E COLOGY AND EVOLUTION OF VECTOR BORNE VIR USES: EMPIRICAL AND MATHEMATICAL APPROAC HES TO UNDERSTANDING THE PERSISTENCE OF CLINICALLY IMPORTANT ARBOVIRUSES By GABRIELA MAXINE BLOHM A DISSERTATION PRESENTED TO THE GRADUATE SCH OOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2017
2017 Gabriela Maxine Blohm
To my grandparents
4 ACKNOW LEDGMENTS I thank my parents, Mary Lou Adams de Blohm and Jorge Tomas Blohm, for teaching me to constantly push myself. I thank my siblings, Melissa, Melanie and John Robert, for supporting, inspiring and challenging me in ways that have helped me grow My grandfather Tomas Blohm and my grandmother Cecilia Montemayor de Blohm sparked my interest in the sciences, thus I would not have pursued this professional direction without their influence I am thankful every day for the support, encouragement, loyal ty and friendship of my partner Juan Carlos Oteyza and our dog Zamba. I thank Jose Miguel Ponciano for his dedication t o thinking rigorously and to his patience in teaching me how to connect models and biology. I will be eternally thankful for his encourag ement and for his influence as a mentor. I also thank Marta Wayne for giving me the opportunity to pursue research for her friendship, and for her influence on my personal and professional growth I thank Jim Maruniak for his friendship, kindness and for opening the doors to his lab so that I could learn how to work with cells. I thank B arry Alto for running a research lab that is focused on problem solving, cooperation and scientific rigor. I learned so much while watching him manage the complexities bo th scientific and logistical of running a BSL 3 laboratory. I also thank him for providing me with the opportunity to gain experience work ing with cells and WNV in the BSL 3. I thank Mike Miyamoto for his insights and encouragement throughout the develop ment of my dissertation I t hank Juliet Pulliam for her influence on my intellectual development, and research for solving public health problems I thank the students that help ed me collect flies in Georgia and run experiments in the Wayne lab. I am thankful to Kylie Zirbel and Eva Buckner for their help and guidance at FMEL Maia Martcheva and Hayriye Gulbudak from the Department of Mathematics at the University
5 of Florida contributed to the development of the mathematical models on West Nile Virus replication The work in Venezuela would not have been possible without Dr. Alberto Paniz Mondolfi, whose c ommitment to alleviating our humanitarian crisis is unshakeable I thank John Lednicky, Gl enn Morris, Julia Loeb, Sarah White and Tania Bonny for letting me work in their lab, for their insights on experimental design and virus isoloation, and for helping me build a bridge towards the work I hope to do in the future in Venezuela. Finally I tha nk Lucy, my loyal companion through most of this process. I thank uncle Kenny for being an endless source of encouragement and support in my life. I also thank the Venezuelans who risked their lives to bring peace and order back to our beautiful country.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 ABSTRACT ................................ ................................ ................................ ................................ ... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ...................... 12 2 Drosophila melanogaster AND SIGMA VIRUS: A MODEL SYSTEM FOR UNDERSTANDING THE PERSISTENCE OF VERTICALLY TRANSMITTED ARBOVIRUSES ................................ ................................ ................................ ..................... 17 Modeling the Dynamics of Sigma virus ................................ ................................ ................. 19 Methods ................................ ................................ ................................ ................................ .. 21 CO 2 Assay for S igma Virus Infection ................................ ................................ ............. 22 Transmission Assays ................................ ................................ ................................ ....... 22 Multigeneration Patrilineal Transmission ................................ ................................ ....... 24 Effect of Sigma Virus on Female Fecundity ................................ ................................ .... 25 Relating the Model to Experiments ................................ ................................ ................. 26 Results ................................ ................................ ................................ ................................ ..... 27 Transmission Efficiency ................................ ................................ ................................ .. 27 Prevalence ................................ ................................ ................................ ........................ 28 Virulence and Cost to Fecundity ................................ ................................ ..................... 30 Discussion ................................ ................................ ................................ ............................... 34 3 EFFECTS OF PHOSPHORUS ON THE REPLICATION OF WEST NILE VIRUS IN CELL CULTURE ................................ ................................ ................................ ................... 39 Methods ................................ ................................ ................................ ................................ .. 40 Cells and Virus Stocks ................................ ................................ ................................ ..... 40 Phosphorus Manipulations ................................ ................................ .............................. 41 Effects of Phosphorus on Rate of Cell Division ................................ .............................. 43 Effects of Phosphorus on Rate of Virus Replication ................................ ....................... 43 Resu lts ................................ ................................ ................................ ................................ ..... 44 Phosphorus Manipulations ................................ ................................ .............................. 45 Effects of Phosphorus on Rate of Cell Division ................................ .............................. 45 Effects of Phosphorus on Rate of Virus Replication ................................ ....................... 47 4 MODELING THE REPLICATION OF FLAVIVIRUSES ................................ ........................ 50
7 Methods ................................ ................................ ................................ ................................ .. 51 Model of WNV Replication ................................ ................................ ............................ 51 Mathematical Description of the Model ................................ ................................ .......... 52 Stability Analysis for the Endemic Equilibrium ................................ ...................... 53 Simulations ................................ ................................ ................................ ...................... 58 Assessing the Estimability of the Model Parameters ................................ ...................... 58 Results ................................ ................................ ................................ ................................ ..... 60 Discussion ................................ ................................ ................................ ............................... 62 5 CLINICAL CASES: TRANSMISSION OF ZIKA AND OTHER ARB OVIRUSES IN VENEZUELA ................................ ................................ ................................ ......................... 65 Methods ................................ ................................ ................................ ................................ .. 65 Patient Specimens ................................ ................................ ................................ ............ 68 Cell Cultu re ................................ ................................ ................................ ...................... 69 Cell Culture Inoculations ................................ ................................ ................................ 69 Evidence of ZIKV Isolation ................................ ................................ ............................. 70 RT PCR Screens for Chikungunya, Dengue, and Zika virus RNA ................................ 70 ZIKV Sequencing ................................ ................................ ................................ ............ 70 ZIKV Phylogenetic Analysis ................................ ................................ ........................... 71 Serology ................................ ................................ ................................ ........................... 71 Results ................................ ................................ ................................ ................................ ..... 72 Discussion ................................ ................................ ................................ ............................... 75 6 CONCLUSIONS AND DISCUSSION ................................ ................................ ...................... 77 APPENDIX A FORMULATION OF CELL CULTURE MEDIUM ................................ ................................ 81 B M A N U S C R I P T S I N P R E P A R A T I O N : CUTANEOUS MANIFESTATIONS OF ZIKV ....... 83 LIST OF REFERENCES ................................ ................................ ................................ ............. 105 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 111
8 LIST OF TABLES Table page 2 1 Location of collection sites ................................ ................................ ............................... 21 2 2 ANOVA results for single generation transmission efficiency ................................ ......... 27 2 3 Virus prevalence and number of flies per collection.. ................................ ........................ 29 2 4 ANOVA results for the prevalence of sigma virus in the field. ................................ ......... 30 A 1 Formulation of L 15 cell culture medium. ................................ ................................ ......... 81
9 LIST OF FIGURES Figure page 2 1 Procedure for measuring prevalence and trans mission. ................................ ..................... 23 2 2 Prevalence of sigma virus in North central Georgia. ................................ ......................... 30 2 3 Effect o f infection status on fecundity ................................ ................................ .... 31 2 4 Transmission efficiency of sigma virus (single generation). ................................ ............. 32 2 5. Patrilineal transmission of sigma virus. ................................ ................................ ............. 33 2 6 Equilibrium prevalence of Sigma virus as a function of female transmission efficiency and male transmission efficiency. ................................ ................................ ..... 35 3 1 Effectiveness of pho sphorus treatments. ................................ ................................ ............ 45 3 2 Vero cell population size at three phosphorus levels, six days after inoculation. .............. 46 3 3 Effects of pho sphorus on the growth trajectory of Vero cells. ................................ .......... 47 3 4 Effect of phosphorus on West Nile Virus population size in Vero cells. .......................... 48 3 5 Effect of phosphorus on the number of plaque forming virus particles. ........................... 49 4 1 Illustration of a generalized Flavivirus replication cycle. Virus particles ( V ) become attached ( A ) to the cell m embrane ................................ ................................ ..................... 52 4 2 Testing the e stimability of the parameters ................................ ................................ ......... 61 4 3 Relative bias of each parameter when certain parameters are known after running 100 simulations for each set of parameters.. ................................ ................................ ...... 62 5 1 ................................ ................................ ....... 72 5 2 Characteristic ZIKV specific cytopathic effects in LLC MK2 cells. ................................ 73 5 3 Maximum Likelihood (ML) tree inferred from available ZIKV whole genome sequences. ................................ ................................ ................................ .......................... 75
10 Abstract of Di ssertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ECOLOGY AND EVOLUTION OF VECTOR BORNE VIRUSES: EMPIRICAL AND MATHEMATICAL APPROACHES TO UNDERSTANDING THE PERSISTENCE OF CLINICALLY IMPORTANT ARBOVIRUSES By Gabriela Maxine Blohm August 2017 Chair: Jose Miguel Ponciano Major: Zoology Understanding the factors that allow the establishment and transmission of vector borne viruses is imp ortant for developing effectiv e intervention strategies. I n this dissertation, I combine field surveys, laboratory experiments and mathematical models to determine 1) how transmission mode (i.e. vertical transmission vs. horizontal transmission) can affect the persistence of a parasite; 2) how non living environmental factors (specifically nutrient enrichment) can affect the replication of arboviruses and 3) how host level biological factors such as prior exposure to infection can affect disease severity a nd transmission rate. Results suggest that vertical transmission is an evolutionarily stable strategy, even when it brings a reduction in host fecundity. Laboratory experiments on the effects of phosphorus on replication of West Nile Virus reveal that phos phorus is a growth limiting factor for cells; however it does not limit the replication of the virus. Mathematical models and simulations of West Nile Virus replication allow us to i dentify areas in the replication cycle that can be effective targets for a ntiviral therapies. Findings from field based surveillance studies of arbovirus prevalence in Venezuela reveal that Zika virus can be transmitted perinatally, suggesting that disease
11 surveillance and vector control should be coupled with basic clinical vir ology research on arbovirus prevalence.
12 CHAPTER 1 INTRODUCTION The geographic range of a rthropod borne viruses (arboviruses) such as Yellow Fever virus (YFV), Dengue virus (DENV), West Nile virus (WNV), Chikungunya virus (CHIKV) and Zika virus (ZIKV) ha s i ncreased rapidly ( Taylor 2008, Weaver 2010 Patterson 2016) Human migration, global changes in temperature and precipitation cycles urbanization and agricultural development have facilitated the geographic expansion of DENV, CHIKV and ZIKV (reviewed in M ayer 201 7 ). Successful establishment of the se virus es depends on several factors such as virulence, mode of transmission susceptibility of the host and of the vector and the epidemiologica l role of the host ( e .g. as a reservoir, amplifier, or sink) amon g others ( Weaver 2010, Mayer 2017 ). Because these factors operate at vastly different scales of biological organization, it is useful to combine a diversity of approaches when attempting to explain the recent geographic expansion of arboviruses (Lord 2014) Model systems, experimental studies, field surveys, mathematical models and statistical tools are among the approaches that can be combined to improve our ability to predict and control arboviruses. In this dissertation, I use a multidisciplinary approac h to explore how virulence, transmission mode and environmental context can affect the emergence of arboviruses. Virulence the reduction in host fitness by a parasite (Read 1994) can affect the persistence of a parasite. If the virulence of a parasite i s high, it is expected that the opportunities for transmission will decrease due to a reduction in the fitness of the host (Ebert 1999). This tradeoff between virulence and transmission provides an explanation for the persistence of many vector borne virus es where the vectors are thought to experience a relatively small reduction in fitness when infected (Lambrechts 2009). Within the vector population, a rboviruses can be transmitted horizontally ( e.g. orally when the vector is obtaining an infectious blood meal from
13 another host) and/or vertically (i.e. transovarially from an infected mother to her offspring). V iruses that can be transmitted vertically by their vectors include WNV, DENV and YFV (Mishra 2001, Goddard 2003, Murray 2010, Sall 2010 Buckner 2013 ). V ertical transmission can allow the interannual persistence of a n arbo virus, thus expanding its geographic range to new areas and lengthening the duration of an epidemic ( Lambrechts 2009 ) V iruses that combine vertical transmission with horizontal trans mission in the vector population are of interest in evolutionary biology and in public health Sigma virus ( Rhabdoviridae ) is a vertically transmitted, virulent parasite that infects natural populations of D. melanogaster worldwide. It is transmitted verti cally and biparentally (L'Hritier, 1958; L'Hritier, 1970) and is virulent: infected hosts exhibit decreased fecundity (Fleuriet, 1981). The Sigma virus Drosophila melanogaster model system provides an opportunity to investigate the mechanisms permittin g parasite persistence with vertical transmission. Accurately predicting the spread of arboviruses requires an understanding of how the environmental context can mediate virulence and viral replication a process which can be modeled mathematically to im prove the accuracy of predictions when scaling up to the host population level (Lord 2014) Due to urbanization and agricultural development, some ecosystems have become suitable for arboviruses and their vectors (Weaver 2010) Agricultural development has brought a rapid increase in the availability of carbon (C) nitrogen (N) and phosphorus (P) in freshwater ecosystems favoring vector species that thrive in nutrient rich larval environments Among these vector species are the Culex pipiens complex, whi ch includes several important vectors of WNV (Reisen 2013) Ecological Stoichiometry t heory predicts that at higher nutrient levels parasite load should increase ( Lafferty 2009 ); the nature of this relationship is unknown for mosquito vectors where select ive pressures on virulence are
14 expected to minimize the fitness cost of infection. Th e prediction that viruses are P limited within their hosts has been supported in an experimental setting, where P was found to increase the prevalence of Barley yellow dwa rf virus in healthy hosts ( Borer 2010 ). The effects of nutrient enrichment on the abundance of WNV have been investigated at the landscape level in a correlational study (Crowder 2013) ; however to our knowledge, there is little experimental work on the eff ects of P on WNV population growth, which we suspect is correlated with virulence In this dissertation, I experimentally manipulate the concentration of phosphorus in cell culture to determine whether WNV titers would increase under nutrient enriched scen arios. I then present a mathematical model of virus replication in which I identify areas where phosphorus would have the largest effect on WNV replication. Recent technological advances have improved our knowledge of the replication process of Flavivirus es; however direct measurement of virus replication in the laboratory can be cost prohibitive. Mathematical models of the within host dynamics of virus replication can be useful for predicting virus load, which can affect virulence, transmission rate and r ate of molecular evolution. In the case of arboviruses, most mathematical models have focused on transmission dynamics between hosts and vectors. Within host replication of arboviruses (particularly Flaviviruses) has not been well characterized mathematica lly. Several mathematical models of Hepatitis C virus (HCV: Flaviviridae) have captured several important aspects of virus replication, generating accurate predictions of within host dissemination of the virus ( Dahari 2007, Kumberger 2016 ) and of viral ent ry into host cells ( Padmanabhan 2011 ). Using previous work on mathematical models of HCV replication, I model the replication of WNV in cell culture and generate testable predictions about the spread of the virus within a host. I model the trajectory of a v irus population under a set of biologically sound parameter values As has been
