidic and basic keratins (Shaw, 1991). The

arginine-rich head regions appear to be important for

intermediate filament assembly (Nelson and Traub, 1983;

Kaufman et al., 1985; Geisler and Weber, 1988) and have been


shown to interfere with filament assembly if they are

mutagenized in transgenic mice (Gill et al., 1990; Wong and

Cleveland, 1990). Indeed phosphorylation at key sites

bordering the conserved rod domain is likely to be the

endogenous method for controlling disassembly of filaments

within a cell (Hisanaga et al., 1990a; Ignaki et al., 1987).

Peripherin and a-internexin are each apparently capable of

homopolymerization into 10 nm filaments (Parysek and Goldman,

1987; Chiu et al., 1989).

Neurofilament Involvement in Disease

Although no direct causal role has been firmly

established with a naturally occurring neurofilament defect,

various aberrations in neurofilament morphology,

phosphorylation-state and distribution have been associated

with a variety of neuropathologies. The most obvious

correlation between neurofilaments and neuropathology is the

phosphorylation of perikaryal neurofilaments in response to

axotomy (Drager and Hofbauer, 1984) or other forms of

neurodegeneration (Sternberger et al., 1985). When this

phenomenon reverts in systems that can fully regenerate, such

as that following the initial stages of regeneration, the

perikaryal neurofilaments are no longer phosphorylated (Moss

and Lewkowicz, 1983; Shaw et al., 1988). Neurofilamentous

accumulations have been associated with intraneuronal

inclusions found in Alzheimer's disease, progressive


supranuclear palsy (PSP), Pick's disease, Parkinson's disease,

Guamanian amyotrophic lateral sclerosis (ALS) and giant axonal

neuropathy (GAN) (Gambetti et al., 1983; Goldman et al., 1983;

Matsumoto et al., 1990; Asbury et al., 1972; for review see

Goldman and Yen, 1986). Although it is still debatable,

early immunological data suggested that neurofilament epitopes

were present in the paired helical filaments (PHF), a major

component of the neurofibrillary tangles (NFTs) from

Alzheimer's disease. Many studies now indicate that this

signal may have been largely due to cross-reactivity of the NF

KSP region with a similar epitope of the microtubule

associated protein (MAP), tau (Miller et al., 1986; Ksiezak-

Reding et al., 1987; Schmidt et al., 1990), which has been

shown by biochemical methods to be a PHF component (Goedert et

al., 1988).

Putative Functions

Despite the fact that neurofilaments are one of the most

prominent features of the neuronal protein profile (at least

10% of total protein in the central nervous system), their

function is still elusive. Arguments have been made for

mechanical roles in providing strength and support, as well as

controlling axon diameter (Hoffman et al., 1984, 1987).

Phosphorylation effects on the carboxytails have been

implicated as a potential regulatory mechanism to alter the

spacing between filaments via the sidearm projections

(Hisanaga and Hirokawa, 1989). A recently described mutant

strain of Japanese quail (quiverer) which apparently lacks

neurofilaments shows a significantly reduced axon diameter

(Yamasaki et al., 1991). Transgenic mice which overexpress NF-

L, however, do not produce axons of increased caliber

(Monteiro et al., 1990). The clarification of this

relationship between neurofilaments and axonal structure will

require further investigations of both the mutant and

transgenic animal models.

Unlike the microfilamentous and microtubular networks

which have many dedicated associated proteins (such as their

respective myosin and kinesin motor protein families),

neurofilaments lack well-documented associated proteins.

Considering that most microfilamentous and microtubular

functions rely heavily upon interactions with their associated

proteins, it is also likely that our understanding of

neurofilament function will not grow considerably until we

isolate and study neurofilament associated proteins. Despite

the absence of obligate neurofilament associated proteins,

some interactions have been reported. Neurofilaments have

been shown to bind to microtubules via the microtubule-

associated proteins MAP2 and tau (Runge et al., 1981;

Leterrier et al., 1982; Minami et al., 1982; Minamai and

Sakai, 1983; Heimann et al., 1985; Miyata et al., 1986), as

well as to microfilaments via the actin-associated protein

fodrin/spectrin (Frappier et al., 1987). Steiner et al.


(1987) have shown that NF-L interacts with the synaptic

vesicle phosphoprotein synapsin 1. Although it has yet to be

demonstrated in neurons, vimentin has been shown to have the

potential to link the plasma membrane (via binding to the

actin-associated protein ankyrin) to the nuclear envelope (via

binding to lamin B) (Georgatos and Blobel, 1987, 1988); this

property is shared by peripherin (Djabali et al., 1991) as

well as the muscle specific intermediate filament desmin

(Georgatos et al., 1987). Neurofilament proteins have also

been described to contain calcium binding EF-hand-like

sequences (Lefebvre and Mushynski, 1988). Though functional

roles for these regions is not yet known, one might suppose

that they would be involved in a calcium-regulated assembly or

post-translational modification process acting either directly

on associated proteins or on neurofilament proteins

themselves. Some data appear to suggest that NFs can directly

bind nucleic acids (Traub et al., 1983 and 1985). Perhaps

neurofilaments may serve as general cytoplasmic carrier or

reaction surfaces serving to integrate other cytoskeletal

networks with the translational machinery and organelles.

Experimental Goals

Despite the prevalence of neurofilament proteins as well

as antibodies raised against them, much confusion still exists

concerning the role of these proteins in normal as well as

disease states. Much of this confusion may be due to a lack


of systematic experimental dissection of the sequence motifs

in the unique carboxyterminal tails of NF-M and NF-H. The aim

of this study is to inspect these domains in an organized

manner using the techniques of molecular biology, bacterial

expression and immunochemistry (Harris et al., 1991).

Ultimately this project should not only resolve existing

discrepancies in the literature due to the use of non-epitope-

mapped antibodies, but also facilitate a more precise

examination of neurofilament protein interactions and

functions in future investigations.



Neurofilaments of the adult mammal are composed

predominantly of three proteins, usually referred to as NF-L,

NF-M and NF-H, though they may also contain a-internexin,

peripherin and vimentin (see Shaw, 1991 for recent review).

Despite the fact that the three major proteins have been

extensively characterized, we still do not understand many

basic features of neurofilament biology. For instance, we know

little about how neurofilaments interact with other neural

components, virtually nothing about how they are transported

in the processes of neurons, how they are organized into

bundles or why they become perturbed in a variety of disease

states. It is widely assumed that the effectorr" regions of

intermediate filaments are in the hypervariable C-terminal

tail regions (e.g., Traub, 1985; Steinert and Roop, 1988). We

have therefore carefully examined the amino-acid sequences of

the tail regions of NF-M and NF-H as a first step in

elucidating the function of neurofilament tails. Our analysis

has allowed us to suggest a nomenclature for the several

distinct types of sequence found in these tails (Shaw, 1989,

1991; see Figure 1-1), and we have now extended this analysis



to cover a-internexin, which clearly belongs to the same

protein family as NF-L, NF-M and NF-H (Fleigner et al., 1990).

In line with previous studies, we call the region

immediately C-terminal to a-helical coil domain "Tail A"

(Geisler et al., 1983). C-terminal to this region a-

internexin, NF-L, NF-M and NF-H each have a glutamic-acid-rich

segment which we call the E-segment (El in NF-M). This

represents the entire C-terminus in NF-L, but both NF-M and

NF-H follow this with short repeated peptides containing the

sequence lysine-serine-proline, which we call the KSP

segments. In rat NF-H there are nearly 60 KSP and related

peptides arranged in an almost unbroken sequence, whereas rat

NF-M has 2 at each of 2 positions (KSP1 and KSP2). In NF-M

these two KSP sequences are separated by a second glutamic-

acid-rich region, E2. The numbers of KSP sequences and the

sequence of surrounding amino acids are quite variable across

species boundaries; human NF-M, in contrast to rat, has only

one KSP at the KSP1 position and 12 at the KSP2 position

(Myers et al., 1987), and human NF-H also has a different

number and arrangement of KSP sequences when compared to rat

NF-H (Lees et al., 1988). The KSP segments are of particular

interest since, they are major in vivo phosphorylation sites on

neurofilaments (Geisler et al., 1987; Lee et al., 1988a). C-

terminal to the KSP2 region in NF-M are the SP sequences,

which are also in vivo phosphorylation sites, may be present

in variable number in different species and are clearly


related to the KSP sequences (Geisler et al., 1987; Shaw,

1989; Xu et al., 1989). Rat NF-M has only one of these

sequences, as do human and mouse, although pig and chicken

have multiple SP repeats. At the extreme C-terminus of NF-M is

a region rich in lysine and glutamic acid which we have called

the KE segment. The primary sequence of this region is highly

conserved across species boundaries and contains interesting

15 amino-acid repeated sequences (shown as black bars in

Figure 1-1; also see Shaw, 1989). We have previously focused

attention on the KE segment as potentially a functionally

significant region, a speculation strengthened by the finding

that the extreme C-terminus of the newly recognized member of

the neurofilament subunit protein family, a-internexin, is

highly homologous to the extreme C-terminal KE segment of NF-

M, and contains a 15 amino-acid peptide sequence very closely

related to the NF-M KE repeats (Fleigner et al., 1990).

Finally, the analogous region at the extreme C-terminus of NF-

H is rich in lysine, glutamic acid and proline, and we have

therefore named it the KEP segment.

It seems obvious that these distinct types of sequence

must each have different functions. A possible method to find

out more about the functions of these regions is to obtain

neurofilament protein fragments containing only one of these

different types of sequence, although this has not been easy

to achieve using standard biochemical methodologies. Recently

it has become possible to generate recombinant fusion proteins


containing defined parts of the sequences of larger proteins,

a powerful method to produce desired regions of molecules in

bulk. We report here on the production of a preliminary panel

of fusion proteins containing NF-M and NF-H tail sequences.


Construction of Fusion Proteins

The fusion proteins are named systematically. In each

case the first letter indicates species, R for rat and C for

chicken. The second letter indicates the particular

neurofilament protein (H for NF-H and M for NF-M), and the two

numbers describe the first and last amino acid of the

neurofilament sequence included in the fusion protein. Rat NF-

M fusion proteins were constructed from the full-length rat

cDNA obtained from Dr. Ron Liem, and we use the amino-acid

numbering system in the publication describing this clone

(Napolitano et al., 1987). Rat NF-H fusion proteins were

constructed from the partial cDNA isolated by Dr. Ivan

Lieberburg and coworkers (Lieberburg et al., 1989). The

sequence of a full-length rat NF-H cDNA has recently been

described (Chin and Liem, 1990), and we use the numbering

system in this paper. The first amino acid of the Lieberburg

clone aligns with amino acid 559 of the full-length Chin and

Liem sequence. We also obtained a partial cDNA for chicken NF-

M described by Zopf et al. (1987). The numbering of this clone

is as defined in the full-length sequence later published by

the same group (Zopf et al., 1990). For fusion protein

expression we used various members of the pATH plasmid family

(Koerner et al., 1991). These prokaryotic vectors produce the

first 324 amino acids of the E. coli enzyme trp-E, a short

amino-acid sequence defined by a nucleic acid polylinker and

the protein coded for by the cDNA inserted. If the insert does

not contain a stop codon a short amino-acid sequence derived

from the vector follows the insert sequence, but this sequence

is never more than a few amino acids since pATH vectors

contains stop codons in all three reading frames immediately

following the polylinker. The neurofilament regions included

in the rat-derived fusion proteins are shown diagrammatically

in Figure 2-1.