15 done with HCV, t he resulting mathematical model can also be applied in a clinical setting, where there is a growing body of work on the within host dissemination of Flavi vir us es Laboratory experiments and mathematical models can facilitate progress towards controlling the spread of arboviruses; however in the absence of field research, the results from these studies cannot be ground tested and the models cannot be improved. C onducting clinical research in tandem with the development of these models is important. To begin testing our predictions, we initiated an arbovirus surveillance study in Barquisimeto, Venezuela. The focus of this study was to provide information on the ZI KV epidemic in Venezuela, where an immunologically nave population experienced an epidemic of ZIKV, concurrent with endemic circulation of DENV, CHIKV, YFV and Mayaro virus (MAYV) Venezuela is situated in an area of high circulation of arboviruses ( Hotez 2017 ). The effects of co infection and immunological status can thus be investigated via clinical research on patients reporting symptoms of febrile illness. In the final section of this dissertation, I present a case study from an epidemiological study i n which we identify virus loads and transmission mode of ZIKV in an immunologically nave population. The establishment of an arbovirus involves processes that operate at multiple scales of biological organization (e.g. physiological, ecological and evolu tionary) A multidisciplinary approach that combines field surveillance, laboratory experiments and mathematical models can provide effective tools for linking these scales and generating testable hypotheses for future work (Lloyd Smith 2009 Lord 2014 ) I n this dissertation, I combine field surveys, laboratory experiments and mathematical models to address several aspects of the establishment of arboviruses. Specifically, I investigate 1) how transmission mode ( vertical vs. horizontal transmission) can aff ect the persistence of a virulent parasite in the Sigma virus Drosophila
16 melanogaster model system ; 2) how non living environmental factors ( specifically phosphorus enrichment ) can affect the replication of WNV ; 3) how mathematical models can improve our understanding of the replication cycle of clinically important RNA viruses ; and 4 ) how laboratory methods and field studies can be designed to detect emerging viruses in resource limited settings specifically in Venezuela during the ZIKV epidemic in 201 6
17 CHAPTER 2 D rosophila melanogaster AND SIGMA VIRUS: A MODEL SYSTEM FOR UNDERSTANDING THE PERSISTENCE OF VERTICALLY TRANSMITTED ARBOVIRUSES density (Bjornstad et al. 2 002), life history and genotype, by the transmission mechanism of the parasite, and by the impact of the parasite upon its host. Virulence the reduction of host fitness by a parasite may impose a cost on the parasite as well as the host, if decreased h ost lifespan or fecundity results in decreased rates of parasite transmission. However, increased virulence may also benefit the parasite, as virulence is often closely linked to the diversion of host resources for parasite reproduction. The costs and bene fits of virulence to the parasite are thus expected to balance each other to some extent, giving rise to an optimal (for the parasite) intermediate level of virulence ( i.e ., the tradeoff theory for the evolution of virulence ) (Ebert 1999; Combes 2001; Aliz on & van Baalen, 2005). There are other factors, of course, that can influence virulence (Ebert & Bull, 2003), but in some systems the nature of transmission makes costs for the host likely to be costs for the parasite as well. This is inevitable when tran smission is vertical across host generations. Virulence is particularly costly to parasites that are transmitted vertically (from parents to offspring) because parasite fitness is so closely tied to host fitness: hosts that die before reproduction doom the ir parasites as well, and if parasite reproduction involves passage through host eggs, reduced host fecundity also reduces parasite fitness. The persistence of vertically transmitted parasites that remain virulent despite their mode of transmission thus pr esents both selection act to reduce virulence, if it is costly to the parasite, too?). Observing such a system often leads to a working hypothesis of the existenc e of some form of cryptic horizontal
18 transmission ( i.e. transmission within generations rather than from parent to offspring (Mangin et al. 1995). Parasites that combine vertical transmission with horizontal transmission, involving both within and among species transmission, are not uncommon. Examples of viruses that include both vertical transmission within a dipteran population and horizontal transmission between dipteran vectors and human hosts include West Nile and yellow fever viruses (Mishra & Moury a, 2001; Goddard et al. 2003; Murray et al. 2010; Sall et al. 2010). Another factor that can affect persistence is diversity in the modes of vertical transmission. In sexual host species, vertical transmission can be uniparental (typically maternal) or bip arental (either parent can infect the offspring). A virulent parasite cannot persist by uniparental transmission alone; at least some horizontal transmission is required. In contrast, biparental inheritance permits persistence of virulent parasites with no horizontal transmission, at least if transmission efficiency is high relative to the cost of infection (Fine, 1975), and moreover changes the selective pressures on both the host and the parasite. Biparentally transmitted parasites have epidemiological an d evolutionary similarities to horizontally transmitted parasites (Fine, 1975). Systems with biparental transmission may provide an unusual opportunity to understand the ecology and evolution of virulence, because transmission to new host lineages occurs v ia sexual reproduction, which can be easier to document and to experimentally manipulate in these systems (which necessarily are structured by mating pairs) than in many modes of horizontal transmission. The sigma virus Drosophila melanogaster model syst em provides an ideal opportunity to investigate the mechanisms permitting parasite persistence without horizontal transmission. Sigma virus (Rhabdoviridae) is a vertically transmitted, virulent parasite that infects natural populations of D. melanogaster w orldwide (reviewed in Fleuriet, 1996). It has long been known
19 to be transmitted biparentally (L'Hritier, 1958; L'Hritier, 1970). Sigma is virulent: infected hosts express a variety of symptoms consistent with lowered fitness, including a decrease in fecu ndity (Fleuriet, 1981). This is the kind of fitness cost in the host that also imposes a potential fitness cost on the parasite. Sigma virus infections have been observed in several other species of Drosophila (e.g., D. affinis and D. Athabasca ) (Williamso n, 1961; Flix et al. 1971b; Longdon et al. 2010), suggesting that this is a widespread host parasite syndrome. Our specific goal in this study is to ascertain if transmission efficiency and virulence as estimated from field and laboratory studies are cons istent, even broadly, with levels of prevalence observed in nature. To achieve this goal, we develop a dynamic, deterministic, discrete generation model that incorporates sex specific transmission efficiency and cost of infection with respect to fecundity for projecting disease prevalence across time. Using data from a natural population of flies, we estimate prevalence, track transmission efficiency across generations and lineages, and also quantify the effect of sigma infection on female fecundity (a key aspect of virulence). We use the results of these lab studies to parameterize the model, and then compare the equilibrium prevalence predicted by the model to the prevalence measured in our field samples. In the discussion, we sketch future extensions of t he modeling framework that may be needed to account for some of the empirical patterns we observed in the lab studies. Modeling the D ynamics of Sigma virus al. (1999), but differs from the former in that we do not distinguish among hosts differing in that we incorporate costs to female flies of being infected, and also p ermit a broader range of transmission efficiency when both members of a mating pair are infected. Given the current prevalence ( pt the fraction of flies that are infected, which is assumed to be independent of sex),
20 the prevalence in the next generation ( pt+ 1) is the ratio of the per capita proportion of infected offspring to the per capita production of all offspring. Infected offspring are produced in three ways: 1) an infected female with fecundity ni and transmission efficiency eF mates with an uninfec ted male; 2) an uninfected female with fecundity nu mates with an infected male with transmission efficiency eM ; or 3) two infected flies mate with fecundity ni and transmission efficiency eB (the probability that an offspring of two infected parents is in fected). We assume that infection affects only female fecundity, resulting in an asymmetry in the equations below with respect to male and female transmission parameters. We further assume that there is no sex specific effect of infection on survivorship, so infected, fertilized eggs have an equal chance of entering the mating pool, regardless of their sex. We also assume that mating is random with respect to infection status: therefore, the probability that a randomly chosen mating pair consists of an infe cted and an uninfected fly is pt (1 pt ), and the probability of both parents being infected is pt The per capita number of infected eggs is the sum of products of mating probabilities, fecundities and transmission efficiencies for events producing infect ed offspring, which is [ pt (1 pt )( nieF + nueM )+ pt 2 nieB ]. The per capita number of eggs produced is the weighted average of the uninfected and infected fecundities [(1 pt ) nu +ptni ]. Taking the ratio of these quantities gives the Equation 2 1 ( 2 1) whe re q = n i /n u is the fecundity of infected relative to uninfected females, which could include differential oviposition rate and even egg viability (as long as egg viability depends on infection of the female, not the eggs). As q decreases, the cost of the virus to the host (virulence) increases. We analyze the properties of this model in the Results section We then relate this
21 model to data, including estimates of prevalence, sex specific transmission efficiency, and the impact of infection on female fec undity. Methods Our focal population for the study consisted of a population of D. melanogaster in north central Georgia, USA. To measure the prevalence of sigma virus, we sampled natural populations of D. melanogaster from six sites along US Hwy. 129/441 ( Table 2 1 ). Table 2 1. Location of collection sites. All sites are along U.S. Highway 441/129 between Eatonton and Athens, GA. Distance is indicated in miles along the highway from the southernmost site (site 1). Site ID Distance (mi) Coordinates 6 50 33 94.671' N 5 35 33 49.463' N 4 23 33 46.528' N 3 22 33 46.264' N 2 20 33 44.447' N 1 0 33 24.578' N We collected five times during the summer of 2009: June 12, June 24, July 10, July 24 and September 18. Twenty four hours before sampling, w e placed 3 baits containing fruit and yeast at each site to attract D. melanogaster We then swept for all visible flies using Drosophila fly nets (Bioquip). We transferred the animals to plastic shell vials containing standard molasses cornmeal medium, and brought them to the University of Florida in Gainesville.All animals were assigned to individual vials in the laboratory within 72 hours of the collection time.
22 Because exposure to CO 2 is lethal to flies that are infected with sigma virus (L'Hritier & Teissier, 1945), animals were anesthetized using cold treatment, a standard alternative to the more common procedure of CO 2 anesthesia. We first placed the insects in empty vials in ice for up to 5 minutes. Once the insects stopped moving, we transferred them to custom made metal blocks which had been chilled for >1 hour at 0C and covered with moist KimWipes. We discarded all species except D. melanogaster, which we placed individually in Drosophila plastic vials with standard molasses cornmeal medium. CO 2 A ssay for S igma V irus I nfection Infection status was determined by CO 2 assay. Each fly to be assayed was placed in an empty vial, which was flooded with CO 2 for 5 minutes, and then returned to ambient oxygen and CO 2 levels for 15 minutes. Flies that retu rned to normal activity levels (walking and flying) were scored as uninfected; flies that were either dead or paralyzed (unable to walk or fly but still moving) after CO 2 exposure were scored as infected. Because the assay kills or paralyzes infected flies flie s were allowed to reproduce before they were assayed for infection. Transmission A ssays Field collected adults will hereafter be referred to as the parental ( P ) generation. By holding these flies in individual vials for five days prior to CO 2 assay ( which is lethal to infected animals), we were able to rear their progeny (the F 1 ), assay the F 1 for sigma virus infection, and thus estimate trans mission efficiency ( Figure 2 1 ). We assayed the F 1 by CO 2 as described above to determine their infection stat us, 10 12 days after establishment of the vials. Transmission was estimated as the fraction of F 1 offspring produced by an infected parent that were infected (calculated separately for male and female parents).
23 Females captured from the field are usually inseminated, and thus can produce viable offspring without mating in the laboratory. We therefore placed the field collected females individually in vials and allowed them to oviposit for five days. The infection status of the sires of these F 1 was thus un known. To determine rates of transmission by field collected males, we crossed each male with two uninfected virgin females from the effectively isogenic 58 stock (Wayne et al. 2007), and left them in the vial together for five days. F 1 progeny were collec ted on day 14 after the vial was established. We determined the single generation transmission efficiency of males and of females obtained from three collection trips: June 24 (trip 2), July 10 (trip 3), and July 24 (trip 4), 2009. Figure 2 1 Procedure for measuring prevalence and transmission. This diagram shows how prevalence and transmission were measured, in females (left side) and males (right side). Heavy arrows denote movement of animals; thin arrows denote passage of
24 time. P denotes the generatio n captured from peach stands (i.e., from the field). The F1 are the progeny of the fi eld caught animals and the F2 are their grand progeny. Prevalence and P to F1 transmission were measured for both female and mal e P animals The progeny of these 11 isomal e lines (all of which were from trip 4) are denoted as F1*. Second, P to F1* includes only virgin F1* animals (as indicated by the V) from day 15 (in contrast to P to F1 transmission efficiency, which included non virgin progeny from days 10 14 ) Our stati stical model tests for the effects of sex and trip on transmission efficiency, using the standard linear modeling function (lm) in R (R Development Core Team 2008). We treated trip as a factor and first determined the statistical significance of the intera ction between sex and trip. We then reduced the model by eliminating the non significant interaction between sex and trip ( P= 0.237) to keep an additive model of the effects of sex and trip. We checked our data and the residuals graphically for deviations from the assumptions of linear models (normality, homogeneity of variance: plot.lm function in R: R Development Core Team 2008), and observed no large deviations. Multigeneration Patrilineal T ransmission As noted in the introduction, given the complete la ck of evidence for any horizontal transmission, biparental transmission is required for persistence of the sigma virus in natural populations. To further characterize biparental transmission, we monitored patrilineal transmission efficiency across three ge nerations originally sired by infected male P flies from the July 24 th trip (trip 4; Figure 2 1 ). A subset of eleven infected males assayed for transmission as described above were used to create eleven independent "isomale" lines. After the original tran smission assay using F 1 progeny collected on day 14, additional unmated progeny (denoted as F 1 virgins) were collected twelve hours later (10 13 F 1 virgins per vial). We next set up single pair matings by pairing each F 1 fly with a single uninfected vir gin of the opposite sex from the effectively isogenic 58 stock. Five days after establishment of the F 1 vials, the parents
25 were removed and assayed for infection using the CO 2 assay. Only F 1 vials of infected parents were retained. The percentage of F 1 flies infected is referred to as the P to F 1 transmission efficiency. When the offspring (the F 2 ) flies eclosed 12 14 days later, we determined the infection status of 10 haphazardly chosen F 2 flies from each vial using CO 2 assay. The sexes of the F 2 pro geny were not recorded. The percent infected F 2 offspring is referred to as the transmission efficiency from the F 1 to the F 2 generation. We analyzed the data from these experiments using three separate paired t tests; all comparisons were made within ea ch of the 11 lineages. We first determined whether the probability of acquiring sigma virus infection from the P males was equal for both male and female F 1 flies. We then compared the transmission efficiency ( F 1 to F 2 ) from infected F 1 female to that of infected F 1 male flies, to test for sex specific differences in transmission efficiency in animals who acquired infection from the sire (male parent). We finally compared the transmission efficiency of P males to that of their male F 1 progeny, to estima te changes in patrilineal efficiency from one generation to the next. Effect of Sigma V irus on Female F ecundity With vertical transmission, impairment of female fecundity hampers the persistence of the parasite. Accordingly we measured the effect of sigma virus on female fecundity (number of eggs per female). We compared 4 infected isofemale lines to 4 uninfected isofemale lines of D.melanogaster collected near Athens, GA in the summer of 2007. Isofemale lines are created by placing a single, inseminated w ild caught female in a vial and propagating her offspring en masse. Thus, these lines represent a small random sample from the population, and are expected to be genetically distinct from one another. We conducted four blocks of measurements, each with fo ur replicate bottles (total: 4 lines x 2 infection statuses x 4 blocks x 4 replicate vials/block = 128 vials). For two generations prior to assay, flies were propagated in vials set up at constant
26 density (5 females + 5 uninfected males per vial), oviposit ing for five days on standard cornmeal molasses medium. For the assay, individual female flies, four days post eclosion, were placed in inverted milk bottles over small Petri dishes containing cornmeal molasses food. After six hours, we counted the number of eggs that were laid by each female. While performance of lines within treatments was consistent across blocks, the mean of treatment groups varied between blocks, precluding a traditional ANOVA treatment. Relating the Model to E xperiments We parameteri zed the prevalence model ( Equation 1 ) with the fecundities and transmission efficiency measured in the lab to predict sigma prevalence at the non zero equilibrium. To obtain point estimates and confidence intervals for transmission efficiencies, we used ma ximum likelihood estimation (bbmle: Tools for general maximum likelihood estimation. R package version 0.9.5.1; http://CRAN.R project.org/package=bbmle (Bolker, 2008) to fit beta distributions (p arameterized in terms of mean and variance parameters; Morris, 1997) to the proportions of offspring infected from crosses with either an infected female (for e F ) or male (for e M ). We based our model parameters on the estimates and confidence intervals (CI s) for the means of these distributions. We used the same method to derive a point estimate and confidence intervals for the observed prevalence. We similarly used maximum likelihood, but based on a negative binomial distribution, to derive point estimates and CIs for the fecundity of infected and uninfected flies. We calculated the expected prevalence and rate of increase at low prevalence by substituting the mean observed parameter values in the expression for the equilibrium prevalence. We used nonparame tric bootstrapping ( n = 10 7 ) using all observed values of fecundity and transmission efficiency to find confidence intervals on these quantities.