RM:677-845. The rat NF-M cDNA was cleaved with Sau3AI,

and the 511 base pair fragment corresponding to the KE segment

was identified and ligated into BamHI cut pATH 1. The protein

coded by this segment starts a few amino acids after the

sequence DKKKAESP (the rat SP segment) and includes the entire

remaining carboxyterminus of the molecule, a total of 169

amino acids, corresponding to what we have defined as the KE

segment. This clone was used to generate the two shorter

fusion proteins RM:677-761 and RM:677-732 by chemical cleavage

as described below.

RM:549-845. The full-length rat NF-M cDNA was cut with

EcoRI and XhoI to produce a 1064 base pair fragment with an

XhoI site on the 5' and an EcoRI site on the 3' end. The

Chymotryptic tail


Coll I 0ol2 Ta I El KSPI E SP2 SP and KB I am
U RM.77-732


Chymotryptic tail

RH:559 -1072 ,

RH:559 -794 RH:84-1072
m-------- I I ---m--
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiini ThT hu4

col cad 2 Ta a

H and IKP *emtft

Figure 2-1. Map of rat NF-M and NF-H fusion proteins and
proteolytic fragments.




KWP sondi


protein coded for by this insert includes the KE, the SP, KSP2

and part of the E2 sequences. This was ligated into pATH23

that had been digested with SalI and EcoRI.

CM:381-605. This was derived from the partial cDNA of

Zopf et al., (1987). A 673 base pair fragment was cut out of

this clone with Sau3AI corresponding to bases 94 to 766. This

codes for the last segment of the a-helical rod domain, the

"tail A" region, the chicken El and E2 segments and the 4 NF-M

KSP sequences at KSP2, accounting for 225 amino acids (the NF-

M KSP1 segment is less evolutionarily conserved than the NF-M

KSP2 segment and is missing from chicken). The clone ends

between the KSP sequences and before the start of the multiply

repeated chicken SP sequences. The 673 base pair cDNA fragment

was gel purified and ligated into BamHI treated pATH 2.

RH:559-1072. We isolated three individual 2.2 kilobase

cDNA clones coding for rat NF-H from a commercial

bacteriophage lambda gt-ll rat brain cDNA library (Clontech)

by screening nitrocellulose filter replicas with a mixture of

polyclonal antibodies directed against all three neurofilament

triplet proteins. The 9-galactosidase fusion proteins

expressed by these clones reacted very strongly with N52, a

phosphate independent NF-H monoclonal antibody. Further

analysis showed that two of these cDNAs were, somewhat

surprisingly, absolutely identical in size, restriction digest

pattern and initial 5' and 3' sequence to the rat 2168 base-

pair NF-H clone described by Lieberburg et al. (1989). There


is no EcoRI site within the full-length rat cDNA clone, and we

did not use the same cDNA library as the Lieberburg group.

Presumably this finding reflects either specific nucleic acid

cleavage or blockage of polymerase during or possibly prior to

library construction. We decided not to use our own clones

since this would have required costly and time-consuming full

nucleic acid sequencing. Furthermore, a pATH construct

containing the full-length Lieberburg clone had already been

produced and was kindly donated by Dr. Ron Liem (Braxton et

al., 1989). These authors excised the Lieberburg NF-H cDNA

using EcoRI and ligated it into EcoRI-digested pATH 1.

RH:846-1072. The Lieberburg rat NF-H cDNA was cleaved

with Sau3AI and EcoRI to produce a 1272 base pair fragment

which corresponds to the KEP segment of NF-H preceded by the

last four irregular NF-H KSP sequences of the KSP repeats.

This was ligated into pATH 23 following treatment of the

vector with BamHI and EcoRI.

RH:559-794. The first 704 b.p. of the rat NF-H cDNA of

Lieberburg et al. (1989) was excised using EcoRI and Sau3AI.

This corresponds to the most perfect tandem KSP repeats

towards the amino terminus of the rat NF-H tail, containing 37

repeats with the consensus AKSPAE. The appropriate cDNA was

ligated into EcoRI/BamHI cut pATH 20.

Isolation and Verification of Fusion Protein Clones

Transfected HB101 E. coli were grown on ampicillin

plates, and 15-20 colonies were selected for screening.

Cultures were grown overnight in 5ml of M9 media plus 50Lg/ml

ampicillin and 20gg/ml tryptophan. These cultures were diluted

1:100 in fresh M9 media with no added tryptophan but with

ampicillin as before. Twenty gg/ml indoleacrylic acid (IAA)

was added to induce fusion protein expression. After overnight

growth at 370C cells were pelleted and dissolved in SDS-PAGE

sample buffer and run out on 7.5% gels. Clones which produced

a large extra, protein band at about the molecular weight

expected were selected for further study (see Figure 2-2A for

typical example). Immunoblots from such clones were tested

with polyclonal and monoclonal antibodies against the

appropriate protein, and those producing convincing positive

reactions at least with polyclonal antibodies were grown up to

prepare plasmid DNA. The correct insert size was then verified

electrophoretically following cleavage with an appropriate

restriction endonuclease. Alternately, insert size was rapidly

verified using the polymerase chain reaction (PCR) with probes

recognizing trp-E and polylinker sequences (i.e., on either

side of the insert; see Figure 2-2B). The 5' primer was

AGCCGCCAGATTGAGATC and the 3' primer was

TAATTCTCATGTTTGACAGCT; both primers will work on all current


members of the pATH family. We also ran out fractionated

fusion protein (see below) on SDS-PAGE and transferred to PVDF

for amino-acid analysis; in every case we obtained amino-acid

profiles extremely close to the expected values. Finally,

RM:677-845 and RM:549-845 were directly sequenced using the

United States Biochemicals Sequenase kit following the

manufacturers instructions for double stranded DNA. We used

the two oligonucleotide primers described above and found the

3' and 5' ends of the constructs to be exactly as predicted.

Fusion Protein Expression

Fusion protein expression was induced in 500ml cultures

by addition of 10-20p.g/ml IAA following withdrawal of

tryptophan (see Koerner et al., 1991). Cultures were grown for

up to 36 hours to obtain maximum fusion protein yield. Fusion

proteins were harvested by the lysozyme lysis inclusion body

procedure described in Ausubel et al. (1990). Inclusion body

material was dissolved in freshly deionized 6M urea, 10mM

sodium phosphate, 1mM PMSF, and ImM EDTA pH=7.5 and subjected

to ion exchange chromatography on DEAE-cellulose as described

previously (Shaw and Hawkins, 1992). Proteins were eluted with

a NaCl gradient from O.OM to 0.25M and fractions were examined

by SDS-PAGE (see Figure 2-3). The cleanest fusion protein-

containing fractions were extensively dialyzed against

phosphate buffered saline containing ImM PMSF and were used

for further experiments and antibody production.

Chemical Cleavage of Fusion Proteins

Hydroxylamine under basic conditions cleaves polypeptide

chains at asparagine-glycine (NG) sequences (Bornstein and

Balian, 1977; Moks et al., 1987). There are two convenient NG

sequences in the KE segment of NF-M, although no other part of

the NF-M or NF-H tails contains these sequences. Appropriate

preparations of neurofilaments or fusion proteins were taken

up in hydroxylamine cleavage solution (2M hydroxylamine, 0.2M

Tris pH=9.00) and incubated at 370C for 1 hour. The reaction

was stopped by neutralization with 1M HC1, and cleaved

proteins were mixed with gel sample buffer and run out on SDS-

PAGE for immunoblotting.

The last 50 amino acids of the rat NF-H tail could be

clipped off using cyanogen bromide (CNBr). The RH:846-1072

fusion protein was dialyzed against PBS, concentrated and then

cleaved by CNBr by adding 190 Al fusion protein (approximately

0.5 mg/ml) and 190 mg CNBr (Sigma) to 760 pL of 88% formic

acid. This mixture was mixed in a sealed tube and left

overnight at ambient temperature in the fume hood. The sample

was then aliquoted into 10 tubes and dried down in a vortex

evaporator at 600C for at least 6 hours. Aliquots were sealed

and stored at -200C until use.


We transfected bacteria with our nucleic acid constructs

and grew up minicultures of individual ampicillin-resistant

colonies under conditions designed to induce fusion protein

expression. After overnight growth bacteria were pelleted and

analyzed by SDS-PAGE. Figure 2-2A shows the result of

screening 11 colonies which had been transfected with the

plasmid designed to produce the RH:559-794 fusion protein.

Lanes 1-9 show colonies all of which produce a very prominent

protein band missing in lane 12, which was transfected with

vector lacking a neurofilament cDNA insert. Lanes 10 and 11

show bacterial colonies which failed to express the desired

insert. We verified the identity of these clones using several

methods. All of the fusion proteins could be stained with

appropriate polyclonal antibodies as well as certain

monoclonal antibodies to neurofilaments. We checked the size

of the cDNA insert using restriction mapping or the polymerase

chain reaction (Figure 2-2B). The inserts were of the expected

size in every case. Fusion proteins were purified from

bacterial inclusion body preparations following dissolving in

phosphate-buffered 6M urea. Ion exchange chromatography on

DEAE-cellulose proved to be an efficient method to purify the

fusion proteins (Figure 2-3).


S 1 2 3 4 5 6 7 8 9 10111212345 S

Figure 2-2. Construct screening methods. A: Coomassie
Brilliant Blue stained SDS-PA gel showing a screening for
fusion protein expression by RH:559-794 clones. Lane S is a
standard containing GFAP (G, 50 kDa). Lanes 1-9 contain a
major band (FP) corresponding to the fusion protein which is
absent in the control (bacteria transfected with empty pATH
vector) in lane 12. Lanes 10 and 11 are Ampicillin-resistant
colonies that did not incorporate the correct insert.
B: Ethidium bromide stained agarose gel of PCR products
derived from RM:677-845, RH:846-1072, RH:559-794, RM:549-845
and CM:381-605 (lanes 1-5, respectively). The three arrowed
standards (lane S) are, bottom to top, 506, 1018 and 1636 bp.
Calculated PCR product sizes are 618, 1367, 788, 1158 and 762
bp respectively, in excellent agreement with the gel.

FP' mim -4


12 3 4 5 6 7 8 910

Figure 2-3. Coomassie Brilliant Blue stained SDS-PA gel
showing a DEAE-cellulose purification of RH:846-1072. Lane S
is a standard containing bovine GFAP (G, 50 kDa) and NF-H (H,
220kDa). Lanes 1-10 are sequential fractions produced by the
0.0 M to 0.25 M NaCI gradient. The fusion protein (FP) is
eluted cleanly before most of the bacterial contaminants.




Asx Thr
Ph Pro

Lu / \ Ala
le Met Val

RH:550-794 RH,846-1072



Figure 2-4. Plots depicting predicted versus measured amino
acid compositions for trpE alone, RM:677-845, RH:559-794 and
RH:846-1072. A reference wheel shows the direction in which
the relative percentage of each amino acid is plotted.




The fusion proteins were transferred to PVDF membranes

and processed for amino-acid analysis and in every case gave

profiles very close to that predicted (Figure 2-4). We have

also verified the nucleic acid sequence of RM:677-845 by

direct nucleic acid sequencing and found it to be absolutely

as expected. In many cases we obtained multiple clones, as in

the case shown in Figure 2-2A, which produced identical

protein profiles.

Table 2-1. Real and Apparent SDS-PAGE Molecular Weights of
Fusion Proteins:

Fusion # Amino Calculated MW SDS-PAGE MW
protein Acids (kDa) (kDa)
trpE 336 37.3 37
RM:677-845 503 55.7 68
RM:677-761 419 46.5 55
RM:677-732 390 43.3 48.5
RM:549-845 633 70.5 82.2
CM:381-605 550 60.5 75
RH:559-1072 833 91.5 133
RH:559-794 566 60.3 71.3
RH:846-1072 567 63 76.2
TD-RH:846-1072 411 45 59
TDCNBr-RH:846-1022 216 24 50

We measured the SDS-PAGE molecular weights of the fusion

proteins as well as the proteolytic fragments of RM:677-845

and RH:846-1072 and found that, although trp-E alone ran very

close to predicted from its molecular weight, in nearly every

case the fusion protein ran more slowly (see Table 2-1). We

conclude that all parts of the NF-M and NF-H tails exhibit

this unusual SDS-PAGE mobility. We calculated the contribution

to SDS-PAGE mobility due to each neurofilament insert and


compared that to the known molecular weight. In every case the

insert ran between 1.4 and 1.8 times slower than expected from

its molecular weight.


Six cDNA constructs containing overlapping but distinct

tail domains for NF-M and NF-H were successfully cloned and

expressed in bacteria. All constructs were verified by PCR

amplification of a cDNA insert of an appropriate size. Some

cDNA constructs were at least partially sequenced to verify

the ligation sites. Fusion proteins were checked for

appropriate molecular weights, amino-acid compositions and

positive immunostaining on western blots using existing

neurofilament polyclonal antibodies. Finally, since all

fusion proteins exhibit the unusual electrophoretic properties

of the parent molecules (discussed below) and since all of

them, when injected into rabbits and mice, generate

neurofilament antibodies (see Chapter 3), there can be little

doubt about the correctness of the constructs.

Rat NF-H and NF-M run at about 200kDa and 145kDa,

although the real molecular weights are 115.3kDa and 95.6kDa

as determined from the full amino-acid sequence (Chin and

Liem, 1990; Napolitano et al., 1987). Previous studies have

shown that the tail sequences of NF-M and NF-H are responsible

for this unusual gel mobility (Kaufmann et al., 1984). Here we

show that this unusual property is distributed throughout the

entire tail segment. The common features of the NF-M and NF-H

tail sequences are the tail A sequences, the KSP repeats, the

glutamic-acid-rich E regions and the overall high charge

density. Our results suggest that the high charge density is

much more important for slowing down these proteins in SDS-

PAGE, since all neurofilament tail fusion proteins,

irrespective of their content, run unusually slowly on SDS-

PAGE. We also note that the chymotryptic tails seem to run

relatively even slower on SDS-PAGE than any of our fusion

proteins, suggesting that either the intact tail somehow

interacts with the gel matrix in a manner different from the

fusion proteins, or that phosphate groups which are resistant

to enzymatic dephosphorylation also contribute to SDS-PAGE gel

retardation in the native molecules.

The recombinant fusion proteins described here can be

used not only to epitope-map existing neurofilament

antibodies, but also to generate antibodies against targeted

domains which may be underrepresented by existing markers.

These constructs are also of great potential use in

biochemical and cell biology studies of neurofilament

dynamics, phosphorylation and function.



In spite of there being a considerable number of NF-M and

NF-H antibodies available both commercially (e.g. Sigma and

Boeringher Mannheim) and from large laboratories doing

intermediate filament research (e.g. Weber and Osborn, Lee

and Trojanowski), nearly none of these antibodies have been

previously characterized beyond their ability to stain the

rod, chymotryptic tail or phosphorylated KSP repeats. The

majority of neurofilament antibodies react with the highly

charged, phosphorylated KSP epitopes, which are shared with

tau and MAP2. Cross-reactivity between antibodies made

against high molecular weight NFs (NF-M and NF-H) and MAPs

(tau and MAP2) has caused considerable confusion in

immunochemical studies on degenerating neurons, like those of

Alzheimer's disease. In addition, the chymotryptic tails of

NF-M and NF-H have been difficult to further dissect protein

chemically because of the large number of repeated elements

and lack of convenient cleavage sites. With the panel of

fusion proteins and fragments we have generated, our

laboratory was able to refine the epitope maps of existing


NF-M and NF-H monoclonal antibodies. After screening over 300

existing monoclonals, we have been able to further map 22 of

them; the majority of these antibodies (the RMO series) were

made by Dr. Virginia Lee and co-workers. By using fusion

proteins containing regions other than the KSP repeats, we

have also been able to make new antibodies that recognize

unique epitopes in both NF-M and NF-H.


Available NF Antibodies

NR4, NN18, NE14 and N52 are commercially available from

several vendors (including Boerhinger-Mannheim and Sigma).

These and the other N series antibodies have been partially

characterized in a series of previous publications (Debus et

al.; 1983, Shaw et al., 1984, 1986). The RMO series (#1-310)

of NF-M and NF-H monoclonals were made available by Dr.

Virginia Lee. SMI31, SMI32, SMI33 and SMI34 can be obtained

from Sternberger-Meyer Immunocytochemicals. RT97 and 8D8 were

the kind gift of Dr. Brian Anderton. The a-IFA clone was

obtained from the American Type Culture Collection and used to

produce antibody rich hybridoma supernatents (Pruss et al.,

1981). The polyclonal antibodies (H301, H298, GP64 and GP63)

were made against purified rat neurofilament subunits isolated

from sciatic nerve (Shaw et al., 1981). Neurofilament

antibodies were used at empirically determined concentrations

which were the minimum required to give saturated staining of

NFs in frozen sections of rat brain.

Western Blotting

Fusion proteins purified by DEAE-cellulose chromatography

and their proteolytic fragments were run out on either 4-20%

gradient (BioRad) or constant percentage (6% or 10%) combless

SDS polyacrylamide gels and transferred to nitrocellulose

filters using standard methodologies. Filters were cut into

strips and probed with various antibodies. Second antibodies

were appropriate goat anti-rabbit, goat anti-guinea pig and

goat anti-mouse antisera coupled to alkaline phosphatase.

Colored reaction product was generated using nitroblue

tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (second

antibodies and chromagens obtained from Sigma).

Production of Fusion Protein Antibodies

Fusion proteins purified as described above were injected

at about 200pg per animal into rabbits and mice using standard

methodologies. All animals injected so far have produced

strong anti-neurofilament antibodies after only two

injections. One rabbit was injected with each of the purified

fusion proteins RM:677-845, RH:559-794 and RH:846-1072. The

antisera resulting from these are referred to as anti-NF-M KE,

anti-NF-H KSP and anti-NF-H KEP respectively. Rabbit

antibodies were characterized by immunoblotting and on frozen


sections of rat brain and spinal cord. We have raised

monoclonal antibodies against the RM:677-845 and RH:846-1072

fusion proteins. Mice were injected following a similar

protocol to the rabbits. Hybridomas were screened on ELISA

using bovine NF-M or rat NF-H, and positive clones were

retested firstly by immunofluorescence on frozen sections of

rat brain and then by immunoblotting on the appropriate fusion

protein and porcine neurofilament protein. Clones of interest

were subcloned twice by limiting dilutions in 96 well dishes

on a monolayer of 3T3 cells.


We used purified fusion proteins to screen our large

panel of neurofilament antibodies (see Figure 3-1 for typical

results). We have previously shown that all of the N series

antibodies stain only the chymotryptic tails of NF-M and NF-H,

and that most of these antibodies show reduced or no staining

following enzymatic dephosphorylation of neurofilament

proteins (Shaw et al., 1986). The antibodies which do not

recognize enzymatically dephosphorylated neurofilaments fail

to stain our fusion proteins. NE14, NF1, RT97, SMI31 and SMI34

fall into this family of antibodies, in line with previous

suggestions that these antibodies recognize predominantly

phosphorylated KSP sequences (Lee et al., 1988a). Three of the

antibodies tested (SMI32, SMI33 and N52) showed very strong

staining of the RH:559-1072 and RH:559-794 clones both of


a b c d e fghij k 1mn o

pqrs t u

Figure 3-1. Selected results with RH:559-794, RH:846-1072 and
RM:677-845. Lanes a-o are blots of RH:559-794 (left arrow).
Lane a=H298 (rabbit anti-NF-M); b=H301 (rabbit anti-NF-H);
c=NN18; d=SMI34; e=SMI33; f=SMI32; g=SMI31; h=8D8; i=RT97;
j=NE14; k=NF1; 1=NC43; m=NL34; n=NA34; o=N52. Lanes p-u are
blots of a mixture of purified RH:846-1072 (upper arrow) and
RM:677-845 (lower arrow). Lane p=GP63 (guinea pig anti-NF-M);
q=GP64 (guinea pig anti-NF-H); r=H298; s=H301; t=NN18; u=N52.

P- -Y !



-- "


Figure 3-2. Preliminary cross-species characterization of
rabbit polyclonal anti-rat NF-H KEP, NF-H KSP and NF-M KE.
Lanes 1-3 are Coomassie Brilliant Blue stained preparation of
neurofilament from pig, cow and chicken respectively. Arrows
indicate the positions of porcine NF-(H), NF-(M) and NF-(L).
Lanes 4-6 are corresponding immunoblots reacted with anti-NF-H
KEP, anti-NF-H KSP and anti-NF-M KE. respectively.


which contain multiple KSP repeats (Figure 3-1, lanes e, f and

o). These three antibodies, however, gave divergent results on

the other fusion proteins. N52 failed to stain either of the

fusion proteins containing NF-M-type KSP sequences (RM:549-845

and CM:381-605), suggesting that it is specific for some

feature of the NF-H KSP motif. N52 showed weak, but definitely

positive, staining of RH:846-1072, which contains the last

four irregular rat NF-H KSP sequences (Figure 3-1, lane u).