27 Results Transmission E fficiency We measured transmission efficiency (proportion of offspring which were infect ed) for males and females Both sexes transmitted the virus, but with different efficacy. Males transmitted sigma virus to a mean of 51% ( 8% std. error) of their offspring, while females transmitted sigma virus to 95% ( 5% std. error) of their offspring ( Table 2 2 ) Table 2 2. ANOVA results for single generation transmission efficiency. There was a statistically significant effect of sex on the transmission efficiency of sigma virus (females transmitted at a higher rate than males). Source Df Sum Sq Mean Sq P Sex 1 1.507 1.507 < 0.0001 Trip 2 0.136 0.068 0.0081 Error 37 0.456 0.012 Multigeneration, patrilineal propagation of sigma virus was possible for multiple generations in the lab, using "isomale" lines (each of which was propagated from an infe cted sire obtained in the field ( P generation) mated with a single virgin female from the effectively isogenic laboratory stock 58 ). All 11 P generation males transmitted sigma virus to some offspring (each representing new female lineages; Figure 1 3, le ft panel). Sons and daughters ( F 1 flies) were equally likely to become infected with sigma virus when the sire was infected (paired t test; P = 0.79). Intriguingly, daughters of infected males had a higher transmission efficiency than did sons ( F 1 to F 2 transmission efficiency; Figure 1 3, middle panel; paired t test, P = 0.032). Moreover, transmission efficiency from sire to offspring was significantly lower in the F 1 to F 2 generation than in the P to F 1 generation (Figure 1 3, right panel; paired t te st, P = 1.149 x 10 6 ).
28 Prevalence Prevalence was estimated five times in the summer of 2009, for flies collected from six sites (data presented in Table 2 3 ). The average prevalence of sigma virus was 28% (Figure 1 2 ). Prevalence did not differ significa ntly by sex, by site, or by trip; nor were any of the interaction terms significant ( Table 2 4 ). However, prevalence did vary among collection trips, ranging from zero infected flies found (3 out of the 20 samples that yielded at least one fly of each sex; Table 2 3 ), to 71% prevalence (July 24, site 4; Table 2 3 ). There are no obvious trends predicting prevalence ( Figure 1 2 ).
29 Table 2 3 Virus prevalence and number of flies per collection. Each site is labeled according to its distance from the southernmost site. We measured prevalence as the proportion of flies infected with sigma virus for each date and site combination. Dashes indicate that the number of flies of at least one sex was zero. Date Distance (mi) Uninfected F Infected F Uninfected M Infected M Prevalence 12 Jun 0 13 2 14 3 0.16 20 6 0 5 3 0.21 22 0 0 1 1 -23 0 0 1 0 -35 14 4 25 4 0.17 50 13 0 8 0 0 24 Jun 0 3 0 0 3 0.5 20 0 1 1 1 0.67 22 0 0 0 0 -23 2 0 2 0 0 35 0 0 0 0 -50 2 3 4 2 0.46 10 Jul 0 3 3 6 0 0.25 20 2 1 3 0 0.17 22 1 1 2 0 0.25 23 2 0 4 0 0 35 0 0 0 0 -50 3 5 4 2 0.5 24 Jul 0 6 3 11 2 0.23 20 8 3 2 1 0.29 22 0 0 0 0 -23 0 2 2 3 0.71 35 0 0 0 0 -50 4 0 13 5 0.23 18 Sep 0 0 0 0 0 -20 7 1 2 0 0.1 22 1 0 4 3 0.38 23 1 0 0 0 -35 0 0 0 0 -50 17 14 37 21 0.4
30 Figure 2 2. Prevalence of sigma virus in North central Georgia. The p roportion of D. melanogaster that are infected with sigma virus is shown ( 2 SE). Variance in prevalence at the 35 mile site (* point without error bars) could not be estimated, because we collected D. melanogaster during only one of the collection trips to this site. Sample size (number of flies) is indicated next to each point. The mean proportion of flies infected with sigma virus was 0.28 0.1 Tabl e 2 4. ANOVA results for the prevalence of sigma virus in the field. There were no statistically signifi cant effects of date, site, or sex on the prevalence of sigma virus, nor any significant interaction terms Source Df Sum Sq Mean Sq P Trip 4 2.71 0.678 0.17 Site 5 0.38 0.075 0.95 Sex 1 0.23 0.228 0.43 Trip x Site 13 4.62 0.355 0.48 Trip x Sex 4 0.99 0.247 0.6 Site x Sex 5 0.31 0.062 0.96 Residuals 10 3.42 0.342 Virulence and Cost to Fecundity Four infected and four uninfected isofemale lines were scored for fecundity. Females from the infected lines laid a mean of 115 eggs each ( 58; standard error of the mean), while females from the uninfected lines laid a mean of 169 eggs each ( 85; standard error of the mean). The performance of individual lines was consistent across the four assay blocks.
31 Generally speaking, lines that were infected had l ower fecundity than un infected lines ( Figure 1 3 ). However, in one out four blocks, infected and uninfected lines performed similarly. Figure 2 3 Effect of infection status on fecundity. Boxes represent the means of lines within treatments (i.e., uninf ected, white boxes; vs. infected, gray boxes) for each of the four blocks. Error bars represent two standard deviations from the mean of each group of four lines. Circles represent outlier lines within treatments (i.e., lines more than 1.5 times the group interquartile range), and are shaded to indicate infection status as described above. Fecundity is generally higher in uninfected lines than in infected lines, except in the third replicate.
32 Figure 2 4 Transmission efficiency of sigma virus (single gen eration). Mean transmission efficiency by females (circles) and males (squares), calculated from the first ten F1 offspring to eclose ( 2 SE). Sample sizes (number of flies) are indicated next to each point. Flies from collection trip 1 were not assayed f or transmission efficiency; while transmission efficiency wasassayed from collection trip 5, a different protocol was used such that the data were not comparable. Interestingly, although it was previously been stated that the sons of infected males do not transmit the virus (L'Hritier, 1970; Fleuriet, 1981), we have demonstrated that multiple generations of patrilineal transmission are possible, at least for virus and flies from the Athens, GA population. However, transmission frequency is lower from the second to third generation of patrilineal transmission than from the first to the second ( Figure 2 5 ).
33 Figure 2 5 Patrilineal transmission of sigma virus. Transmission efficiency (the proportion of the first 10 offspring that were infected) is illustra ted on the Y axis, which has the same range in all panels. Thin black lines correspond to each of the 11 isomale lineages. The grand means ( 2 SE) of the lines are shown in heavy grey lines. The left panel represents transmission efficiency between the P and F 1 generations, testing the hypothesis that patrilineal transmission is equal in sons (open squares) and daughters (open circles; P = 0.79). The middle panel compares transmission efficiency from the F 1 to the F 2 specifically between sons (open squ ares) and daughters (open circles) of infected sires from the field; transmission is higher from F 1 daughters than F 1 sons ( P = 0.032). Finally, the right panel represents the decrease in transmission efficiency (open diamonds: combining male and female offspring) from the first to the second generation (i.e., P to F 1 relative to F 1 to F 2 ; P < 0.0001). A more complex model than (1) above would be needed to encompass this intriguing transgenerational effect on transmission, as well as to incorporate the distinction among classes of hosts with different patterns of within host dynamics first explored consideration of distinct patterns of within host dynamics may be needed to fully explain our results. We also have demonstrated that infection acquired solely from the sire persists for at least two generations in the lab, although transmission efficiencies decline with each generation. The 1 1 1 1
34 persistence of paternally acquired virus is particularly important because, as reiterated by our model results and previously demonstrated by others, biparental inheritance is required for persistence of vertically transmitted parasites (Fine, 1975). One biological consequence of such male transmission is that given occasional interspecific hybridiza tion, the virus could be transmitted across species boundaries. As noted above, sigma virus is currently known to infect several species of Drosophila (Williamson, 1961; Flix et al. 1971b; Longdon et al. 2010). It would be interesting to have a fuller un derstanding of the phylogenetic scope of sigma virus among closely related species of flies, coupled with a deeper understanding of patterns of genetic variation in the virus within single host species. Future studies will incorporate population genetic co mponents that include genetic variation of both the host and the parasite. Together with the ecological factors sketched above, genetic differences could play a strong role in explaining differences among sites in sigma virus prevalence. For example, viral variation that causes detectable differences in transmission and/or infectivity is well known (Goldstein, 1949; Brun & Plus, 1980; Wilfert & Jiggins, 2010a). Host genetic variation for male transmission and for resistance is also well documented (L'Hriti er, 1970; Gay, 1978; Brun & Plus, 1980; Fleuriet,1996; Wayne et al. 1996; Bangham et al. 2007; Bangham et al. 2008; Carpenter et al. 2009; Wilfert & Jiggins, 2010b). However, data on allele frequencies in virus and host, as well as potential host parasite interactions (or lack thereof; Wilfert & Jiggins, 2010b), are required to make meaningful progress in this direction. Discussion Our point estimate for prevalence based on lab estimates of sex specific transmission and cost of infection on fecundity is con sistent with prevalences observed in the field; however, lab estimated parameters also produce very wide confidence intervals, and the equilibrial prevalence they predict tends to exceed observed field values (Figure 2 6 )
35 Figure 2 6. Equilibrium prevalen ce of Sigma virus as a function of female transmission efficiency and male transmission efficiency. The open circle is the prevalence predicted using the point estimates of the parameters and equations. The filled circle represents the observed mean preval ence and mean female transmission efficiency; error bars represent 95% binomial confidence intervals at the observed prevalence. The dashed curve shows the expected prevalence as a function of female transmission, given that male transmission is equal to i ts point estimate (0.5). Regardless, the overlap suggests that our model provides a sensible springboard for more detailed investigations of virulence and transmission in the sigma virus Drosophila system, including a wider range of known and suspected b iological factors. Our estimates of prevalence and transmission efficiency were obtained from the same population of flies within a single season, and our estimates of virulence were obtained from the same population (albeit two years earlier). We found t hat sigma infects a mean of 28% of D. melanogaster individuals in north central Georgia (approximate latitude 33.9 N), with no differences in prevalence between males and females. The prevalence of sigma virus in 2009 (when flies were collected for our lab study) was higher than in 2005, when it infected only 6.2% of the population of D. melanogaster in Athens, GA (Carpenter et al. 2007); or in 2006, when it infected 7% of the population (Wayne
36 unpublished data). Thus, prevalence in these natural populations varies greatly from season to season. While it is possible that prevalence in the 2009 season was anomalous, note that transmission efficiencies were obtained from the same population for the same season (though virulence estimates are from lines collect ed in 2007), so the data are at least self consistent. Our estimate of virulence (32%), while high, is not grossly dissimilar from previous estimates of virulence in terms of egg to adult viability. For unstabilized females, the decrease in fecundity has b een estimated at 19% ( 4.6%), and for stabilized females, at 10% ( 2.4%; Fleuriet, 1981). It is interesting to place our field estimates of prevalence into a broader geographical context. The prevalence of sigma virus in our study area, which is among the southernmost sites s tudied to date (33.9 N), is also among the highest yet recorded, at 25%; though again prevalence varies among years (6.2 % in 2005; Carpenter et al. 2007); 7% in 2006 (Wayne, unpublished data). In Athens, Greece (37.9 N), sigma was found in 14.9% of the po pulation (Carpenter et al. 2007); in Galicia, Spain (42.66 N), prevalence was 4.3% (Carpenter et al. 2007); in Languedoc, France (43.7 N), prevalence ranged from 10 20% (Fleuriet, 1976; Fleuriet, 1980; Fleuriet, 1996) ; in Ithaca, NY (47.4 N), prevalence was a mere 1.7% (Yampolsky et al. 1999); and finally in Essex, UK (51.79 N), prevalence was 7.0%, though prevalence was zero in nearby Kent, UK (Carpenter et al. 2007). Latitude is of course an imperfect proxy for climatic variables that presumably can exe rt an influence on prevalence, but inspecting this suite of studies suggests to us the existence of a latitudinal or climatic gradient in prevalence of sigma virus. (Note that Galicia is on average considerably wetter and cooler than both Languedoc and low land Greece.). One published outlier to this trend comes from Clermont, FL (Apshawa Road; 28.61 N), the southernmost site, where prevalence was estimated at 1.5% (Carpenter et al. 2007). There are reasons to believe that prevalence may be low for several reasons distinct to that study. First, it was estimated early in
37 the season (March 2005); secon d, D. melanogaster is locally rare in the a rea; and third, the viral isolate was quite distinct from all other samples sequenced (Carpenter et al. 2007; Longdon et al. 2010), and so might well have distinct epidemiological properties. Our analysis of the m odel assumed that a population had reached equilibrium, and that fitness costs of parasitism are fixed. Natural populations are likely to be strongly disequilibrial. Fly populations are likely to fluctuate greatly in numbers, in response to variation in c limate, fluctuations in resource availability, and other factors, and may go extinct and become replenished by re colonization. If sigma virus is lost by chance from a population at low numbers, when it recolonizes there will be a lag before it reaches equ ilibrium, and during this lag prevalence will be less than the local equilibrium. We measured fitness costs of parasitism to female fecundity in the lab, but it is plausible that additional costs could be incurred in natural conditions when females are ex posed to a wide range of stressors. This too would tend to depress viral prevalence. For example, variation in prevalence in Mexico City might be explained by decreased overwintering survival of infected relative to uninfected flies and thus reduced freque ncy of infection of founders of the following spring population (Flix et al. 1971a), combined with low dispersal among populations. Likewise, male mating success might be more strongly affected by infection in the field than in the lab, and this could lea d to low infection in sparse populations (as in the Clermont, FL population noted above). We note also that if there is spatial or temporal variation in R (or q ), and local populations equilibrate in prevalence rapidly in response to such variation, then b inequality from the concave down shape of (5) where R > 1, prevalence averaged across sites will be less than the prevalence estima ted from averaged values of the parameters (Inouye, 2005). Without empirical estimates for his model parameters, L a broad range of plausible values. He did, however, include "stability" in his model, where stability
38 is defined as a persistent level of infection within female hosts, including near certainty of transmission to offsprin g. In contrast, we chose to model sex specific transmission efficiencies and associated errors, as informed by our empirical data, rather than including stabilization status per se Interestingly, we rarely observed female transmission efficiency of 100%. Similar to the Yampolsky et al. (1999), our model shows that if transmission by one parent is less than 100%, at least some transmission by the other parent is required for persistence (eq. ). Our estimates of transmissi on efficiency are similar to those of Yampolsky et al. (1999), who inferred that the transmission efficiency for the two sexes must be around 0.67 based on the rate of the spread of infection within their experimental populations. Our estimates range from just over 0.5 (for males) to 0.95 (for females), and so bracket this value. As previously noted (L'Hritier, 1970), virus acquired solely from the sire is transmitted at lower rates by daughters than is virus acquired from the dam (Figure 2 5 ).