N52 therefore does not absolutely require either large numbers

or highly regular NF-H type KSP repeats. SMI32 showed weak

staining of RM:549-845, which contains the rat NF-M KSP2

sequence, did not stain the chicken fusion protein and gave

very weak and equivocal staining of RH:846-1072. We conclude

that SMI32 is -somewhat less specific for NF-H KSP sequences

than N52, but that it may have more rigid requirements in

terms of the length or organization of those sequences. SMI33

gave the simplest pattern; it stained all fusion proteins

which contain typical neurofilament KSP sequences irrespective

of whether they were in NF-M or NF-H, with one exception.

RM:677-845 actually contains one KSP sequence (GDKSPQE, 717-

723). However this is not evolutionarily conserved, is not

preceded by neutral or hydrophobic amino acids and is

apparently not an in vivo phosphorylation site (Xu et al.,

1989). It seems that this tripeptide arose fortuitously and is

probably not related to the other KSP sequences. NN18, a

phosphorylation-independent monoclonal antibody reactive with


the chymotryptic tail of NF-M stains the RM:677-845 and

RM:549-845 fusion proteins strongly and specifically,

localizing the NN18 epitope to somewhere with in the NF-M KE

segment (Figure 3-1, lane t, for higher resolution mapping see

below). Anti-IFA, a monoclonal antibody which recognizes a

determinant in the a-helical rod region of all intermediate

filaments (Pruss et al., 1981) stains only the CM:381-605

fusion protein. This finding is completely in agreement with

the results of Geisler et al. (1983), who localized the anti-

IFA epitope to the extreme C-terminus of the a-helical rod

region, which is included only in CM:381-605. Finally we

previously described NL34, NA34 and NM46 as antibodies which

recognize NF-H and/or NF-M tails, but which are partially or

totally phosphate-independent (Shaw et al., 1986). Two of them

failed to stain any of the fusion proteins and the third,

NA34, showed weak staining for the fusion proteins containing

NF-H KSP sequences (e.g. Figure 3-1, lane n).

We have raised antibodies against the rat fusion proteins

RM:677-845, RH:559-794 and RH:846-1072. We refer to these

antibodies as anti-NF-M KE, anti-NF-H KSP and anti-NF-H KEP

respectively. All three antisera produce high-titre

neurofilament staining on frozen sections of rat brain and

strongly and specifically recognize the appropriate immunogen

and intact rat neurofilament subunit on immunoblots (not

shown). We examined the ability of these antibodies to

recognize neurofilament subunits in cow, pig and chicken.


Anti-NF-H KEP recognized pig and cow NF-H strongly on

immunoblots, but gave a very weak signal on chicken NF-H

(Figure 3-2, lane 4). Anti-NF-H KSP proved to stain all

mammalian NF-H bands very strongly, but also showed weaker

staining for mammalian NF-M, except in the case of bovine NF-

M, which was stained much more strongly than rat or pig

(Figure 3-2, lane 5). Anti-NF-H KSP weakly recognized both

chicken NF-H and chicken NF-M. In contrast the anti-NF-M KE

strongly recognized NF-M in all species tried and showed no

reactivity for'NF-H (Figure 3-2, lane 6). All three antibodies

recognized axonal neurofilaments in frozen sections of rat

brain very strongly and specifically. In line with the

immunoblot results all three antibodies gave clearly

neurofilamentous staining on frozen sections of chicken brain,

but anti-NF-M KE stained very strongly in contrast to anti-NF-

H KSP and anti-NF-H KEP which were much weaker (not shown).

We have also raised monoclonal antibodies against the NF-

M KE (RM:677-845) and NF-H KEP (RH:846-1072) segments. We

injected mice and made hybridomas using standard procedures,

and screened supernatents on either native pig NF-M, rat NF-H

or the fusion proteins. Many hybridomas recognized native NFs

in ELISA assays, and from these we subcloned eight NF-M KE and

2 NF-H KEP monoclonals which stained neurofilaments strongly

in frozen sections of rat brain and also recognized only the

appropriate native NF subunit strongly on immunoblots. We also

isolated several hybridomas which stain the trp-E carrier


protein, one of which, called 4A10, we subcloned and partially

characterized. In order to further localize the epitopes for

these antibodies, we cleaved the RM:677-845 fusion protein

using hydroxylamine, a reagent that cuts between asparagine-

glycine (NG) residues under basic conditions (Bornstein and

Balian, 1977), and the RH:846-1072 fusion protein with

cyanogen bromide.

The extreme C-terminus of rat NF-M contains two conserved

NG sequences, at 732-733 and 761-762. Complete cleavage at

these sites should therefore release an 84 amino acid C-

terminal fragment and a 29 amino acid fragment immediately

preterminal to this. Both of these sites are within the

RM:677-845 clone and complete NG cleavage should therefore

produce three fusion protein fragments which contain

neurofilament sequences corresponding to 677-732, 733-761 and

762-845. The N-terminal 677-732 fragment should remain

attached to the trp-E carrier protein. We performed the NG

cleavage as described above and ran the product on SDS-PAGE

for protein visualization (Figure 3-3, lanes U and C). We

noted three major protein bands in the SDS-PAGE molecular

weight range 50-70kDa, which made it clear that the cleavage

was only partial under our conditions. The highest band

comigrated with the RM:677-845 fusion protein and was

therefore presumably uncleaved protein. We assumed that the

lower bands represented RM:677-845 cleaved at one or other of

the two NG sites. To test this we transferred all three


polypeptides to PVDF membranes and obtained amino-acid

composition data. The profiles obtained (not shown) were

consistent with this interpretation, and we therefore refer to

the three bands in Figure 3-3 as RM:677-845, RM:677-761 and

RM:677-732. These three large protein bands could be

conveniently used for immunoblotting and epitope localization

experiments with our panel of monoclonal antibodies (the

released 29 and 84 amino acid C-terminal fragments are rather

small and much less easy to immunoblot under our standard

conditions). The results were quite clear; NN18, RMO1 and

RM059, stain all three bands, two of our new monoclonals stain

the two larger bands and the remaining 6 stain only the

largest band. In addition many of the RMO monoclonal

antibodies, including the better described RM0255 and RM054,

also only stain the largest band. Dr. Lee has localized the

epitope for RM0255 to the last 20 amino acids of NF-M using a

peptide containing those last 20 amino acids in a competitive

ELISA assay (personal communication). Finally, and as

expected, our trp-E antibody stains all three bands. These

results firmly map the epitopes for the 11 antibodies as shown

in Figure 3-5. We always saw staining for a fourth band just

below RM:677-732 with an apparent molecular weight of about

44kDa (labelled T in Figure 3-3), which is a minor component

biochemically. We conclude that this protein is produced by

cleavage within the trp-E molecule, by removal of about 200

amino acids. In line with this interpretation the preterminal

,-~- RM:677-845 0 II immm INOM

RM:677-7614 n *N
RM:677-732 *U

U C 1 2 3 4 5 6 7 8 9 10 11 12

Figure 3-3. Production and use of hydroxylamine cleaved
RM:677-845 to map antibodies to NF-M C-terminal. Lanes U and
C show Coomassie Brilliant Blue stained gels of uncleaved (U)
and cleaved (C) RM:677-845 fusion protein isolated by ion
exchange chromatography in 6M urea as described. Note partial
cleavage, so that a band of apparent molecular weight 66kDa is
seen in both lanes. Note also the presence of two lower
molecular weight fragments in C which run at 55kDa and
48.5kDa. Lanes 1 to 12 are immunoblots from 7.5% acrylamide
curtain gels of preparations identical to lane C. The
antibodies are 1=NN18, 2=5C6, 3=IG12, 4=5G9, 5=IG9, 6=4H4,
7=5E7, 8=5B12, 9=RM054, 10=RM0255, 11=3H11 and 12=4A10. The
position of the three fusion proteins is as indicated. T
denotes a fragment produced by N-terminal cleavage of trp-E.



Figure 3-4. Immunoblots of cyanogen bromide (CNBr) cleaved
RH:846-1072 to map NF-H KEP monoclonals 3G3 and 5B8. Lane A is
stained with 3G3 and lane B with 5B8. Arrowheads mark the
positions of the uncleaved fusion protein RH:846-1072 (top),
a degradation product of RH:846-1072 missing part of the trp-E
head, TD-RH:846-1072 (middle), and the CNBr cleavage fragment
of TD-RH:846-1022, TDCNBr-RH:846-1022 (bottom). No protein
band was present in the top position in samples treated by

antibodies NN18, 5C6 and IG12, but none of the extreme C-

terminal antibodies stained a further protein band below T;

this is consistent with our conclusion since a fusion protein

missing part of the N-terminus should also generate a further

similar but lower apparent molecular weight degradation ladder

to that shown in Figure 3-3. Upon re-examination several of

our other fusion protein preparations contained a second minor

protein band running at about 22kDa lower than the major

component, also consistent with a cleavage event within the

trp-E molecule, probably generated enzymatically either in

vivo or during our inclusion-body preparation.

Using cyanogen bromide (which cleaves after methionine)

we were also able to produce a fragment of the NF-H KEP fusion

protein which is missing the last 50 amino acids (RH:846-

1022). Western blots of cyanogen bromide treated RH:846-1072

were stained with our two monoclonal KEP antibodies (Figure 3-

4). Both antibodies (3G3 and 5B8) stain the uncleaved fusion

protein (RH:846-1072) and its trp-E-degradation product (TD-

RH:846-1072), but only 3G3 stained the large fragment (TDCNBr-

RH:846-1022). -This pattern implies that the epitope of 5B8

may be in the last 50 amino acids, whereas that of 3G3 is

likely to be within the preterminal KEP region. It is of note

that out of all the antibodies we screened, these two

monoclonals are the only antibodies (other than the rabbit

polyclonal anti-NF-H KEP) whose epitopes lie solely within

the KEP domain of NF-H. Although they contain a degradation


product (TD-RH:846-1072, indicated by the middle arrowhead on

Figure 3-4) missing part of the the trp-E head, samples

treated with CNBr contain no detectable uncleaved RH:846-1072

(indicated by the position of the top arrowhead in Figure 3-

4); the lack of any uncleaved material in the CNBr treated

samples is likely due to the presence of 6 methionine cleavage

sites in the trp-E head domain.

We tested all newly produced monoclonal antibodies on

several of the fusion proteins as was done for the antibodies

we already had available. As expected, all of the KE

antibodies recognized the RM:549-845, which includes the

sequences used as immunogen. None of the KE monoclonals showed

the slightest reactivity with the RH:846-1072 fusion protein,

which contains the NF-H KEP segment, which can be considered

as at least analogous to the NF-M KE segment. Similarly, the

KEP monoclonals 3G3 and 5B8 do not stain any of the fusion

proteins that do not contain the NF-H KEP segment. We could

not do the same experiment in such a straightforward manner

with our new polyclonal antibodies, since they all have some

recognition for trp-E in addition to the neurofilament

sequences. We therefore tried various older polyclonal

antisera from animals which were injected with native NF-M or

NF-H, but which nonetheless produced antibodies which cross-

reacted with the two proteins (Shaw and Weber, 1981). Despite

this cross-reactivity the anti-NF-H crude sera were completely

specific for the NF-H fusion proteins and the anti-NF-M crude

Chvmotrvtic tail

R I M.77-845

NF-M H IN _----
OoO 1 OCR 2 Too a Bl KSP1 B2 KSP2 SP ad Kameals


FA SMI-33 NN18 5 9 R12
RMO1 I32 512 i 12
RMO 3H11 19
5E7 5'o 4

Figure 3-5. Epitope maps of NF-M and NF-H tails. Only
antibodies which show strong and unequivocal staining are
shown. The epitope of monoclonal IFA has been previously
mapped (Geisler et al., 1983).

Chymotryptic tail

RKr550 -1072

4 2 RH:559 -794

Caa zi l' N and KS .amiu


RMO 39
RMO 81
RMO 195
RMO 224

RH 48-1072

3 1


Figure 3-5--continued.



sera were completely specific for the NF-M fusion proteins

(Figure 3-1, lanes q-s). Clearly the cross-reactivity

exhibited by these antibodies was not due to any immunological

similarity between the extreme C-termini of NF-M and NF-H.

These NF-M and NF-H polyclonal antibodies showed some degree

of recognition for fusion proteins contain KSP sequences,

suggesting that KSP sequences are at least partially

responsible for the cross-reactivity of these antisera (e.g.,

Figure 3-1, lanes a,b).


Previous data have shown that the majority of

currently available neurofilament antibodies recognize

phosphorylated KSP sequences in the tails of NF-M and NF-H

(e.g. Garden et al., 1985; Shaw et al., 1986; Lee et al.,

1988a; Coleman and Anderton, 1990). The immunodominance of the

KSP region has caused some confusion; many proteins apart from

neurofilaments contain the KSP motif, notably the microtubule

associated proteins tau and MAP2, and immunological cross-

reactivity between these proteins and NF-M and NF-H has been

well documented (Kziesak-Reding et al., 1987; Nukina et al.,

1987; Lee et al., 1988b). Clearly antibodies which are known

to recognize non-KSP parts of NF-M and NF-H would be very

useful for a variety of studies. We now have a growing panel

of such antibodies, allowing a more refined interpretation of


many previous studies and the design of better experiments in


The detailed characterization of the commercially

available NF-H monoclonal antibody N52 has not been previously

described, although we have previously shown that it

recognizes the native chymotryptic tail of axonal NF-H, and

that its binding does not appear to be affected by enzymatic

dephosphorylation (Shaw et al., 1986). We now show that N52

belongs to the large class of antibodies that recognize the

regions of NF-H which contain Lys-Ser-Pro (KSP) repeated

sequences. Given that the antibody recognizes the KSP

segments, it is perhaps rather surprising that binding is not

affected by the level of NF-H phosphorylation. This antibody

also weakly recognizes our RH:846-1072 fusion protein which

contains the last four disorganized KSP sequences, suggesting

that N52 is relatively insensitive to the exact primary

sequence of the NF-H KSP repeats. However N52 does not

recognize the NF-M KSP motifs in either of the NF-M clones

which contain these sequences, showing that some feature of

the NF-H repeats is important for efficient binding. In

contrast SMI33, a widely used phosphorylation independent

monoclonal antibody, stains all fusion proteins which contain

typical KSP sequences, although the recognition of RH:846-1072

was extremely weak, so that staining is not visible on Figure

3-1 (lane g). RM:549-845 contains only two KSP sequences, and

CM:381-605 contains four KSP tripeptides within the sequence

VKSPPAKSPPKSPPKSPV, which is somewhat divergent from the

mammalian norm. Clearly SMI33 is a useful probe capable of

detecting as few as two neurofilament KSP repeats and is

relatively insensitive to the exact organization of these

sequences. SMI32, a monoclonal antibody which selectively

recognizes the dephosphorylated forms of NF-M and NF-H

(Kziezak-Reding et al., 1987; Nukina et al., 1987), recognizes

the RH:559-1072 and RH:559-794 clones, but not the RH:846-1072

clone, suggesting that this antibody either requires more KSP

sequences or some other feature of the highly-regular KSP

repeats found only towards the N-terminus of the NF-H KSP

segments. This antibody weakly recognizes RM:549-845, which

contains the NF-M KSP2 sequences (Figure 3-5), and does not

stain the chicken NF-M clone. This antibody therefore appears

to be directed primarily against longer or more organized NF-H

KSP sequences, but has a limited ability to bind to NF-M KSP

sequences. This data provides the first detailed

characterization of N52, and confirms and extends our

understanding of the binding requirements of SMI32 and SMI33.

Our results localize the epitopes for RMO1, RM059 and

NN18, a widely used NF-M monoclonal antibody, to the extreme

carboxyterminus of NF-M between amino acids 677 and 732. As

noted above, we have previously found that certain neurons

stain poorly or not at all with NN18 although they stain well

with other NF-M antibodies (Weber et al., 1983, Eaker et al.,

1990). We can now conclude that some modification of the


region between 677 and 732 is responsible. This could involve

unusual protein folding, post-translational modification or

the presence of a binding protein, and may be of functional

significance. The work described here is a necessary first

step towards understanding what this modification might be. It

will also be useful in the understanding of some previous

surprising results, for example the finding of an NN18

immunoreactive microtubule-associated protein in crayfish

neurons (Weaver and Viancour, personal communication) and a

strongly NN18 immunoreactive 58kDa protein (NF5) in the

myoplasm of developing ascidia (Swalla et al., 1991).

The production of the rabbit polyclonal anti-NF-H KEP and

the two KEP monoclonals 3G3 and 5B8 represents a significant

addition to the neurofilament antibody repertoire. These NF-H

specific antibodies will allow a better determination of NF-H

staining in tissue than the KSP reactive antibodies previously

available. These unique probes may also prove useful in

determining the possible effector role of this previously

indistinguishable domain.

Several of the antibodies used here have been shown to

recognize epitopes in the tails of NF-M and/or NF-H, but are

unable to recognize non-phosphorylated neurofilaments. This

family includes SMI31, SMI34, NE14, NF1, NL34, NC43, RT97 and

8D8. Their failure to recognize our fusion proteins is

therefore completely in line with these previous findings.

Some members of our previous panel of antibodies were either

unaffected or showed only a partially-reduced ability to

recognize enzymatically dephosphorylated neurofilaments (Shaw

et al., 1984, 1986). NL34 and NM46 belong to this group;

although they recognize chymotryptic tails of NF-M and NF-H,

they were both completely negative on the fusion proteins

tried here. It is possible that these antibodies recognize

epitopes containing phosphate groups which are very difficult

to remove enzymatically; it is known that enzymatic

dephosphorylation does not remove all phosphate groups from

neurofilaments (Georges et al., 1986). NA34 is an unusual

antibody which recognizes enzymatically dephosphorylated NF-M

and NF-H weakly in most species, but stains NF-M strongly in

pig (Shaw et al., 1986). NA34 stained RH:559-1072 and RH:559-

794, but only very weakly. We do not understand why this

antibody fails to stain the NF-M fusion proteins, but can at

least conclude that it has some capacity to stain NF-H KSP


Our polyclonal antibody against the NF-M KE segment

stains chicken NF-M strongly and specifically (Figure 3-2). In

contrast, the antibodies to the rat KEP and KSP segments

stained mammalian NF-H strongly, but recognize chicken NF-H

very weakly. We have previously shown that the KEP segment of

the NF-H tail is much less conserved in primary sequence than

the KE segment of NF-M and proposed that these two regions are

distinct enough in sequence to be structurally unrelated

(Shaw, 1989). Directly in line with these suggestions we now

show that the KEP segment of NF-H is immunologically much more

variable across species boundaries than the KE segment of NF-

M, and that the two regions show no immunological cross-

reactivity with any of the antibodies employed here. The

immunological conservation of the NF-M KE segment documented

here suggests structural conservation in line with our

proposal that this region subserves an important function

(Shaw, 1989). Finally we can speculate that chicken NF-H, for

which we currently have no sequence data, will be fairly

divergent from the mammalian prototype in the regions

homologous to the NF-H KSP and KEP segments.

We noted that the anti-NF-H KSP antibody showed strong

staining for cow, but not pig or rat NF-M. This result is in

line with an observation we made several years ago; all of our

phosphate dependent monoclonal antibodies specific for NF-H

also strongly recognize cow NF-M but not NF-M from other

mammals (Shaw et al., 1984). We made this odd finding in

several different individual cows, and other groups also have

noted this phenomenon (Lee et al., 1986). In the meantime we

have found that many of the phosphate-dependent antibodies to

NF-H recognize phosphorylated KSP sequences (unpublished). We

now show that antibody raised against mammalian NF-H KSP

repeats strongly stains bovine NF-M. However, polyclonal and

monoclonal antibodies to the KE segment all recognize bovine

NF-M, so that other regions of the bovine NF-M tail are

probably not unusual. Taken together a possible explanation


for these findings is that the KSP repeats of bovine NF-M are

much more related to those of the typical NF-H sequences than

in other mammals, but that other parts of the bovine NF-M tail

do not deviate from the mammalian norm. It will be interesting

to read the bovine NF-M sequence should it become available.

The production of fusion proteins like those described

here will allow the mapping of other neurofilament antibodies

and the production of more antibodies of predefined

specificity. The fusion proteins and antibodies raised against

them will be valuable probes in future studies of the

structure, dynamics and functions of neurofilament tails. The

novel polyclonal antisera should also be ideal for screening

cDNA expression libraries to clone neurofilament or

neurofilament-related proteins, and will also be very useful

for immunocytochemical double-labelling and microinjection




Antibodies to neurofilament proteins are often used as

markers for neurons. It is generally recognized that not all

neurons contain neurofilament triplet proteins. Many studies

have looked at neurofilament triplet protein expression in

various regions of the rat nervous system, but have come up

with different conclusions, especially in the cerebellum

(Matus et al., 1979; Shaw et al., 1981; Trojanowski et al.,

1985; Shaw et al., 1986; Vitadello and Denis-Donini, 1990).