39 CHAPTER 3 EFFECTS OF PHOSPHORUS ON THE REPLICATION OF WEST NILE VIRUS IN CELL CULTURE West Nile v irus is a positive sense RNA virus belonging to the genus Flavivirus of the family Flaviviridae which contains more than 100 species of viruses Clinically important members of this family include vector borne viruses and others which are transmitted through direct contact. such as Yellow Fever virus (YFV), Dengue virus (DENV), Japanese Encephalitis virus (JEV) and Zika virus (ZIKV) are transmitted by mosquitos Clini cal symptoms of infection with these viruses include febrile illness and in some cases neurological disease (Knipe 2001) WNV is the most widespread arbovirus in the world; it is transmitted by several species of mosquitos in North America (Turell 2005) a nd it has been detected in least 100 species of wild and domestic animals (Root 2013 ) In a recent study, Crowder et al (2013) investigated the relative effects of rainfall, host abundance and land use type on the prevalence of West Nile Virus. They found a strong positive correlation between land use type and infec tion rates of humans and horses: interestingly, land use type was a better predictor of WNV abundance than were the other factors, even when controlling for host density. Nutrient enrichment br ings an increase in nitrogen, phosphorus and other macronutrients. The concentration of phosphorus [P] correlates positively with the concentration of RNA in tissues and with per unit mass metabolic demands (Sterner 2002 ). Phosphate is an abundant componen t of agricultural fertilizers and is a rate limiting nutrient in many ecosystems: a small addition of P relative to nitrogen leads to large increases in growth rate. Given the stoichiometry of cells and viruses and the high per unit mass concentration of p hosphorus in viruses, it is reasonable to expect that virus replication could be phosphorus limited. T he effects limiting nutrients such as phosphorus can affect the ass embly and maturation of virus particles within a
40 host. Phosphorus is important for intracellular regulation of pH, thus affecting the maturation process of virus particles. Investigating how phosphorus might mediate the strength of the relationship between cells and virus could shed light on how nutrient enrichment might affect vector competence. Methods All experiments in this chapter were conducted in the Biosafety Level 3 (BSL 3) laboratory at the Florida Medical Entomology Laboratories at the Universit y of Florida campus in Vero Beach, FL following established procedures Preparation and filter sterilization of cell culture media was conducted in the Biosafety Level 2 (BSL 2 ) laboratory. T he protocols for measuring phosphorus and adjusting pH of cell fr ee and virus free cell culture media were optimized at the Ecosystem Ecology Laboratory at the Department of Biology at University of Florida campus in Gainesville, FL. Cells and V irus S tocks Vero cells (ATCC, passage 191) derived from the kidney of Chlo rocebus sabaeus (African green monkey), an adherent cell line that forms a monolayer and is permissive to infection with WNV, were maintained at 35 o C in a water jacketed incubator using s 15 medium (1X) L 15) does not require supplementation with CO 2 Vero cells grown without CO 2 supplementation usually take 1 2 days longer than cells grown with CO 2 supplementation to reach confluency (G. Blohm unpublished data). Established procedures for working with Ve ro cells were adjusted accordingly throughout the experiments. Virus stocks for these experiments were isolate d from Culex pipiens quinquefasciatus in Indian River County, FL, in 2006 (isolate # 2186 obtained by plaque purification ) which had been stored at 80 o C. The titer of the isolate was obtained immediately upon thawing, before the
41 start of all experiments, and separate aliquots were stored to avoid RNA degradation by freeze thawing. To obtain titers, p laque assay s were conducted following establishe d procedures with the following modifications: agarose overlay was mixed with 2X L 15 media prepared from powdered L 15 ; infection was allowed to progress for 1 week, as countable plaques were observed 5 7 days post infection (as opposed to 2 3 days post i nfection normally observed in CO 2 supplemented experiments ). Titer of the laboratory stock was consistent with previously recorded tit er of the same stock : the working laboratory stock of WNV contained 10 7 plaque forming units p er mL, suggesting that degra dation of viral RNA had not occurred since storage. I n the absence of CO 2 supplementation, the laboratory stocks of WNV produced cytopathic effects (CPE) between 5 and 8 days post infection. The onset of CPE (cell r ounding lysis, cell membrane shrinking and detachment from monolayer) occurred 1 2 days later than recorded in previous studies with CO 2 supplementation. Otherwise all characteristics of the virus are similar to those described in previous studies conducted in a CO 2 supplemented environment (S. Richards and G. Blohm unpublished data) Phosphorus M anipulations C ells and virus were exposed to changes in phosphorus by adjusting the concentration of total phosph orus salts in L 15 medium, which contains two forms of phosphorus salts: 0.06 g/L anhydr ous monobasic potassium phosphate and 0.19 g/L anhydrous dibasic sodium phosphate ( Appendix A: Table 1 ). We adjusted the concentration of sodium phosphate for our experiments. Because o smotic potential and pH are affected when changing the concentration of salts in cell culture media, we adjusted the pH of the cell culture media with cell culture grade liquid sodium and with HCl This ensured that at all three phosphorus levels any changes in cell metabolism and virus attachment rates could be attributed to changes in sodium phosphate (NaPO 4 ) only
42 R emoval of phosphorus from the cell culture medium requires chelation of phosphate salts, causing the formation of precipitates that are cytotoxic To prevent the formation of cytotoxic precipitates, w e r equested custom made L 15 medium without the addition of sodium phosphate We then adjusted the concentration of phosphate by additi on, rather than removal, of phosphate salts To achieve phosphorus levels that are close to those of commercially available L 15 media, we added 0.06 g to each L of custom made L 15 and adjusted pH accordingly The highest concentration of phosphorus that we could reach without irreversibly changing the pH of the media wa s at approximately 0.19 g /L of anhydrous dibasic sodium phosphate A t this high concentration of phosphorus we observed a decrease in pH (from 7 .6 to 5 .5 ) that required the addition of 5 10 mL sodium bicarbonate. Once the pH reached desired levels, w e f ilter sterilized the media using a 0.22M sterile polyethersulfone membrane filter (Millipore Billerica, MA ) and verified total phosphorus concentration by persulfate digestion as described below To measure total phosphorus in the cell culture media, we followed a modified version of the persulfate digestion protocol that is commonly used for quantification of total (organic and inorganic) phosphorus during waste water monitoring surveys Briefly, 1:10 triplicate serial dilutions of NaPO 4 were prepared wit h custom made L 15 media (for negative controls, distilled water was used as a matrix), ranging from 0.001 g/L to 1.0 g/L in a clear 96 well plate compatible with microplate reader. Sulfuric acid (11 N) and ammonium molybdate antimony potassium tartrate (8 g/L) were added to the samples, which are then sealed to block the entrance of CO 2 and incubate d for 30 min in the dark. Abso rbance readings were then taken for determination of detection limits and standard curve for total mg P/L.
43 After verification of p hosphorus treatment levels and pH adjustment, c ell culture medium was then supplemented with 10% heat inactivated fetal bovine serum (FBS) ( Gaithersburg, MD ), 1 % of 200 mM L Glutamine ( ), 5 % of Amphotericin aithersburg, MD) and 1% P enicillin S treptomycin N after bringing into the BSL 3 laboratory at FMEL Effects of Phosphorus on R ate of C ell D ivision We estimated the effect of phosphorus on the population level growth ra te of Vero cells Because we were interested in obtaining an estimate of the effect size of phosphorus on cell population size, as well as the shape of the trajectory of the cell population under different phosphorus environments, we conducted two experimen ts (end point and time series). In the end point experiment, w e seeded eighteen T 25 flasks (six replicates for per phosphorus level) at 10 4 cells per mL. Six days after seeding the flasks, we suspended the cells D) a nd counted the number of cells per mL: t wo 100L samples of the cell homogenate were obtained for estimation of the cell population size. L ive cells were counted using a hemacytometer and 0.4% trypan blue solution ( Gaithersburg, MD ), a cell impe rmeable vital stain which is taken up only by cells that are not viable, thus allow ing the distinction between live and dead cells. In the time series experiment, we seeded twenty four T 25 flasks with 10 4 Vero cells per mL (8 flasks for each of the three phosphorus levels) and destructively sampled the cells using trypsin EDTA at four time points for each level of phosphorus : 24, 48, 96, and 168 hours (7 days) post seeding Cell counts were conducted as indicated above using trypan blue. Effects of Phosp horus on Rate of V irus R eplication To determine the effects of phosphorus on the rate of replication of WNV, we conducted two experiments ( time series and end point ). In the time series experiment we measured the
4 4 growth trajectory of the virus through time by real time RT PCR at three phosphorus levels. In the second experiment we quantified the virus by plaque assay at two levels of phosphorus (Low and High) after a 72 hour growth period. To control for indirect effects via changes in the rate of cell div ision, we varied the seeding density of the cells and allowed them to become 90% confluent before infecting with WNV in both experiments C ells in the Low P treatment were seeded at 10 5 cells per mL, while cells in the Medium P treatment were seeded at 5 x 10 4 cells per mL and cells at the High P treatment were seeded at 10 4 cells per mL. This allowed all wells to reach confluency at approximately the same time. The wells were inoculated with WNV at an MOI (multiplicity of infection) of 1. The supernatant w as then sampled at 8 hour intervals for 72 hours and the virus population was quantified by real time RT PCR. In the end point experiment, we seeded six well plates at 5x10 4 Vero cells per mL for the Medium P concentration and at 1x10 4 virus particles per mL for the High P concentration, at 10 replicates per treatment (20 wells + 4 negative controls that were not infected with virus). When the cells were confluent, we infected each well with 200L virus stock. After a two hour incubation, we removed the ino culum and refreshed the cell culture media at corresponding P levels. On the 7 th day post infection, we sampled the supernatant and quantified the number of virus particles by plaque assay (methods described above). Results Adjusting the concentration of phosphorus in the cell culture media led to changes in pH. When compensating for the changes in pH, we were able to isolate the effects of phosphorus alone on the cells and virus populations in cell culture. Cells responded to an increase in phosphorus lev els; however the number of virus par ticles in cell culture did not.
45 Phosphorus M anipulations Concentration of total phosphorus in cell culture media varied within 25% of the average for each treatment. At higher levels of phosphorus, the variance in concen tration was greater than at lower levels (Figure 3 1) Phosphorus level Figure 3 1 Effectiveness of phosphorus treatments. The concentration of total (inorganic and organic) phosph orus for each treatment level. The average concentration of phosphorus i n the L ow (open diamonds) treatment was 15.1 mg/L; Medium (gray diamonds) was 58.2 mg/L and High (black diamonds) was 119.5 mg/L. Effects of P hosphorus on R ate of C ell D ivision The average rate of cell division increased nonlinearly with an increase in t he concentration of phosphorus A two fold increase in the concentration of phosphorus led to a 50 65% increase in cell population size after 6 days (Figures 3 2 and 3 3). Effects of phosphorus on cell population size were more variable at higher levels of phosphorus. Due to a low sample size, the shape of the growth trajectory of the cell population at each phosphorus level cannot be determined (Figure 3 3); however the results of both experiments are consistent in that an
46 increase in the concentration of total phosphorus in the cell culture media leads to an increase in the rate of cell division during a six day period. Figure 3 2. Vero cell population size at three phosphorus levels six days after inoculation. Cells in Low P (open diamonds) conditions reached an average of 41 ( 3 std. error) cells per 0.0009 mL (4.6x10 4 cells/mL). Cells in Medium P (gray diamonds) conditions reached an average of 83 ( 4 std. error) cells per 0.0009 mL (9.2x10 4 cells/mL). Cells in High P (black diamonds) conditions re ached an average of 124 ( 10 std. error) cells per 0.0009 mL (1.1x10 5 cells/mL).
47 Figure 3 3. Effects of phosphorus on the growth trajectory of Vero cells. During the first three days after seeding, the number of cells in Low P, Medium P and High P is close to 0. In fact, the number of cells in the Low P treatment is slightly higher ( 147 cells/0.0009 mL) than in the High P treatment (112 cells/0.0009 mL). After six days, the average number of cells in the Low P treatm ent did not change; however the vari ance increased. The average number of cells in the High P treatment increased to 208 cells/0.0009 mL). Ef fects of Phosphorus on Rate of V irus R eplication There was no effect of phosphorus on the growth trajectory of the WNV population in cell culture (Figu re s 3 4 and 3 5 ) Results by real time RT PCR show that the virus population in the spent media remains constant for the first 20 hours. The exponential phase of the growth curve occurs at approximately 40 hours post infection, when the virus population in creases from approximately 10 4 virus particles per mL to an average of 10 6 virus particles per mL. The virus population then stabilizes at approximately 75 hours post infection, when cell lysis begins to
48 occur (G. Blohm, personal observation) and the virus is unable to continue replicating due to a depletion in the population of available host cells. Figure 3 4 Effect of phosphorus on West Nile Virus population size in Vero cells Number of plaque forming units per mL (determined by q RT PCR) of WNV iso late #2186 Low phosphorus is represented by the open circles; medium phosphorus is represented by gray circles, and high phosphorus is represented by black circles. The virus population was grown on confluent monolayers of Vero cells in 6 well plates
49 Figure 3 5. Effect of phosphorus on the number of plaque forming virus particles. Virus populations grown in the Medium P (gray diamonds) treatment reached similar numbers to those grown in the High P (black diamonds) treatment. There was no effect of phos phorus on the final population size of the virus after 7 days of incubation.