Because neurofilaments, when they are present, tend to be

phosphorylated in the axonal compartment but not

phosphorylated in the somatodendritic compartment, much of the

initial confusion appeared to be resolved by considering

whether the antibody's staining is dependent upon the

phosphorylation state of the neurofilaments (Goldstein et al.,

1983; Sternberger and Sternberger, 1983; Durham, 1990). In

addition, the use of subunit-specific antibodies also makes

the staining results more intelligible (Shaw et al., 1981;

Trojanowski et al., 1985; Carden et al., 1987). Despite these

improvements significant differences in results are still

evident and remain unresolved. Our panel of phosphate-

independent, subunit- and domain-specific polyclonals and

monoclonals should provide data that are more reliable than

previously possible.


For the brain experiments adult cats and Sprague Dawley

rats were anaesthetized with pentabarbitol and perfused with

heparinized normal saline followed by 4% paraformaldehyde in

PBS. Brains were dissected out, post-fixed in the same

solution for four to twelve hours at 40C, transferred to a 20%

sucrose solution at 40C for 6 hours and then placed in a 30%

sucrose solution at 40C for 24 hours. Brains were then quick-

frozen in liquid nitrogen-cooled isopentane and embedded for

cryostat sectioning. Sections of 6-12 pm were collected

either into PBS (for free-floating incubations) or onto

gelatin-subbed slides that were then kept frozen at -200C

until staining. Some rat brains were prepared by quick

freezing without fixation to check for fixation artifacts. A

few Long-Evans rats were used to verify findings in a separate


For examination of structures intimately associated with

bone, such as olfactory epithelia and organs of the inner ear,

tissues were placed in a 40C solution of 7.5% EDTA (pH=8.0)

for at least 2 days, following perfusion and at least 1 hour

post-fixation in 4% paraformaldehyde.


Whole-mounts of retina were prepared by removing the eyes

from perfused animals and removing the cornea and lens to

expose the retina to the post-fixation solution (4%

paraformaldehyde) for an overnight incubation at 40C. Retinas

were usually detached and could be then be easily dissected

from the eye for free-floating immunostaining.

For the myenteric plexus experiments segments from the

small intestine of neonatal [ages in days: 0(newborn), 4, 7,

and 21] and adult (age 60-120 days) rats were taken and

prepared in either of two ways. In the first method, samples

were frozen in supercooled isopentane, mounted and 5-8 pm

frozen sections were cut. Sections were collected on glass

slides and fixed with acetone for staining. In the second

method, intestine was threaded onto glass capillary tubing

and, under the dissecting microscope, the longitudinal muscle

layer and accompanying myenteric plexus were removed. The

longitudinal muscle/myenteric plexus layer was pinned on

Sylgard (Brownell Electric, Orlando, Fl) and fixed overnight

in Zamboni's solution (40C) (Costa et al., 1980; Stefanini et

al., 1967). The next day, the tissue was washed free of the

Zamboni's fixative using 80% ethanol and then the tissues were

dehydrated and rehydrated through graded alcohols prior to

staining as the whole mount preparations. Additionally, rat

brain from the different age groups was taken and prepared

using the two fixation methods described above and sections

were immunostained as controls.


Immunohistochemistry was performed either on free-

floating tissue in solution or on sections on slides that had

been thawed and encircled with a PAP pen (The Binding Site

Inc., San Diego, CA). Tissues were blocked with 0.1% BSA and

0.3% Goat serum in PBS for 20 minutes at 370C. Following a

PBS rinse, primary antisera, often containing approximately

0.1% Triton X-100, were applied for at least 45 minutes at

370C (or overnight at 40C) and then washed three times in PBS

for 10 minutes at room temperature. Fluorescent secondary

antibodies (Jackson ImmunoResearch, West Grove, PA) were

applied at final dilutions of 1:40 in 0.1% BSA and 0.3% Goat

serum in PBS plus 0.1% Triton for at least 45 minutes at 37C

and washed in PBS three times for 10 minutes at room

temperature. Double-label applications were performed in the

same manner, but with primary and secondary antibody mixtures.

Sections and whole mounts were coverslipped using an anti-

bleaching mounting medium (1 mg/ml para-phenylene-diamine in

90% glycerol, 10% Tris 200mM, pH = 7.9), sealed with nail

polish and stored at 4C. Some sections were processed using

an avidin-biotin peroxidase enhancement method (Vectastain

Elite Vector Laboratories) of detecting the primaries,

rather than using fluorescence.

All polyclonal rabbit sera against the native

neurofilaments (H297, H298, H301), the fusion proteins (anti-

rat NF-H KSP, anti-rat NF-H KEP, anti-rat NF-M KE, anti-

chicken NF-M E/KSP) and some monoclonal ascites fluids (DA2,


NAP4, NN18, SMI32) were used at a final dilution of 1:200 in

PBS, whereas hybridoma culture media supernatants for the

fusion protein monoclonals were typically used undiluted. The

a-internexin monoclonal (#135) was a gift from Dr. Ron Liem.

The GABA antibody (Eugenetech, Inc.) was used at 1:750 for a

primary incubation of at least 3 days at 40C.

Acetycholinesterase (AChE) staining was performed on

free-floating sections followed by the immunohisto-fluorescent

labelling exactly as described above. AChE staining involved

pre-incubation in 0.5 M Na citrate, 0.3 M CuSO4, 0.05 M

K3Fe(CN)6, 0.1% Triton X-100 in 0.1 M Acetate buffer (pH=6.0)

for 20 minutes at room temperature, followed by a 15 minute

incubation in 20 ml of pre-incubation solution containing 1.4

mg ethopropazine and 14 mg acetylthiocholine iodide, a 5

minute wash in 0.05 M TRIS (pH=7.6) and developing in 0.5% DAB

in PBS plus 0.1% Triton (Geneser-Jensen and Blackstad, 1973).

Alkaline phosphatase treatments of unfixed and fixed

sections were incubated in E. coli Alkaline phosphatase (Sigma

#P-5931) at 2 mg/ml in TBS for 4 hours at 370C, overnight

(approximately 12 hours) at 40C and then for an additional

hour at 370C prior to immunostaining.


Overview.- Neurofilament triplet staining was evaluated in

many cell types and fiber tracts; most structures were

similarly immunoreactive for NF-L, NF-M and NF-H (Table 4-1).


Cortex Pyramidal cells

Dentate granule- NF-M >> NF-H and NF-L
Dentate Interneurons- NF-H and NF-L > NF-M
CA Pyramidals- NF-H and NF-L > NF-M
CA Interneurons- NF-H and NF-L > NF-M
Mossy fibers- NF-M >> NF-H or NF-L

Purkinje cells- Anti-NF-H KSP >>> all other NF antibodies
Granule cells- Anti-NF-H KSP positive in cat, but not rat
Parallel fibers'- Lower third are positive for NF triplet; NF-H > NF-L > NF-M
Basket cells- Anti-NF-H KSP positive in cat, but not rat
Basket cell axons- Anti-NF-H KSP negative in cat
Stellate cells- Anti-NF-H KSP positive in cat, but not rat
Large Golgi cells- Anti-NF-H KSP positive in cat, but not rat
Small Golgi cells'- NF-H > NF-L >>> NF-M; seen only in rat, not cat
Lugaro cells- Anti-NF-H KSP positive in cat and rat
Deep cerebellar nuclei
Mossy and Climbing fibers

Striatum Internal capsule

Tectum Posterior and Superior colliculi

Tegmentum Locus coeruleus and Substantia nigra

Trigeminal nuclei (mesencephalic, motor, principal, spinal)

Pontine nuclei

Spinal Cord tracts and a-motor neurons

Dorsal Root Ganglia

Retina Ganglion cells and Horizontal cells

Spiral Ganglia
Type I neurons- Immunonegative with anti-NF-H KEP, 3G3 and 5B8
Type 1 neurons

Vestibular Ganglia and Organs Hair cells and Afferent endings (Saccule, Utricle, Ampules)

Olfactory Bulb
Glomeruli, Mitral cells, Tufted cells- Anti-NF-H KSP > all other NF antibodies

Olfactory epithelial neurons'

All of the structures listed above were positive for NF-L, NF-M and NF-H, except where noted otherwise.

'Although all parallel fibers were immunopositive for a-internexin, only the larger diameter fibers in the deep third
of the molecular layer were NF triplet positive.
With all NF-M monoclonals small Golgi cells appeared immunonegative; only a very weak signal was detected with
NF-M polyclonals.
'Only a minority of olfactory epithelial neurons from P4 rat were NF immunopositive.

In some cases, however, there are notable differences in the

staining patterns for the different subunits. As is

consistent with the existing literature, the overall pattern

most often seen was that of a dephosphorylated NF-H KSP

antibody staining preference in cell bodies and a fiber tract

selection by phosphorylated NF-H KSP antibodies (Figure 4-1).

The NF-H KEP, NF-M KE, and NF-L antibodies were particularly

striking, however, in that even though there was a significant

cell body reaction, many of them stained fibers more intensely

than cell bodies (Figures 4-1 to 4-5 and 4-12). Unlike

antibodies directed against the phosphorylated KSP regions of

NF-M and NF-H, the staining pattern of NF-H KEP, NF-M KE and

NF-L antibodies are unaffected by alkaline phosphatase

treatments; perhaps this preferential staining of axons by

markers for non-phosphorylated epitopes reflects a

quantitative difference in the concentration of neurofilament

proteins in the axonal processes compared to the

somatodendritic compartment.

Cerebellum. Beyond the general patterns of

somatodendritic versus axonal staining, two new and unusual

findings were discovered in the rat cerebellum. Firstly, in

horizontal sections we were able to see a subpopulation of

parallel fibers (the larger caliber ones located deeper in the

molecular layer) that were immunoreactive to NF-H and NF-L

(Figure 4-2), and to a lesser degree NF-M (Figure 4-3); this

NF triplet positive subpopulation is compared by double-label

to the entire veil of parallel fibers, not just large caliber

ones, stained by an a-internexin antibody (Figure 4-4).