50 CHAPTER 4 MODELING THE REPLICATION OF FLAVIVIRUSES Flaviviruses ( Flaviviridae ) are single stranded RNA viruses that ar e approximately 12 kb in length, containing a capsid prot ein and a lipid envelope. At the cellular level, attachment and entry proceed by receptor mediated endocytosis (primarily clathrin mediated attachment) followed by uncoating in the cytoplasm, where replication and assembly take place. The virus particles then matur e in the endoplasmic reti culum and are released as either infectious or defective virus particles Direct measurement of each step in the virus replication process is often cost prohibitive. In cases where specific aspects of virus replication ca nnot be directly measured in the laboratory, it is useful to work with mathematical models and statistical tools for parameter estimation I n a study of Hepatitis C virus (HCV) Kumberger et al. (2016) model the within host dynamics of HCV using a system o f differential equations where the cells are distinguished according to their infection status (Uninfected, Infected, and Infectious), with viral entry, viral replication and viral export as rate parameters connecting the three state variables of the host cells. The model is used to predict cell to cell transmission in different types of tissue within a host. This approach proved effective at predicting the within host spread of HCV that had been measured in the literature: in studies where viral maturation was measured by electron microscopy, and where clinical measurements of within host dissemination of the virus were taken. In Padmanabhan et al. (2011), the molecular process of viral entry was the focus of the model. The authors model the kinetics of vir us entry into host cells by including CD81 expression as a determinant of viral entry. Cells resistant to infection due to reduced CD81 expression were included in the model, and as in the previous study, the mathematical model was able to predict results obtained in the literature.
51 In the section that follows, w e model the replication cycle of WNV following an approach that is similar to that of Kumberger 2016 with a few differences detailed below We focus specifically on the virus population and assume a single state for the cells. We model the growth trajectory of a population of WNV particles in cell culture as a continuous process wherein virus particles attach to the cell, replicate and mature inside the cell, and then are released into the supernat ant as infectious virus particles. The rationale for this approach was to divide the replication cycle of the virus into distinct categories that could be measured in the laboratory, thus providing a model that can be tested and improved with published dat a on WNV replication dynamics and within host proliferation Methods Model of WNV R eplication We divide the virus population into three sub populations (V, A, and R) according to their location and infectivity ( Figure 4 1 ) Upon contact with the host cell infectious virus particles V t bind to membrane receptors at time t and become incorporated via endocytosis at a rate The low pH of the endosome causes the virion envelope to fuse with the endosomal membrane, causing the nucleocapsid to become uncoate time. The viral RNA is then released into the cytoplasm, where RNA replication takes place at a rate Thi s rate of assembly and maturation is weighted inversely by the concentration of attached virus particles : I t is well known that WNV replication and maturation are error prone processes, thus to account for the production of de fective (non infectious) virus particles, we include in the equation for R t to denote the loss of virus partic les from the system. I nfectious virus particles
52 or are degraded by RNases and other environmental factors 1 illustrates the trajectory of the virus population in each o f the categories. Figure 4 1. Illustration of a generalized Flavivirus replication cycle. Virus particles ( V ) become attached ( A ) to the cell membrane. Virus RNA is then released into the cytoplasm, where virus replication occurs. Newly assembled, immatu re virions enter the endoplasmic reticulum ( R ) and progress along a pH gradient until becoming released back into the extracellular matrix as infectious virus particles. Mathematical Description of the M odel The model consists of a system of differential equations, each with a positive term and one or more negative terms (Figure 4 2) The positive terms represent the rate of entry of virus particles into a given stage (A, R or V) while the negative terms denote either losses of virus particles from the sy stem or transitions of virus particles from one stage to the next. We chose a continuous time model because WNV establishes a persistent infection, meaning that a cell can be infected multiple times, producing several overlapping generations of virus parti cles before cell death.
53 (4 1) Stability Analysis for the Endemic E quilibrium Our WNV model is a system of differential equations that expresses the rate of change of the number of infectious virus particles per mL the number of virus particles i nside the cell as well as the number of attac hed virus particles (Figure 4 1). Setting the equations in the system e qual to 0 and solving for the values of amounts to finding the values of the state variables for which their rate of change is 0. These are termed the equilibrium values of the state variables. The trivial equilibrium for the model is If however, all these case, the non trivial equilibria a re in Equation 4 2 ( 4 2) Provided that the numerator of V* is positive, then a simple inspection of all these equilibrium abundances show that these will all be positive. In other words, provided these equilibrium abundances will correspond to an endemic equilibrium. In these differential equation models, the equilibrium values can correspond to concentrations to which the system arrives some time after starting the experiment and subsequently stays there indefinitely, or they can correspond to concentration values that the system quickly gets away
54 he stability analysis is to determine when the values of and above correspond to a stable equilibrium. Below, we show that the same condition, must be met for the endemic equilibrium above to be stable. This stability condition is important because it delineates when a positive amount of free virus particles will persist in the system. In other words, this inequality represents an explicit condition that, if met will guarantee the long term persistence of free virus particles in the system (as well as attached and intracellular). Hence, the reproductive number in the system is in Equation 4 3. ( 4 3) Assuming that the system is started close to the equilibrium values, the goal of the stability analysis is to study the long term behavior of the deviations of the system state from these equilibrium values. If the deviations grow indefinitely over time, the equilibrium is unstable. If however the devi ations of the system state from the equilibrium values decay towards 0 over time, then the equilibrium is stable, i.e. the trajectories of the state variables will eventually converge to their equilibrium values. The system of eq uations (4 4) follows
55 ( 4 4) T o study the behavior of the deviations from equilibrium over time, we approximate using a Taylor series expansion around Using vector notation by setting the Taylor series approximation yields the following system of equations, written in matrix format in Equation 4 5 : (4 5) Computing these partial derivatives gives the system of equations in 4 6 ( 4 6)
56 As it turns out (Kot 2000), the sign of the eigenvalues and of the Jacobian matrix evaluated at the equilibrium determine the trajec tories of those deviations because the solution trajectory for the deviations, written in vector format is of the form shown in Equation 4 7 (4 7) where the vectors are constants depending on the mode l parameters. When all the eigenvalues are negative, as all the terms converge to 0, and hence, all the elements in the vector of the deviations of equilibrium converge to 0. In other words, as time grows large, if all the eigenvalues of the Jacobian matrix are negative, then the system trajectories converge to the equilibrium values and the equilibrium is said to be stable. Knowing whether the three eigenvalues are negative a mounts to verifying that the Routh Hurwitz criterion (Kot, 2000) holds. For our thr ee dimensional model system, the s e criteria are stated in Equation 4 8 : (4 8) This determinant is readily found to be in Equation 4 9 (4 9) Collecting the terms in in this expression gives a polynomial of degree three of the form Straightforward algebraic manipulations of the above expression give
57 and The three solutions to the characteristic equation (Equation 4 10) are the three eigenvalues of ( 4 10) The Routh Hurwitz criterion states that these eigenvalues are all negative (i.e., the equilibrium is stable) provided Because all the model parameters are positive on), it immediately follows that both and are negative, and hence their product is positive. Now, determining the sign of is more challenging because in the expression above, in the numerator there is a negative sign. However, a part of this numerator is immediately recognized as the negative of the numerator of Substituting in the expression for the explicit values value s of and simplifying and expressing its equation in terms of gives Equation 4 11 (4 11) Equation 4 11 is negative because all the parameters are b iologically valid in our model only when they are positive, and provided Then, having established that and that if it follows that the Routh Hurwi tz criterion holds provided
58 Hence, the endemic (positive) equilibrium is asymptotically stable whenever the basic reproductive number is greater t han 1, and our proof is complete. Simulations the level of confidence in each parameter when varying sampling frequency. At the start of the experiment, the number of attached virus particles A 0 and the number of intracellular virus particles R 0 are both 0. We initiate virus population growth with a multiplicity of infection (MOI) of 3 virus particles per cell; thus V 0 = 3. To set the initial parameter values, we manually assigned a combination of values that would yield biologically feasible trajectories, where V t would stabilize to a carrying capacity and A t would remain below V t Assessing the E stimability of the M odel P arameters The estimability of the mod el parameters was assessed in the context of the data generated by the serial passage experiment. Under this experimental setting, a typical data set consists of a number, say n of virus samples collected from the supernatant at regular time intervals (e .g. if samples are collected every 8 hours for a period of 72 hours that gives us 10 samples if time 0 is included). Then, to assess estimability of the model parameters, the idea is to simulate the trajectories of A R and V as well as regular random samp les from those trajectories at intervals matching the experimental conditions. With those simulated samples, one can then estimate the model parameters and later judge, from the statistical properties of the estimates, the quality of the estimation. Furth ermore, here we modified the sampling frequency in order to assess how does the quality of the estimate changes as a function of the number of samples taken, during the same total time period an experiment is running. The specific details of the simul atio ns are given below.
59 The time series of samples with observation error were then used to estimate via least squares the model parameters. As stated in Ponciano and Capistran 2011, least squares estimation amounts to specifying a Normal likelihood. These t wo steps (simulation and estimation) were the basic building blocks of our simulation experiments. The statistical qualities of the estimates under any given experimental setting can be assessed by running these two steps a large number of times (here we u sed 100 times for all of our simulation settings). Then, the mean and variance of the estimates, relative to the true parameter values gives precise estimates of the bias and variability of the model parameter estimates. We assessed the quality of the parameter estimates under two different simulation settings. In the first setting (simulated 100 times), we assumed that for a total of hours virus particles were sampled every 1.44 hours, to get a total of 50 observations. A lthough gathering 50 samples of the supernatant in 72 hours represents substantial experimental work, having a large number of samples gave us a benchmark for assessing the estimability of the parameters. Indeed, in time series statistics, the more sample s over time are gathered, the better the quality of the estimate of the dynamics and of the model para meters are. The second experiment consisted of simulating a total of 25 observations taken every 2.88 hours during 72 hours. This setting corresponded t o a realistic experimental setting. For this experimental setting we also ran 100 simulation and estimation steps. Finally, for each sampling setting (25 and 50 samples), we assessed the quality of the estimates when one of the parameters was assumed to be known (or estimated empirically from a different experiment), for each one of the model parameters. Thus, for each one of the two settings we initially programmed 7 simulation experiments with 100 runs each (for a total of 700 s imulation and estimation steps). The
60 programs were all written in the language R and as it stands, the computer code takes about 24 hours to run per simulated setting. Results The p arameters that gave biologically feasible trajectories of t he model are shown in Figure 4 4 which presents a simple test of estimability of the model parameters, using 50 simulations. To obtain this figure, I conducted 50 simulations of the trajectories with 25 where the deterministic trajectories were contaminated each tim e with different levels of observation error, (up to 20%) For each simulated time series, I estimated the model param e ters using only as observations the time series of contaminated with observation error. The figure shows that all the estimated parameters fall around the true value, which gave us a quick indication that the model parameters were indeed estimable when one has as data only the time series of the free virus particles. Going from 25 to 50 samples in time only imp roves the quality of the inference by a little bit, but not substantially (results not shown) When estimating all parameters, all appear unbiased. The variance of the parameter estimates, however, varies widely which indicates that the time series of viru s particles contains more information for some parameters than others. For example, Figure 4 3 shows that has a large variance and is sometimes over estimated three to fourfold. Because is the most difficult parameter to estimate, when this parameter i s known, the remaining parameters are estimated with the highest precision and unbiased for all experiments. Assuming that is known yields the worst estimates. Fixing also gives good parameter estimates but would be difficult to measure empirically. Fixing individually improves the precision the estimates of is the parameter that is fixed
61 Figure 4 2 Testing the estimability of the parameters. For 5 0 simulat ed time series of length 25 all the model parameters were e stimated and compared to the true value (horizontal lines in red) used to generate the simulations In all cases, the parameter estimates fell within the range of values that resulted from the simulations.
62 Figure 4 3 Relative bias of each parameter when certain parameters are known after running 100 simulations for each set of parameters In the upper left panel, it is assumed that all parameters are unknown. The most difficult parameter to estimate (most uncertain) ) Discussion In general, all parameter estimates are unbiased; however the prec ision varies according to which parameter is known (Figure 4 3 ). precision of the model. This parameter appears only once in the system of equa tions and is Thus, fixing it imposes a numerica l restriction on the potential optimal solution. These results emphasize that teasing apar t the estimates of can be challenging. If one could estimate and the variability in the estimates of so that now the estimates are on av erage unbiased and with a precision that frames the estimate within 0.5 and 1.5x relative to t he true value (Figure 4 5 ).
63 Obtaining values for the parameters in the laboratory poses a different challenge; however the results of the simulations provide gui dance on which parameters are most informative, all else being equal. As shown in Chapter 3, V t can be measured directly as plaque forming units or, alternatively, by TCID50. To determine the number of defective virus particles that are being produced, the difference between the number of genome copies by real time RT PCR and the number of plaque forming units can be calculated. This will allow an estimation of the rate at which virus particles leave the system either via incomplete maturation or assembl y, or due to degradation of viral RNA. Measuring R t becomes more costly: electron microscopy can be used to visualize the process of virus particle maturation through the endoplasmic reticulum. Direct measurement of A t and of the kinetics of RNA replicatio n in the and maturation, it will be possible to obtain statistical estimates of the parameters that cannot be measured in the laboratory. Estimation of the pa rameters of a biological, dynamic model using time series data is a common approach in ecological research ( Ferguson et al. 2014, 2015a, 2015b ). We apply this approach in the field of virology, where we attempt to estimate parameters in the virus replicat ion cycle. W e show that when the data at hand consists of time series of observations of one of the stages of the virus replication cycle (V t ) it is possible to estimate with a suitable degree of precision the value of t he critical parameters governing th e transitions and changes in the virus life history processes. Although the model is a first and simple representation of the WNV life cycle, our results represent a substantial step into the formulation of a general understanding of the dynamical process es involved in such life cycle.
64 Our simulations and assessment of the statistical properties of the Maximum Likelihood (ML) estimates of the model parameters show how to target efforts aiming at complementing time series statistical analyses with experi mental methods. Because the variance of the parameters varies widely from one parameter to the other, the changes in the observed time series of free virus particles contains more information to estimate some parameters than others. Via our approach where we assumed that one of the parameters at a time was known, we were able to isolate the set of parameters that it would be more valuable to estimate using expe rimental work. We hope that the results presented here offer guidance and ideas as to which one of these parameters or intra cellular life cycle stages it would be most interesting to target in the laboratory. The model analysis conveys important qualitative predictions. In our analysis, we were able to derive a basic reproductive number for the int ra cellular dynamics of the WNV. This number, being a direct function of the key parameters of the WNV life cycle, there is much to say and investigate regarding how the persistence of the infectious virus particles can be affected when one or several of these processes is changed. The value of the model then, besides its quantitative prediction is the gain of a qualitative understanding of the conditions under which virus persistence occurs. These predictions could then be scaled up in order to link the i ntra cellular processes with processes at higher scales involving the vector itself.
65 CHAPTER 5 CLINICAL CASES : TRANSMISSION OF ZIKA AND OTHER ARBOVIRUSES IN VENEZUELA Z ika virus (ZIKV) continues to spread throughout tropical and subtropical regions of the world. Transmission of ZIKV is known to occur through mosquito bites, from pregnant mother to fetus, through sexual contact, blood transfusion, and accidental laboratory exposure (WHO 2016). It is still unclear whether the virus can be transmitted through saliva, or if post natal transmission between mother and child can occur during breastfeeding, delivery or close contact between the mother and her newborn. Prior infection with other flaviviruses might affect the severity of ZIKV disease (Dejnirattisai 2 016). Little is known about how co infection with other viruses can affect disease severity and transmission rate. Our understanding of ZIKV infection is further complicated by the low viremia exhibited in the blood of infected patients. Virus loads are hi gher in saliva and in urine; furthermore, infected patients appear to shed virus in the urine for months after the acute phase (CDC 2016). Because detection by rt PCR in patient specimens can sometimes lead to false negatives when viremia is low, we couple d detection by rt PCR with virus isolation in cell culture. Results include a case of possible neonatal transmission via breast milk, as well as a possible increase in the incidence of autoimmune complications such as psoriasis (Appendix B ). Methods In Mar ch 2016, a network of doctors was formed and notified of the possible presence of ZIKV in the city of Barquisimeto, Venezuela (pop. 800,000). Participating doctors were briefed on procedures for screening patients for clinical symptoms of possible ZIKV inf ection (Table X) Informed consent forms and questionnaires were developed and distributed to the network. Specimen collection methods were submitted to the IRB in Barquisimeto and in Florida.
66 Patients with clinical symptoms of febrile arbovirus infection were sent to the laboratory at Hospital Internacional Barquisimet o Specimens are being collected following aseptic techniques outlined in hospital protocols and immediately frozen at 150 degrees C in liquid nitrogen. The specimens are shipped by courier to the University of Florida, where they are being tested for ZIKV and other virus infections. Specimens are thawed and handled according to previously established protocols in the Lednicky laboratory at the College of Public Health and Health Professions at the University of Florida, Gainesville. Once screened for ZIKV using primers published by Balm et al. (2012), the specimens will be screened for other Flaviviruses and Alphaviruses using a duplex RT PCR protocol with genus specific primers developed by Vieira de Morais Bronzoni et al. (2005). For virus isolation from urine and serum, four cell lines (MRC 5, Vero E6, LLC MK2 and C6/36) will be inoculated at 60% confluence with aliquots of the specimens. Inoculated cells will be allowed to incubate for 2 hours; checking every 15 minutes for cytotoxic effects of the inoculum. Total inoculation time is adjusted according to cytotoxic effects of the inoculum. Infections will then be allowed to progress for 2 6 weeks depending on the onset of cytopathic effect s (CPE). Once CPEs are observed, the spent media will be tested by rt PCR using the primers mentioned previously. On March 25, 2016, a 32 year old female patient from Barquisimeto, Venezuela presented with a 1 day history of symptoms associated with acute ZIKV infection: malaise, arthralgia, conjunctival hypere mia, and maculopapular rash (Figure 4 1 ). At that time, she was exclusively breastfeeding her 5 month old child, who was asymptomatic. Breast milk, serum, and urine were collected from the mother on March 28, 2016 (4 days after onset of acute Zika Fever symptoms), and serum and urine from the child the same day, and analyzed as detailed in time RT PCR (Ct
67 26.73, level of detection : Ct 36.8). Serologic analyses revealed the mother had IgM and borderline IgG antibodies against ZIKV and no detectable antibodies to Chikungunya virus. She was also IgG positive but IgM negative for Dengue virus (DENV). The mother remained symptomatic wit h arthralgias and malaise lasting 10 days, with the macular papular rash and urine were positive for ZIKV by real time RT PCR, with Ct 35.57 and 35.36. The child rem ained asymptomatic throughout the observation period. Vero E6 and LLC MK2 cells were (S1). Cytopathic effects (CPE) characteristic of ZIKV infectio n were observed 9 and 12 days post inoculation of MK2 cells (S2 Figure 2); the breast milk was fractioned, with CPE most obvious in cells inoculated with the lipid enriched fraction. The presence of ZIKV vRNA was confirmed in all cultures by RT PCR. Full genome sequencing of ZIKV isolated from breast milk and (GenBank # KX702400 and KX893855) revealed 99% identity, with only two synonymous nucleotide substitutions at third codon positions between the two strains. Both strains were different from the genomic sequences of other ZIKV strains in the laboratory. Moreover, sequencing of the NS5 gene of the other isolates indicated identical virus was in all specimens from both mother and child; mock infected cells did not develop CPE, and their spent media was RT PCR negative for ZIKV vRNA. Phylogenetic analysis showed that the two strains cluster with high bootstrap support (99%) within a larger clade of Colombian sequences (S2 Figure 3) Intere stingly, the mother had no history of traveling to Colombia. However, Barquisimeto is on a major trade route to Colombia where a large number of Venezuelans have regularly been traveling recently due to food and medicine shortages in Venezuela. We cannot r ule out the
68 possibility of transmission of an identical viral strain to both mother and infant by a mosquito. However, the mother and child lived in an air conditioned, screened house, where the risk of mosquito transmission would be minimized; they had no recent travel history, and the baby spent most of the time consistent with previous findings (Lednicky et al. 2016) ; however to date no studies have shown that transmission through breast milk is possible. Matching isolates from her breastfeeding infant is most consistent with post natal transmission from mother to child. Interestingly, the child, in contrast to the mother, was asymptomatic; assuming that transmission was via breast milk, this provides reassurance that asymptomatic infection can occur in a healthy 5 month old infant infected via breast milk. P atient S pecimens Blood and urine from mother and child, and breast milk were obtained 28 March, 2016 (4 days after onset of Z ika Fever symptoms). Blood (8 mL from mother, 4 mL from baby) were collected into acid citrate dextrose yellow top tubes (ACD Vacutainer blood collection tube, Becton Dickinson and Company, Franklin Lakes, NJ). Urine was collected from the mother followi ng a standard mid stream clean catch method, and by pediatric urine collector bag for the baby. Breast milk was collected into a sterile container after the areola and nipple were cleaned. The specimens were immediately transported to the laboratory at Hos pital Internacional Barquisimeto where approximately 1.5 mL of each specimen was aseptically transferred to sterile cryopreservation vials, and all were subsequently stored in the vapor phase of a liquid nitrogen cryotank within 2 hours of collection. A du plicate specimen of the mother s serum was tested for parvo, cytomegalo, Epstein Barr, varicella zoster and herpes simplex virus types 1 and 2 at Laboratorio Genomik, Maracay, Venezuela following established methods 1 The remaining
69 specimens were shipped o n dry ice by an express courier to the University of Florida, where the frozen specimens were stored at 80 degrees C upon their receipt on 15 July, 2016. Cell C ulture Mammalian cell lines LLC MK2 (CCL 7) and Vero E6 (CRL 1586) were obtained from the Ameri can Type Culture Collection (ATCC, Manassas, VA) and propagated as monolayers at 37 C and 5% CO2 in Advanced Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 2 mM L Alanyl L Glutamine (GlutaMAX, Invitrogen, Carlsbad, CA, USA.), antibiotics [PSN; 50 g/ml penicillin, 50 g/ml streptomycin, 100 g/ml neomycin (Invitrogen, Carlsbad, CA, USA)], and 10% (v/v) low IgG, heat inactivated gamma irradiated fetal bovine serum [FBS (HyClone, Logan, UT, USA)] Cell C ulture I noculations When confluency was at 8 0%, growth media was removed from cell cultures in T25 cell culture flasks (cell growth surface of 25 cm 2 ) and replenished with 300 l of complete media. In preparation for inoculation of the cells, an aliquot of the breast milk was centrifuged to partitio n lipid rich and aqueous phases. This was performed to determine whether ZIKV, if present, would survive in a particular fraction of whole breast milk. Separate individual flasks of the cells were inoculated with 200 L aliquots of the specimens (whole mil k, lipid enriched milk, aqueous phase milk, plasma, and urine), and incubated at 37 C for 2 hours, with manual rocking of the flasks performed at 15 minute intervals. After 2 hours, 3 ml of additional growth media was added and the cells inoculated at 37 C Mock infected cells were maintained in parallel. The cells were observed daily and re fed with reduced serum media (3% FBS). At the University of Florida laboratory, the cells are maintained and observed for 1 month before being reported as
70 negative for virus isolation. Spent media from the inoculated cells were sporadically tested by RT PCR as described previously 2 Evidence of ZIKV I solation Based on the laboratory s cumulative experience with the isolation of ZIKV and as described previously 2 virus s pecific cytopathic effects (CPE) consisting of perinuclear vacuoles prior to cell death were expected in ZIKV infected LLC MK2 and Vero E6 cells. The ZIKV specific CPE develop earlier and are more pronounced in LLC MK2 cells. Cytopathic effects characteris tic of ZIKV were observed in all cells inoculated with mother and child s specimens; they were first observed in LLC MK2 cells 11 days post inoculation and 1 3 days later in Vero E6 cells. CPE were initially most obvious in LLC MK2 cells inoculated with li pid enriched breast milk. No CPE were observed in the mock infected negative control cells. Representative images of the infected cells are provided in S2. RT PCR S creens for Chikungunya, Dengue, and Zika virus RNA To screen the cultures for Chikungunya vi rus [CHIKV], Dengue viruses 1 4 [DENV 1, 2, 3, and 4], and ZIKV genomic RNAs, vRNA was extracted from virions in spent cell growth media using a QIAamp Viral RNA Mini Kit (Qiagen Inc., Valencia, CA). The extracted vRNAs were tested by RT PCR using prime rs and procedures for CHIKV 3 for DENV types 1 4 4 and for ZIKV 5,6 Preliminary screens were performed using Omniscript rev erse transcriptase (Qiagen) and OneTaq DNA polymerase (New England Biolabs, Ipswich, MA) for PCR. ZIKV Sequencing At 14 days post ino culation, vRNA was extracted from virions in the spent media of LLC urine and sequenced as described previously2. Briefly, high fidelity enzymes [Accuscript High Fidelity
71 reverse transcriptase in the presence of SUPERase In RNase inhibitor (Ambion, Austin, TX) and Phusion Polymerase (New England Biolabs)] were used for reverse transcription and PCR. The using RNA ligase mediated RACE (RLM RACE), whereas the 3 polyadenylated bidirectionally using Sanger Sequencing using a genome walking strategy and primers described previously2 ZIKV P hylogenetic A nalysis All currently available ZIKV full genome sequences were downloaded from NCBI ( http://www.ncbi.nlm.nih.gov/ ). Sequences were aligned using ClustalW 7 followed by manual editing with Bioedit 8 The Maximum Likelihood phylogenetic tree was inferr ed from the full genome alignment using the best fitting substitution model with the IQTREE program 9 as in our previous study 2 Statistical robustness and reliability of the branching order within the tree were assessed by bootstrapping (1000 replicates) and fast likelihood based Shimodaira Haswgawa (SH) like probabilities 10 with IQ TREE. Serology Zika virus serologic testing was performed using the Zika Virus ViraStripe IgG, IgM Test Kit (Viramed Planegg, Germany). This immunoassay allows the detection of IgG or IgM antibodies against ZIKV specific antigens in human serum. ViraStripe carries the following purified Zika Virus specific antigens: E, EIII, EIII* (Envelope antigens) and NS1 (Non Structural antigen 1). The control section of each strip includ es a serum control, three conjugate controls and a cut off control. Results of the tests were positive for IgM and borderline for IgG, with low titers suggesting seroconversion.
72 Results On March 25, 2016, a 32 year old female patient from Barquisimeto, Ven ezuela presented with a 1 day history of symptoms associated with acute ZIKV infection: malaise, arthralgia, conjunctival hyperemia, and maculopapular rash (Figure 4 1) Figure 5 ssion for publication in March 2016. At that time, she was exclusively breastfeeding her 5 month old child, who was asymptomatic. Breast milk, serum, and urine were collected from the mother on March 28, 2016 (4 days after onset of acute Zika Fever sympto ms), and serum and urine from the child the same day. Blood and urine from mother and child, and breast milk were obtained 28 March, 2016 (4 tested for parvo, cytomega lo, Epstein Barr, varicella zoster and herpes simplex virus types 1 and 2 at Laboratorio Genomik, Maracay, Venezuela. Mammalian cell lines LLC MK2 (CCL 7) and Vero E6 (CRL 1586) were propagated as monolayers. In preparation for inoculation of the cells, an aliquot of the breast milk was centrifuged to partition lipid rich and aqueous phases. This was
73 performed to determine whether ZIKV, if present, would survive in a particular fraction of whole breast milk. After 2 hours, 3 ml of additional growth media wa s added and the cells inoculated at 37C. Mock infected cells were maintained in parallel. The cells were observed daily and re fed with reduced serum media (3% FBS). Virus specific cytopathic effects (CPE) consisting of perinuclear vacuoles prior to cell death were expected in ZIKV infected LLCMK2 and Vero E6 cells. The ZIKV specific CPE develop earlier and are more pronounced in LLC MK2 cells. Cytopathic effects characteristic of ZIKV were observed in all cells inoculated with mother and ; they were first observed in LLC MK2 cells 11 days post inoculation and 1 3 days later in Vero E6 cells. CPE were initially most obvious in LLC MK2 cells inoculated with lipid enriched breast milk (Figure 5 2) No CPE were observed in the mock infected ne gative control cells. Representative images of the infected cells are provided in Figure 5 2. Figure 5 2 Characteristic ZIKV specific cytopathic effects in LLC MK2 cells. A. Mock infected LLC MK2 cells, 12 days post seed. Original image magnification a t 400X. B. LLC MK2 cells inoculated with breast milk lipid enriched fraction, 12 days post seed. Arrows point out early (small) and late (large) perinuclear vacuoles characteristic of ZIKV infection. Original image magnification at 400X.
74 At 14 days post i noculation, vRNA was extracted from virions in the spent media of LLC sequenced. Sequences were aligned using ClustalW followed by manual editing with Bioedit. The Max imum Likelihood phylogenetic tree was inferred from the full genome alignment using the best fitting substitution model with the IQTREE program. Phylogenetic analysis showed that the two strains cluster with high bootstrap support (99%) within a larger cla de of Colombian sequences (Figure 5 3 ).
75 Figure 5 3. Maximum Likelihood (ML) tree inferred from available ZIKV whole genome sequences. The ML tree was inferred from full genome sequences with the best fitting nucleotide substitution model selected by a hierarchical likelihood test using the program IQ TREE. All ZIKV full genomes available in GenBank were used. For simplicity only the portion of the tree showing the South American lineage is shown. Sequence labels are colored by country of origin, accord ing to the legend on the left. The sampling year of each sequence is also given in the sequence label. Internal nodes highlighted by red circles have strong bootstrap support (>99%). Arrows indicate the mother and child ZIKV isolates. Branch lengths are sc aled in nucleotide substitutions per site according to the bar in the figure. Discussion Interestingly, the mother had no history of traveling to Colombia. However, Barquisimeto is on a major trade route to Colombia where a large number of Venezuelans hav e regularly been traveling recently due to food and medicine shortages in Venezuela. We cannot rule out the
76 possibility of transmission of an identical viral strain to both mother and infant by a mosquito. However, the mother and child lived in an air cond itioned, screened house, where the risk of mosquito transmission would be minimized; they had no recent travel history, and the baby spent most of the time within the house. To date no studies have shown that transmission through breast milk is possible. M atching isolates from her breastfeeding infant is most consistent with post natal transmission from mother to child. Interestingly, the child, in contrast to the mother, was asymptomatic; assuming that transmission was via breast milk, this provides reassu rance that asymptomatic infection can occur in a healthy 5 month old infant infected via breast milk.