Secondly, and of much more interest, we found staining of what

appear to be a population of previously undescribed cells that

we think correspond to what have been called small Golgi

neurons at the electron microscopic level (Palay and Chan-

Palay, 1974). These neurons were even more intriguing because

they demonstrated a difference in the amounts of NF-L, NF-M

and NF-H immunoreactivity, with NF-H consistently being much

stronger. In fact a-internexin (Figure 4-5) and most NF-M

(Figure 4-6) antibodies appeared negative, but NF-L and all

NF-H antibodies, irregardless of the phosphate-dependence or

prior phosphatase treatment, robustly stained these cell

bodies (Figures 4-5 through 4-8). This finding is similar,

but not identical, in Purkinje cells, in which the soma and

dendritic trees also stain strongly for NF-H, but not for NF-

M, NF-L and a-internexin; this difference is only seen,

however, by using antibodies against the dephosphorylated NF-H

KSP (Figures 4-1, 4-4 to 4-6, 4-10 and 4-11). Because the

small Golgi neurons were not previously described as being

strongly immunoreactive for NF-H, considerable effort was

expended to confirm their identity. The cells were of the

right size and distribution (granular layer, most concentrated

in vestibular lobules IX and X) (Figures 4-7 and 4-8). In

addition, AChE staining distinguished a subset of nodular

mossy fibers (as described by Brodal and Drablos in 1963)

< ^ -*, 1'-l'. <,
**A r ^'
;I ^ *\ "*"f c "'*i?

Figure 4-1. Axonal versus somatodendritic staining patterns of
NF antibodies. A and B: Double-label immunofluorescence of a
sagittal section of rat cerebellum using a monoclonal to the
NF-H KEP, 3G3 (A) and the anti-NF-H KSP (B). C and D: Double-
label of a horizontal section of rat cerebellum stained with
a monoclonal against phosphorylated NF-H KSP, NAP4 (C) and the
anti-NF-H KSP serum (D), which prefers dephosphorylated NF-H.
Approximate magnification: A and B = 40x; C and D = 100x.

Figure 4-2. NF-H and NH-L antibody staining of a subpopulation
of parallel fibers. Double-label immunofluorescence of a
horizontal section of rat cerebellum using the polyclonal anti
NF-H KEP (A) and the monoclonal to NF-L, DA2 (B). Approximate
magnification: 100x.

Figure 4-3. NF-H and NF-M antibody staining of a subpopulation
of parallel fibers. Double-label immunofluorescence of a
horizontal section of rat cerebellum using the polyclonal anti
NF-H KEP (A) and the monoclonal to NF-M, 3H11 (B). Approximate
magnification: 100x.

Figure 4-4. NF-H and a-internexin antibody staining of
parallel fibers. Double-label immunofluorescence of a
horizontal section of rat cerebellum using the polyclonal
anti-NF-H KSP (A) and the monoclonal against a-internexin,
#135 (B). Compared to the #135 monoclonal which stains all off
the parallel fibers, the NF-H antiserum only reveals a
subpopulation of parallel fibers in the deep third of the
molecular layer. Approximate magnification: 100x.

Figure 4-5. NF-H versus a-internexin antibody staining of a
horizontal section through the rat nodulus. Double-label
immunofluorescence using the a-internexin monoclonal #135 (A)
and the polyclonal anti-NF-H KSP (B); The small Golgi neurons
are visible only with the NF-H antibody. Approximate
magnification: 63x.

Figure 4-6. NF-H versus NF-M antibody staining of the rat
uvula. A and B: Double-label immunofluorescence of a sagittal
section with the polyclonal anti-NF-H KSP (A) and NF-M
monoclonal NN18 (B). C and D: Double-label immunofluorescence
of a sagittal section with the polyclonal anti-NF-H KSP (C)
and NF-M monoclonal 3H11 (D). The small Golgi neurons are
visible only with the NF-H antibody. Approximate
magnification: A and B = 20x; C and D = 100x.

Figure 4-7. NF-H positive small Golgi neurons.
Immunofluorescence of the rat nodulus containing small Golgi
neurons stained with the anti-NF-H KSP polyclonal. Note the
number and morphology of these cells. Approximate
magnification: A = 63x; B and C = 200x.

I, "

Figure 4-8. Distribution of NF-H positive small Golgi neurons
compared to acetylcholinesterase (AChE) positive nodular-type
mossy fiber terminals by double-label. A: Immunofluorescence
of an anti-NF-H KSP stained sagittal section of rat cerebellum
demonstrating that the small Golgi neurons are predominantly
found in the granule cell layer of the vestibulocerebellum
lobuless IX and X, arrow). B: AChE staining of the same
section as in A, showing an enhanced reactivity in the granule
cell layer of lobule IX and X as compared to the other
lobules. C: AChE positive mossy fiber endings (arrows). D:
Double-label of same section as in C, showing the location of
NF-H KSP positive small Golgi cells (arrowheads) as well as
some antibody-labelled climbing fibers. Approximate
magnification: A and B = 20x; C and D = 200x.

Figure 4-9. Lugaro cell in the rat cerebellum.
Immunofluorescent labelling of a Lugaro cell (arrow), as well
as some Purkinje cells, some transverse basket cell axons and
some small Golgi neurons with anti-NF-H KSP. Approximate
magnification: 200x.

Figure 4-10. Anti-NF-H KSP labelling of cat Purkinje cell
axons and a large Golgi neuron. A: Intense staining of an axon
(arrows) as it exits a Purkinje cell body. B: Faint staining
of a large Golgi neuron (arrow). Approximate magnification:
A = 250x; B = 150x.

-* ".> .a .'<

Figure 4-11. Anti-NF-H KSP labelling of cat granule and Lugaro
basket cells. A: Granule cells (open arrows) are faintly
labelled by neurofilament antibody; note that their processes
are thinner than those of small Golgi cells (Figure 4-7) which
taper more slowly into a dendritic trunk. B: Staining of an
horizontally-oriented interneuron (arrow) at the base of the
Purkinje cell layer which is either a Lugaro cell or subclass
of basket cell. C: Labelling of two typical basket cell
(arrows). Approximate magnification: A = 75x; B = 100x; C =

Figure 4-12. Double-label of NF-H versus NF-M immunoreactive
patterns in rat hippocampus. Anti-NF-H KSP (A) stains many
different hippocampal cell and fiber types, but is less
reactive with the dentate granule cell layer and the mossy
fibers concentrated in the very center of the dentate. The
NF-M antibody 3H11 (B) stains the same section in a nearly
complementary fashion, reacting most strongly with the mossy
fibers and the dentate granule cell layer. Approximate
magnification: 40x.

Figure 4-13. Labelling of assorted neurons by NF antibodies.
A: Intense labelling of an occasional primary neuron of the P4
rat olfactory epithelium with DA2, a NF-L monoclonal; besides
the soma, both the dendritic knob (arrowhead) and the axon
(arrow) is visualized by the antibody. B: Staining of a layer
V pyramidal neuron from rat cerebral cortex with NF-M KE
monoclonal 3H11. C: Intense staining of rat olfactory bulb
structures with anti-NF-H KSP polyclonal. Labelled structures
include glomeruli (asterisks), mitral cells (arrow) and tufted
cells (arrowhead). D: Intense labelling of CA1 pyramidal
neurons from rat hippocampus with anti-NF-H KSP serum.
Approximate magnifications: A and B = 200x; C = 40x; D = 100x.

Figure 4-14. Anti-NF-H KSP staining of retinal ganglion and
horizontal cells from rat and cat. Visualization of the same
area of a whole mount rat (A and B) or cat (C and D) retina in
two different focal planes shows staining of ganglion cells
(open arrows, A and C) and horizontal neurons (arrows, B and
D). Approximate magnification: A and B = 100x; C and D = 200x.

Figure 4-15. Immunofluorescent staining of rat inner ear
structures by Anti-NF-H KSP. Structures labelled include: 1)
eighth cranial nerve; 2) vestibular ganglion; 3) cochlea; 4)
sacculus; 5) utricle; 6) ampulla of semicircular canal.
Approximate magnification: 20x.

Figure 4-16. Rat vestibular ganglion and organs labelled by NF
antibodies. A: Vestibular ganglion stained with anti-NF-M KE.
B: NF-M monoclonal NN18 stained calyceal afferent nerve
endings for hair cells of the crista of an ampulla. C: NN18
staining of the afferent fibers and calyces around hair cells
of the utricular macula. D: Apparent staining of hair cells
as well as the afferent fibers of the sacculus by 5B8, a
monoclonal against the KEP domain of NF-H. Approximate
magnification: A, B and C = 63x; D = 100x.

Figure 4-17. Double-label immunofluorescence of spiral
ganglion by Anti-NF-H KEP and NF-M monoclonal NN18. A
demonstration of intense labelling of all spiral ganglion
neurons and afferent fibers by NN18 (A), but only strong label
of fibers by anti-NF-H KEP (B). Approximate magnification:

Figure 4-18. Double-label immunofluorescence of rat spiral
ganglion by anti-NF-M KE and NF-H KEP monoclonal 5B8. A and
B are double-labels of the same section, as are C and D.
Although both type I (heavy white arrows) and type II (thin
black arrows) neurons are labelled by anti-NF-M KE (A and C),
only type II neurons can be seen with 5B8 (B and D). Note
that there appears to be a set of fibers which are labelled by
the anti-NF-M KE, but not 5B8 (open arrow, A). Approximate
magnification: 200x.

Figure 4-19. Double-label immunofluorescence of myenteric
neurons stained with Anti-NF-M KE and NN18. Whole mounts of
myenteric plexi stained with anti-NF-M KE (A, C, and E) and
NN18 (B, D, and F). Anti-NF-M KE stains myenteric neurons in
all ages. Neonate intestine (A and B) shows comparable
positive immunostaining in neurons (arrows) with anti-NF-M KE
and NN18. Intestine from 7 day old rats (C and D) shows a
decrease (to background) of NN18 staining in neurons (arrows).
Adult intestine (E and F) demonstrates no detectable NN18
signal in myenteric neurons. Scale bar = 20 pm.

Figure 4-20. Double-label immunofluorescence of myenteric
neurons stained with Anti-NF-M KE and RMO1. Whole mounts of
myenteric plexi stained with anti-NF-M KE (A, C, and E) and
RMO1 (B, D, and F). Anti-NF-M KE stains myenteric neurons in
all ages. Neonate intestine (A and B) shows comparable
positive immunostaining in neurons (arrows) with anti-NF-M KE
and RMO1. Intestine from 7 day old rats (C and D) and adult
(E and F) demonstrates no detectable RMO1 signal in myenteric
neurons (arrows). Scale bar = 20 p~m.

Figure 4-21. Double-label immunofluorescence of myenteric
neurons stained with Anti-NF-M KE and RM059. Whole mounts of
myenteric plexi stained with anti-NF-M KE (A, C, and E) and
RM059 (B, D, and F). Anti-NF-M KE stains myenteric neurons in
all ages. Neonate intestine (A and B) shows comparable
positive immunostaining in neurons (arrows) with anti-NF-M KE
and RMO1. Intestine from 7 day old rats (C and D) and adult
(E and F) demonstrates a decreased, but detectable RM059
signal in myenteric neurons (arrows). Scale bar = 20 in.


whose clustered endings often appeared to terminate near the

NF-H immunopositive small Golgi cells (Figure 4-8). With the

exception of an occasional Lugaro cell (Figure 4-9), prominent

neuronal cell types other than Purkinjes and small Golgis

(i.e., granule, basket, stellate, and large Golgi cells) were

not seen with neurofilament antibodies in rat cerebellum.