77 CHAPTER 6 CONCLUSIONS AND DISCUSSION We sought to investigate the role of virulence, transmission mode and environmental context could explain the surg e in arboviral disease that has occurred during the past few decades. We combined field data, laboratory experiments, mathematical models and statistical tools to evaluate our findings and to generate predictions across scales of biological organization. R esults from the work with Sigma virus suggest that mathematical models, when informed by laboratory experiments, can provide a means for extrapolating results to other scales of biological organization as suggested in Lloyd Smith 2009 and in Lord 2014 Fu rthermore, the laboratory based estimates of virulence were informative enough to accurately predict the field measured estimates of virus abundance. Interestingly, despite a high frequency of vertical transmission and high virulence, Sigma virus persists at high densities in the population of D. melanogaster that was sampled. This could be attributed to various aspects of the environmental context such as possible reintroductions of the virus into the D. melanogaster population, seasonality of Sigma virus abundance, and environmental context such as the fact that our field sites were located along a transect that was adjacent to areas of possibly high fertilizer use. Given that the populations that were sampled in this study occurred in areas of high agricu ltural activity, it was sensible to investigate the role of agriculture in determining the prevalence of the virus. To improve our ability to generate predictions concerning arboviruses that are less dependent on strictly vertical transmission, we proceede d to address the next research question by directly studying an arbovirus that exhibits both horizontal and vertical transmission and occurs in areas of high agricultural activity (Reisen 2013) The results from the laboratory experiments with WNV suggest that P could affect virus load indirectly via changes in cell division rates; however we did not detect a direct effect of P on the replication rate of the
78 virus in cell culture. The absence of this effect could be due to the narrow range of P conditions t hat were achievable in cell culture. It is also possible that the changes in P that were experimentally induced in the cell culture media were not reflected inside of the cells. Osmoregulatory processes could have maintained homeostatic internal concentrat ions of I virus replication in cell culture. The limiting nutrient could be C, N or another micronutrient Testing the effects of P on vector competence could yield results that are different from those obtained in cell culture. Effects of P on other aspects of mosquito biology, which could affect vector competence, were quantified experimentally by Peck et al. ( 2005 ) in Cx. p ipiens quinquefasciatus and in Cx. pipiens tarsalis, two competent vectors of WNV An increase in the rate of larval development was observed for one species, where P led to earlier emergence and smaller size at emergence. One issue with this work was that the experimental increase in P led to high mortality, thus sample sizes were low. Future work should include experiments where total P is increased following alternative protocols that cause less mortality to the larvae. Given that larval stoichiometric ratios mirrored those of the environ ment, it would be interesting to determine whether those stoichiometric signatures are maintained throughout the lifespan of the mosquitos, and the effects of larval nutritional ratios on futu re reproductive success of fema l e s should be investigated, given that gonadal tissue tends to be higher in per unit mass concentration of P. Finally, future work should measure other aspects of vector competence such as lifespan. The fact that a positive effect of P was observed at the landscape level and at the indivi dual level in other systems suggests that the results from cell culture could not hold across all scales of biological organization According to metabolic theory (Sterner 2002) we expect a negative relationship between mass specific metabolic demands of P and lifespan. Thus, it is
79 possible that faster growing mosquitoes could have lower vector competence due to fewer opportunities for bloodfeeding. Because of these results, the original model of virus replication was modified so that P was kept out of the system of equations. The literature on mathematical models of within host Flavivirus dynamics is focused primarily on HCV. These models involve systems of differential equations where the replication rate of the virus is modeled as a function of various aspects of cell membrane composition, gene expression, and others. Models of the role of intracellular pH, which is regulated by P, are nonexistent to our knowledge. Our model attempts to incorporate the possible pathways by which P could affect WNV popula tion dynamics within a host Furthermore, we apply sta tistical tools developed in ecological population biology (Ferguson et al. 2014, 2015a, 2015b) to determine the accuracy with which we can estimate the rate of virus attachment, replication, maturation and ex it from an infected cell. Initial estimates of these parameters all included the We were able to estimate the rate of virus attachment, entry, and exit from the cell with relatively higher confidence than othe r parameters The most difficult parameter to estimate is the rate of virus replication and assembly within a cell. This is also difficult to measure in the laboratory. Improvements to our model of intracellular replication could draw from HCV studies (Dah ari 2007, Padmamabhan 2011) ; for viral assembly from phage research ( Zlotnick 2011) and for maturation and assembly in the endoplasmic reticulum (Romero Brey 2016). Results from the study in Venezuela suggest that transmission mode could be a dynamic proce ss that is not only a function of virulence, but a function of immunologic al history of the host. T he role of co infections should also be investigated as preliminary evidence suggests that two of the patients infected with ZIKV were also infected with DE NV4 (Blohm unpublished
80 data) Testing clinical specimens for multiple viral infections when patients present with febrile illness is important for understanding the basic virology and epidemiology of arboviruses. Second, prior immunological history of the patient can determine virulence and therefore the evolutionary trajectory of the host and of the v irus. As indicated in Appendix B patients with prior history of autoimmune diseases experienced greater susceptibility and a wider range of complications in response to infection with ZIKV than patients that do not have prior exposure to either ZIKV or a closely related Flavivirus such as DENV4 When an accurate, specific antibody test for DENV and ZIKV is developed, it will be possible to assess the role of p rior infection with closely related viruses on clinical outcomes of infection. Continued work that is multidisciplinary in its approach will lead to the development of improved predictive models, tools for clinical diagnosis in resource limited settings, a nd a deeper understanding of the factors that give rise to the establishment and persistence of arboviruses. The combination of model system approaches, mathematical models, and laboratory experiments is a fruitful source of information about clinically im portant arboviruses.
81 APPENDIX A FORMULATION OF CELL CULTURE MEDIUM Table A 1. Formulation of L 15 cell culture medium. Components Molecular Weight Concentration (mg/L) mM Amino Acids Glycine 75 200 2.6666667 L Alanine 89 225 2.52809 L Arginine 174 50 0 2.8735633 L Asparagine 132 250 1.8939394 L Cysteine 121 120 0.9917355 L Glutamine 146 300 2.0547945 L Histidine 155 250 1.6129032 L Isoleucine 131 250 1.908397 L Leucine 131 125 0.9541985 L Lysine 146 75 0.51369864 L Methionine 149 75 0.5033557 L Phenylalanine 165 125 0.75757575 L Serine 105 200 1.9047619 L Threonine 119 300 2.5210085 L Tryptophan 204 20 0.09803922 L Tyrosine 181 300 1.6574585 L Valine 117 100 0.85470086 Vitamins Choline chloride 140 1 0.007142857 D Calcium pantothenate 4 77 1 0.002096436 Folic Acid 441 1 0.002267574 Niacinamide 122 1 0.008196721 Pyridoxine hydrochloride 206 1 0.004854369 Riboflavin 5' phosphate Na 478 0.1 2.09E 04 Thiamine monophosphate 442 1 0.002262444 i Inositol 180 2 0.011111111 Inorganic Salts Calcium Chloride (CaCl2) (anhyd.) 111 140 1.2612612 Magnesium Chloride (anhydrous) 95 93.7 0.9863158
82 Magnesium Sulfate (MgSO4) (anhyd.) 120 97.67 0.8139166 Potassium Chloride (KCl) 75 400 5.3333335 Potassium Phosphate monobasic (KH2PO4) 136 60 0.44117 647 Sodium Chloride (NaCl) 58 8000 137.93103 Sodium Phosphate dibasic (Na2HPO4) anhydrous 142 190 1.3380282 Other Components D+ Galactose 180 900 5 Phenol Red 376.4 10 0.026567481 Sodium Pyruvate 110 550 5
83 APPENDIX B MANUSCRIPTS IN PREPARA TION: CUTANEOUS MANIFESTATIONS OF Z IKV G eneralized pustular psoriasis triggered by Zika virus infection Alberto E Paniz Mondolfi 1,2 Marier Hernandez Perez 3 Gabriela Blohm 4 Marilianna Marquez 5,6 Adriana Mogollon Mendoza 5,6 Carlos E Hernandez Pereira 5,6 Julia Rothe DeArocha 7 Alfonso J Rodriguez Morales 8 Affiliations: 1 Department of Tropical Medicine and Infectious Diseases, Hospital Internacional Barquisimeto, Lara, Barquisimeto, Venezuela. 2 Instituto Venezolano de los Seguros Sociales (IVSS), Department of Health, Caracas, Venez uela; 3 Department of Dermatopathology, Miraca Life Sciences Research Institute/Tufts Medical Center, Boston, MA, USA; 4 Department of Biology, College of Liberal Arts and Sciences, University of Florida, Gainesville, FL, USA; 5 Infectious Diseases Research Incubator and the Zoonosis and Emerging Pathogens regional collaborative network. 6 Health Sciences Department, College of Medicine, Universidad Centroccidental Lisandro Alvarado, Barquisimeto, Lara, Venezuela; 7 Psoriasis Unit, Hospital Central "Antonio Mar ia Pineda", Barquisimeto, Lara, Venezuela. 8 Public Health and Infection Research Group, Faculty of Health Sciences, Universidad Tecnolgica de Pereira, Pereira, Colombia. Keywords : Zika virus ; psoriasis; viral, cutaneous. Running title: Psoriasis trigge red by Zika virus
84 Correspondence: Alberto E Paniz Mondolfi MD, MSc, FFTM RCPS (Glasg), Yale New Haven Hospital, Microbiology Laboratory (PS656), 55 Park Street, New Haven CT 06511 ( alberto.paniz mond firstname.lastname@example.org email@example.com). Abstract Z ika virus (ZIKV) is an emerging arthropod borne virus belonging to the flaviviridae family (1) which is expanding in epidemic proportions through tropical and subtropical areas around the world (1). Its cl inical presentation is non specific being usually misdiagnosed with other classical viral exanthems and arboviral infections such as Chikungunya (CHIKV), Dengue (DENV) and Mayaro (MAYV) (1, 2), thus posing a challenge at the time of diagnosis. The emergenc e of ZIKV has been linked to the development of a number of clinical complications, mainly congenital and neurological (1). Yet, besides its self limiting pruritic maculo papular rash, little is known about the biology and cutaneous manifestations of ZIKV disease. Infections are amongst the well known triggers of psoriasis (3). Herein, we report an exceptionally interesting case of psoriasis presenting three weeks after an otherwise uneventful resolution of an acute ZIKV infection. Report A 68 year old wo man presented with a 10 day history of generalized erythroderma and scaly plaques of acute onset. She also complained of general malaise, fever and localized tenderness. Three weeks prior, she had developed a pruritic maculo papular rash along with astheni a, small joint arthralgias and conjuctival hyperemia, which resolved uneventfully after 5 days. At that time a full blood count and chemistry were unremarkable except for mild lymphocyte leukocytosis. Serologic analyses using the Zika Virastripe IgG/IgM t est kit (Viramed, Planegg, Germany)
85 revealed a positive IgM and negative IgG. Additional serologic testing for DENV, CHIKV, EBV, CMV and parvovirus returned negative. RT PCR for DENV and CHIKV were negative but positive for ZIKV. On physical examination the patient exhibited extensive erythorderma and sharply demarcated erythematous silvery scaly round to oval plaques. Most plaques harbored obvious coalescing macroscopic pustules forming large central crusts (Fig. 1 a, b). Lesions started as erythematous macules exhibiting an abrupt centrifugal expansion in a period of hours. Interestingly, most lesions localized to trunk and proximal aspects of limbs (were the original ZIKV associated rash was more pronounced). Histological examination revealed psoriasi form epidermal hyperplasia with horizontally confluent parakeratosis and neutrophil exocytosis (Fig. 1a) at the lesion edges along with intraepidermal (subcorneal) neutrophilic pustules towards the center, and a mixed dermal lymphocytic and neutrophil infi ltrate. At admission, the complete blood count, chemistry and hepatic transaminase tests were normal with an elev ated C agents including hepatitis A, B and C, HIV and syphilis was negative. Antistreptolysin O titers returned negative as well. The patient was initiated on intensive topical therapy with class 1 steroids (Clobetasol propionate 0.05% BID) and methotrexate at a starting dose of 7.5 mg weekly. Her symptoms gradually resolved with persistence of a few large lesions that eventually faded after 15 weeks. She did not report any side effects and laboratory values remained normal.
86 Discussion Psoriasis is a chronic skin disease that affects approximately 2% of the population (4) with many triggering factors, both external and systemic capable of inducing the disease phenotype in suscept ible individuals (4). Compelling evidence suggests that many microorganisms may play a role in the onset or exacerbation of psoriasis (3). Bacterial agents such as streptococci and staphylococcus are considered the most common players implicated in the dev elopment of the disease, presumably through superantigen activation of skin seeking T cells (3). Other agents linked to the pathogenesis of psoriasis include fungi like Malassezia and Candida sp which through colonization may elicit an up regulation of k eratinocyte expression promoting a hyperproliferative state and also producing T cell activating superantigenic factors in a similar fashion to pyogenic bacteria (3). In addition, infection with viral agents such as retroviruses (3), EBV (5), VZV (6), CMV (5), HPV (3) and HSV (7) have also been associated with the onset of psoriasis. Moreover, a recent study has suggested the potential role of CHIKV as a trigger for psoriasis (8). Although viral triggers have been implicated in the pathogenesis of psoriasi s the exact mechanisms driving its progression have not been well established. The exact mechanism driving the pathogenesis of ZIKV at the cutaneous level remains unclear. However recent evidence reveals that human keratinocytes are permissive to ZIKV repl ication in early stages of infection with notable cytopathic effects and induction of apoptosis (9). This initial interplay between the virus and the keratinocyte sets the scene for the development of the psoriatic plaque initiating events,
87 including activ ation and production of type 1 interferons (IFN) via specific induction of pattern recognition receptors (PRRs) and up regulation of expression of IFN stimulated genes (OAS2, ISG15 and MX1) as well as chemokines (such as CXCL10, CXCL11 and CCL5) which prom ote T cell attraction and direct receptor independent like antimicrobial activity (9). In addition, the polyfunctional T cell activation (Th1, Th2, Th9 and Th17 response) seen during the acute phase of ZIKV infected patients (10) along with recent evidenc e that suggest that up to 50% of human in vitro generated immature dendritic cells (DCs) challenged with ZIKV express virus envelope proteins (favoring propagation of the virus in the human skin) (9); further support a role for aberrant activation of derma l DCs which would stimulate auto reactive Th17 cells and cytokines inducing keratinocyte activation and epidermal proliferation. To the best of our knowledge this is the first report linking ZIKV as a possible trigger for psoriasis. Experimental evidence suggests that the virus directly contributes to the release of keratinocyte derived mediators of the inflammatory process and the T cell driven immune reaction that drive the evolution of the psoriatic reaction. The association reported in this case provi des important clinical insights for further studies. References: 1. Musso D, Gubler DJ. 2016. Zika virus. Clin Microbiol Rev 2016 Jul;29(3):487 524. 2. Paniz Mondolfi AE, Rodriguez Morales AJ, Blohm G et al. ChikDenMaZika Syndrome: the challenge of diagnosing arboviral infections in the midst of concurrent epidemics. Ann Clin Microbiol Antimicrob 2016 Jul 22;15(1):42.