Cat cerebellum was examined to see if these findings were

present across species. Nearly all cerebellar neuron types

(Purkinje cells, basket cells, stellate cells, Lugaro cells,

large Golgi neurons and even granule cells) were clearly NF-H

positive in the cat, except that small Golgi cells were not

seen using the anti-NF-H KSP serum (Figures 4-10 and 4-11).

It may be of note that although Purkinje cells and their axons

were even more prominently visible in the cat, the basket cell

axon pinceaus were routinely not seen using the anti-NF-H KSP

serum (Figure 4-10), though they could be seen with other NF

triplet antibodies.

Hippocampus. Another prominent example of a subunit

distribution difference was found in the hippocampus. The

dentate granule cells and their axonal projection to the

Ammon's horn pyramidals, the mossy fibers, were consistently

more reactive with antibodies against NF-M than for NF-H or

NF-L (Figure 4-12). Although NF-M was apparently more

abundant in these cells and their fibers, it was also present

in some other hippocampal interneuron cell bodies, though not

as much as NF-L and NF-H. The pyramidals of Ammon's horn were


most reactive to NF-H and NF-L antibodies, especially the

anti-NF-H KSP serum (Figure 4-13).

Cerebral cortex. The pyramidal cells of the cerebral

cortex were also labelled by NF triplet antibodies. Subtle

differences in staining occurred between cortical regions and

layers, but these differences appeared to correlate with the

numbers of pyramidal type neurons. In general the larger

pyramidal cells of layer V were most completely labelled

(Figure 4-13).

Olfactory neurons. NF triplet antibodies were quite good

markers for several type of neurons in the olfactory pathways,

including primary sensory neurons in the olfactory epithelium

as well as mitral and tufted cells in the olfactory bulb

(Figure 4-13). Although labelling occurred for NF-M, NF-L and

NF-H usually reacted more strongly with cell bodies of primary

and secondary neurons as well as the large dendritic processes

in the glomeruli of the olfactory bulb. We saw labelling of

primary olfactory neuron dendritic knobs, cell bodies and

axons with antibodies against all three NF triplet proteins

(Figure 4-13).

Retina. The retinal ganglion and horizontal cell types

of both the rat and cat were stained by the NF fusion protein

antibodies. Tissue was prepared as a whole mount and various

cell types and fibers were examined by focusing through the

layers. Not only did they appear larger, but the morphology


of the horizontal cells and their processes were more clearly

defined in the cat than the rat (Figure 4-14).

Inner ear. Another area of the rat nervous system in

which our NF antibodies detected unexpected findings was in

the organs of the inner ear, including the spiral ganglion of

the cochlea and the vestibular apparati (Figures 4-15 to 4-

18). Afferents to vestibular organs ampullaee, saccule and

utricle) were strongly labelled and often appeared to form a

calyx at the base of the hair cells; in some cases the hair

cells themselves seemed to be stained (Figure 4-16). The

vestibular ganglion neurons appeared to be a heterogeneous

population based on the spectrum of NF immunoreactivity

(Figure 4-16). A similar finding was more clearly documented

in the spiral ganglion (Figures 4-17 and 4-18). Approximately

90% of the neurons (type I) of the spiral ganglion send

myelinated afferents to the inner hair cells of the Organ of

Corti, whereas the remainder (type II) send unmyelinated

processes to the outer hair cells. Antibodies directed

against the KEP domain of NF-H stained only the type II spiral

ganglion neurons, which are also noticeably smaller than type

I neurons; all other NF triplet antibodies, however, reacted

with both type I and II neurons (Figures 4-17 and 4-18).

Myenteric neurons. As a result of using antibodies which

we had epitope-mapped using the fusion proteins, we were also

able to extend some previous findings in the enteric nervous

system (Table 4-2). Most myenteric plexus neurons from adult



Brain Myenteric Plexus

Antibodies 0 day Adult 0 day 4 day 7 day 21 day Adult

Anti-NF-M KE + + + + + + +
Anti NF-H KSP + + + + + + +
Anti NF-H KEP + + + + + + +
Anti NF-L (H297) + + + + + + +

NN18 + + + -
RMO1 + + + -
RM059 + + + + 4 1 4

SC6 + + + + + + +
1G12 + + + + + + +

1G9 + + + + + + +
5G9 + + + + + + +
3H11 + + + + + +

(+) = positive immunoreactivity, (-) = negative immunoreactivity, (1) = decreased immunoreactivity as
compared to brain controls.

rats demonstrated neurofilament staining with antibodies

directed against different subunits and domains. However, as

previously shown (Eaker et al., 1991) enteric neurons from

adult rats stain weakly with the NF-M monoclonal NN18 compared

to other NF-M, NF-L or NF-H antibodies (Figure 4-19). RMO1

and RM059, now shown to recognize the same segment of NF-M as

NN18, also produce less staining of adult intestine (Figures

4-20 and 4-21, respectively). Neonatal myenteric neurons were

examined with the same neurofilament antibodies, but were

stained by all of them, including NN18 (Figure 4-19), RMO1

(Figure 4-20) and RM059 (Figure 4-21). Myenteric neurons were

observed to be more tightly packed and smaller in neonates as

compared to adult. Intestinal tissues from four-day-old and,

to a greater extent, seven-day-old animals displayed

noticeably decreased, or even absent, neuronal

immunoreactivity with NN18, RMO1 and RM059 (Figures 4-19, 4-20

and 4-21, respectively), while maintaining positive staining

with the other antibodies. By day 21 the morphology and

staining of myenteric neurons was identical to that seen in

adult (Table 4-2). This epitope-masking event was seen in

both Zamboni-fixed whole mounts and unfixed cryostat sections

of intestine, but not in brain controls.


We have found a widespread distribution of neurofilament

positive cells and fiber tracts in nearly every major area of

the adult rat nervous system (Table 4-1). We have

demonstrated staining differences based not only on

phosphorylation of the KSP domains, as has been previously

described throughout the literature (Sternberger and

Sternberger, 1983; Shaw et al., 1986; Lee et al., 1986b), but

also on a differential distribution of the NF-L, NF-M and NF-H

subunit proteins and some of their domains. Apparent

differences between the NF subunit and domain distributions

are likely to reflect the actual difference in protein

concentrations in axons versus cell bodies. Concentration

differences would explain why antibodies with non-repetitive

epitopes (NF-L, NF-M KE and NF-H KEP) would appear to prefer

axons. The resistance of these staining distributions to

alkaline phosphatase treatments supports this conclusion.


Several other new patterns of neurofilament expression were

discovered in the rat cerebellum, hippocampus, spiral ganglion

and the myenteric plexus.

A prominent cell type in the granular layer of the

vestibulocerebellum of rat was conspicuously immunoreactive

against all NF-H antibodies, but had significantly less

reactivity against NF-L antibodies and a nearly undetectable

signal for NF-M. The cells were immunonegative for GABA. They

often appeared to co-distribute with nodular type mossy fiber

terminals as originally described by Brodal and Drablos

(1963). Although there are some descriptive discrepancies, we

believe that these cells are those identified by Palay and

Chan-Palay (1974) as small Golgi neurons, and as "pale" cells

by Altman and Bayer (1977) and Sturrock (1990). Although

large Golgi cells have been described as GABAergic (Gabbott et

al., 1986), these cells were immunonegative for GABA. The

small Golgi cell presents an interesting phenomenon in which

there is clearly intense staining for NF-H, but nearly none

for NF-M. When cat cerebellum was examined with our antibody

panel, most cell types were easily detected; the small Golgi

neuron, however, if present, was not immunoreactive for

neurofilament -proteins. Differences in the intermediate

filament complement of certain cell types across species have

been reported (Shaw and Weber, 1984; Lee et al., 1986a). We

have also found staining of a subpopulation of parallel fibers

in the rat cerebellum, a finding either unseen or seen but


often ignored or misrepresented in data presented by many

previous investigators (Matus et al., 1979; Trojanowski et

al., 1985; Bignami and Clark, 1985; Marc et al., 1986; Langley

et al., 1988; Dautigny et al., 1988; Vitadello and Denis-

Donini, 1990; Roussel et al., 1991; Kaplan et al., 1991).

Furthermore, we were able to visualize some granule cells

staining with anti-NF-H KSP serum in cat cerebellum. Previous

studies have indicated a transient developmental expression of

NF-L and NF-M in granule cells, but only a-internexin has been

found in the adult (Cambray-Deakin and Burgoyne, 1986; Chiu et

al., 1989). Our ability to detect NF-H in the granule cell

bodies in mature animals, adds further evidence that at least

a subpopulation of parallel fibers are likely to contain NF-H.

All staining patterns in the rat were verified on two

different strains (Sprague-Dawley and Long-Evans) as well as

on both fixed and unfixed tissue; none of our findings were

strain-dependent nor altered by fixation.

Examination of the hippocampus with numerous members of

our NF antibody panel, allowed detection of a differential

subunit distribution in the rat hippocampus. Specifically,

NF-M was found to be more prominent in the dentate granule

cells and mossy fibers. In a nearly complementary pattern,

NF-H and NF-L were more obviously present in the cell types

other than dentate granule cells, particularly in the

pyramidal neurons of Ammon's horn and in the various

interneurons of the hippocampus.


Other major areas of the nervous system including

cerebral cortex, olfactory epithelium, olfactory bulb and

retina were well stained by our NF fusion protein antibodies.

Cortical pyramidal cells, especially those of layer V, were

most conspicuously labelled in the cerebrum. Primary sensory

neurons of the olfactory epithelium were labelled, but not

uniformly. The majority of cells did not stain, but those

that did label often were strongly reactive not only in the

soma, but in the processes as well. We saw labelling of

primary olfactory neuron dendritic knobs with NF-H antibodies

as reported in the literature (Bruch and Carr, 1991), but we

also saw staining of cell bodies and axons with antibodies

against all three NF triplet proteins (Figure 4-13). Previous

studies have described the developmental and cell-type

specific complement of intermediate filaments in the retina

(Shaw and Weber, 1983 and 1984). Similar to what had been

described, the rat retinal ganglion and horizontal cell types

were stained by the NF fusion protein antibodies. These data

were extended by examining cat retinal whole mounts, which had

an identical staining pattern; it may be of note that compared

to the rat the morphology of horizontal cells and their

processes are more clearly defined in the cat.

In the spiral ganglion we found that markers against NF-H

KEP domain selectively identified the type II neurons only.

Other NF antibodies recognize all of the spiral ganglion

neurons (type I and II). Similar findings have been reported

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EZZP7KTVS_AG3AC6 INGEST_TIME 2012-02-29T17:53:55Z PACKAGE AA00009058_00001