88 3. Fry L, Baker BS. Triggering psoriasis: the role of infections and medications. Clin Dermatol 2007 Nov Dec;25(6):606 15. 4. Nestle FO, Kaplan DH, B arker J. Psoriasis. N Engl J Med 2009 Jul 30;361(5):496 509. 5. Jiyad Z, Moriarty B, Creamer D et al. Generalized pustular psoriasis associated with Epstein Barr virus. Clin Exp Dermatol 2015 Mar;40(2):146 8. 6. Ito T, Furukawa F. Psoriasis guttate acuta trigg ered by varicella zoster virus infection. Eur J Dermatol 2000 Apr May;10(3):226 7. 7. Zampetti A, Gnarra M, Linder D et al. Psoriatic Pseudobalanitis Circinata as a Post Viral Koebner Phenomenon. Case Rep Dermatol 2010 Nov 6;2(3):183 188. 8. Seetharam KA, Sridev i K. Chikungunya infection: a new trigger for psoriasis. J Dermatol 2011 Oct;38(10):1033 4. 9. Hamel R, Dejarnac O, Wichit S et al. Biology of Zika Virus Infection in Human Skin Cells. J Virol 2015 Sep;89(17):8880 96. 10. Tappe D, Prez Girn JV, Zammarchi L et a l. Cytokine kinetics of Zika virus infected patients from acute to reconvalescent phase. Med Microbiol Immunol 2016 Jun; 205(3):269 73.
89 Cutaneous features of Zika virus infection: A clinicopathological overview. Paniz Mondolfi AE, Blohm GM, Hernandez Pe rez M, Larrazabal A, Moya D, Marquez M, Talamo A, Rothe J, Lednicky J, Morris JG. 1. Introduction: Zika virus (ZIKV) is an arbovirus belonging to the Flaviviridae family, which was first isolated serendipitously in non human primates (Rhesus monkeys) and mosquitoes in 1947 1948 in the Zika forest of Uganda (X). Its first isolation in humans was in 1954 in Nigeria (X), although many cases were later recorded through almost half a century during yellow fever serological survey studies in Africa. ZIKV has eme rged over the past decades causing minor reported in the Western Pacific island of Yap (Federated States of Micronesia), followed later by a larger epidemic i n French Polynesia through 2013 and 2014 (X). Recently in 2015, ZIKV reached the shores of the Americas following the footsteps of other arboviruses such as Chikungunya (CIHKV) and causing great concerns due to its unprecedented pathogenicity and increased risk for causing severe fetal malformations and neurological symptoms. Even though Zika virus infections are usually mild or remain clinically unapparent (X), cutaneous manifestations as for other arbovirosis remain a hallmark of the disease. In this arti cle, we discuss the cutaneous clinical spectrum of ZIKV infection, while comprehensively revising the main biological aspects of the virus and its interaction with human skin cells, the host immune response and pathogenesis. 2. Etiology and Epidemiology:
90 ZIKV is a Flavivirus closely related to the Yellow fever virus in the Spondweni (SPOV) clade of X mosquitoe borne Flaviviruses (X). It is an enveloped virus that is approximately 50 nm in diameter and contains a 10.794 kb single stranded, positive sense R NA genome (X). The frame coding for a polycistronic peptide that yields three structural ( capsid [C], premembrane/membrane [PrM], and envelope [E]) and seven nonstructu ral proteins, NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5 (X). Molecular epidemiology studies suggest that the virus emerged in East Africa around 1920 differentiating into two distinct African clusters, and later believed to have migrated to Asia in the 194 0s originating the Asian lineage strains (X). To date, phylogenetic studies conducted by Lanciotti et al. after the recent Yap State outbreak confirm the existence of three different ZIKV lineages or subclades: East African (prototype Uganda strain), West African (Senegal strains), and Asian (ZIKV 2007 Yap strain) (X). It is believed that ZIKV reached the Americas through introduction of an Asiatic lineage strain in ed during the New World epidemics remain to be elucidated. It is an arthropod borne virus, which is transmitted mainly by Aedes species of mosquitoes. It was first isolated from Aedes africanus (X) and later in a wide range of other Aedes species such as Ae. apicoargenteus Ae. vittatus Ae. furcifer and Ae. luteocephalus being the principal vectors of transmission Aedes aegypti and Aedes albopictus Other authors have reported the presence of the virus in other species such as Culex sp (X), Mansonia unif ormis Anopheles coustani and Culex perfuscus, nevertheless their role as competent vectors remains to be proven. (X).
91 In rural settings, reservoirs include non human primates and many other mammalian species (X); however in urban cycles humans may act as the main hosts (X). Though not much is known about its ecology, ZIKV transmission is thought to occur mainly in sylvatic cycles with consequent spillovers to rural and urban settings influenced by a number of epidemiological factors such as switch in ve ctor usage, the emergence of new strain variants (X) and other geographic and climatic factors influencing biomes and transmission dynamics (X). Nonvectorial transmission has been reported to occur directly from mother to child (X), by transfusion (X), th rough saliva (X), sexually (X) and presumably via breastfeeding (X). 3. Pathogenesis: The exact mechanisms driving the pathogenesis of ZIKV at the cutaneous level remains unclear. Nevertheless a recent study by Hamel et al. has provided important insights into the interaction of ZIKV and human skin cells (X). Similarly to other arboviruses, ZIKV infects the host by vector mediated transmission through blood feeding by female mosquitoes (X) who injects the virus into the skin allowing it to infect cutaneous resident cells such as dermal macrophages and skin fibroblasts (X). In these cells, the virus will drive its first replication cycle, thus triggering an initial immune response from the host as observed with other arboviruses (X). The concurrent delivery of several well potentiation and capacity of arboviruses, such as ZIKV to replicate at the anatomical site of inoculation (X), leading to an increased viremic phase of the disease in the vertebrate host (X) and its dissemination to other tissues causing acute viral symptoms (X).
92 Because skin cells are the first to encounter arboviral pathogens after their inoculation in the host, the identification of susceptible cell types is of pivotal importance on e lucidating the main pathogenic aspects of ZIKV infection. Some of the cells have shown to be permissive to the installment of infection via interaction with specific receptors of the mammalian host (X). Studies in human skin cells using a strain isolated f rom the recent outbreak in French Polynesia have shown that epidermal keratinocytes, immature dendritic cells and dermal fibroblasts are permissive to infection by ZIKV (X). Such permissiveness driving the entry of the flavivirus into the host cell seems t o be driven by its interaction with a variety of cell surface receptors and attachment factors such as the sulfated polysaccharide heparan sulfate which is widely known as a non specific attachment factor for flaviviruses (X). However, over a dozen putativ e receptors and attachment factors have been described to play a role in the entry of flaviviruses into mammalian and mosquito cells to date. More recently, the dendritic cell specific intracellular adhesion molecule 3 grabbing non integrin (DC SIGN), a ty pe of C type lectin receptor, was demonstrated to play an important role on mediating ZIKV internalization in immature dendritic cells (X). This is in lines with previous studies where it has been demonstrated that DC SIGN plays a major role in Dengue viru s (DENV) pathogenesis as well (X). In addition to DC SIGN, other entry / adhesion factors have shown to participate in ZIKV infection and internalization at the skin level. Such is the case of the AXL tyrosine kinase receptor (AXL), Tyro3 tyrosine kinase r eceptor (Tyro3) and the TIM 1 protein (X). In human skin, it has been shown that AXL strongly enhances ZIKV infection in synergy with TIM 1, probably with TIM 1 acting as a viral binding factor and later transferring it to AXL in order to initiate internal ization (X). However, it is presumed that TIM 1 is not indispensable for ZIKV endocytosis and that it rather may act by concentrating virions in the cell surface in order to
93 (X). Another important aspect to consider is the fact that ZIKV can infect and replicate in several of the cell constituents of the cutaneous milieu such as epithelial, mesenchymal, endothelial and immune cell lineages exhibiting at the same time a selective tropism for certain cells which is determined by the profile of receptor expression (X). In this sense, for example while immature dendritic cells express DC SIGN, epidermal keratinocytes and cutaneous fibroblasts lack this receptor but in turn express AXL which has proven to play a major role in ZIKV entry (X). Such a wide range of entry receptors provides the virus with an enhanced capacity to infect a variety of target cells in the skin. Even though the epidermis serves as an important m echanical barrier against physical and environmental aggressions, arboviruses such as ZIKV have managed to subvert this protective blockade by reaching the sensitive cell targets using the arthropods buccal apparatus to penetrate the thickness of this robu st outer compartment. But as briefly mentioned above, the epidermis is home to three major resident cell populations (keratinocytes, melanocytes and Langerhans cells), which play an important role in the initial homing and replication of ZIKV and which des erve to be examined in more detail in the following lines. 3.1. Keratinocytes Keratinocytes constitute the major cell population of the epidermis, playing an essential role in maintaining the infrastructure responsible for maintaining the barrier function through its differentiation into the outmost cornified layer (X). However, it is in the innermost layers of the epidermis were an intricate network of hormones and cytokines orchestrate the interaction of
94 keratinocytes with other specialized cells to driv e key immune and metabolic functions (X). Studies on cellular tropism have revealed that epidermal keratinocytes are the initial target for West Nile virus infection (X) as well as early DENV replication (X). Similarly Hamel et al. have explored the contri bution of primary human epidermal keratinocytes obtained from neonatal foreskin in the early stages of ZIKV infection, proving the permissive nature of human keratinocytes in ZIKV replication (X) as well. In addition, the gradual increase in the production of ZIKV particles in human keratinocytes along with the cytopathic effects observed in experimentally infected cells (X), signal the occurrence of ongoing apoptosis (X). As observed in CHIKV (X), WNV (X) and DENV (X), apoptotic blebs serve as shelter for and humoral immune response (X). In these same lines, it has been speculated that apoptosis in ZIKV follows this same strategy aiming to escape the immune response from the host by increasing their dissemi nation to neighboring healthy cells (X). From an immunological standpoint keratinocytes are known to play a key role in innate immunity via pathogen recognition (X). The type 1 interferon (IFN) pathway is perhaps one of the most important anti viral pathwa ys, with particular relevance in CHIKV (X) and DENV (X) infections. Activation and production of type 1 IFNs are triggered through pattern recognition receptors (PRRs) following recognition of pathogen associated molecular patterns (PAMPs) (X). Amongst the most important pattern recognition receptors are the Toll like receptors as well as the cytosolic retinoic acid inducible gene 1 (RIG1) like receptors MDA 5 (melanoma differentiation associated protein 5) and RIG 1 (X). As with DENV (X) and WNV (X) infect ions, ZIKV strongly induces the expression of important PRRs such as TLR3, MDA 5 and RIG 1 (X). However, type 1 IFN production in ZIKV infection appears to occur in an
95 independent fashion to IRF3 (X) as observed for other different viruses (X), in both fla vivirus infected epidermal keratinocytes and dermal fibroblasts (X). 3.2. Cutaneous fibroblasts Cutaneous fibroblasts also remain in the epicenter of ZIKV pathogenesis. Experimental evidence suggests that this cell population constitutes a site for active viral replication (X). As in keratinocytes ZIKV induces an innate anti viral response with specific induction of PRRs, upregulation of TLR3 mRNA expression, as well as enhanced transcription of RIG 1 and MDA5, thus promoting the initiation of downstream s ignaling pathways aimed to activate the antiviral machinery (X). In a similar fashion, ZIKV infection is known to upregulate the expression interferon stimulated genes (OAS2, ISG15 and MX1), and chemokines (such as, CXCL10, CXCL11 and CCL5) (X) which promo te T cell attraction and direct, receptor independent defensin like antimicrobial activity (X). It is in this dermal infection phase, where as in CHIKV infection fibroblasts located in the deep dermis and basal skin layer may become permissively infected, thus allowing viral replication (X) including other skin resident cells like macrophages, endothelial and muscle cells (X). Interestingly, experimental evidence using electron microscopy to visualize ZIKV infected skin fibroblasts has shown characteristic autophagosome like vesicles in these cells (X). Although autophagy is known to occur as a physiological response to cellular stress following virus amplification (X), recent evidence suggests that such process may serve viruses as a survival strategy to fu rther increase their replication supporting their life cycle (X). Virus induced induction of autophagy, also known as proviral autophagy seems to play a role in replication and translation processes of many arboviruses like DENV, JEV and CHIKV (X). To
96 furt her interrogate whether ZIKV induced autophagy, Hamel and collaborators used confocal microscopy to demonstrate coexpression of the viral envelope protein and the cytosolic microtubule associated light chain 3 (LC3) a mammalian homologue of yeast Apg8p which in LC3 colocalized with ZIKV viral envelope protein; thus confirming that ZIKV is able to trigger its replication by induction of autophagy in cutaneo us fibroblasts (X). 3.3.Melanocytes The role of melanocytes has not been explored in ZIKV infection. However, it is widely known that these specialized cells residing in the epidermis display important immune and metabolic functions capable of influencing response to pathogens as well as contributing to the clinical manifestations of cutaneous infections (X). From an immunological standpoint, melanocytes are equipped to stand on the forefront of an initial viral insult, based on their capacity to attract a nd recruit immune cells such as neutrophils, macrophages and lymphocytes (X), as well as to their ability to act as phagocytic, antigen processing and antigen presenting cells (X). In a similar fashion to keratinocytes and cutaneous fibroblasts, melanocyte s also express PRRs, which after recognition of specific PAMPs stimulate the production of type 1 IFNs (X) thus exhibiting an important role in containment of viral infections. On the other hand, melanocytes throughout the course of infection could also be responsible for the clinical manifestations of disease. Various viruses, including arboviruses such as the alphaviruses, are known to directly infect melanocytes disturbing their metabolic functions and thus originating the pigmentary changes seen usually in infants short after the vanishing of a maculopapular rash like in the case of CHIKV infection (X). One can speculate that the post
97 inflammatory hypopigmentation frequently seen days after the acute ZIKV rash in some patients would share the same pathol ogical basis in the melanocyte. 4. Clinical Features As revealed by sero epidemiological surveys ZIKV human infections are usually asymptomatic. The first descriptive case of the disease was reported in detail in 1956 after experimental inoculation in a h ealthy volunteer, who developed headaches 82 hours post inoculation accompanied by fever that lasted 48 hours with no associated cutaneous signs or symptoms (X). Later in 1964, the disease was reported on another individual after occupational exposure and who did exhibit a characteristic maculopapular rash involving the face, neck, trunk, palms and soles, 24 hours after the onset of headaches and other non specific symptoms such as h Polynesian outbreaks in 2007 that the classic combination of fever, rash, arthritis and/or arthralgia and/or myalgia, conjunctivitis, and fatigue, would come to define the most common clinical picture of the disease (X). ZIKV infection usually recreate s an influenza like illness that is difficult to differentiate from other arboviral (Dengue, Chikungunya) or exanthematic (Measles, Rubella) viral diseases (X); and when clinically apparent, often exhibit mild forms of the disease (X). The incubation perio d usually ranges from 3 to 10 days with the duration of illness lasting about a week (X). In light of the most recent epidemic, the Pan American Health Organization (PAHO) issued interim case definitions based on data obtained from the current epidemic in the Region of the Americas (PAHO). Thus, a suspected case is a patient with a rash (usually pruritic and maculopapular) with two or more of the following signs or symptoms: an elevated body temperature (<38.5C),
98 arthralgia or myalgia, nonpurulent conjunct ivitis or conjunctival hyperemia, headache or malaise and peri articular edema (PAHO). A confirmed case is a suspected case with a positive laboratory confirmation of ZIKV (PAHO). The cardinal cutaneous manifestation of ZIKV infection is the maculopapular rash and prurtitis (X). However, in our experience there is a marked diversity in the characteristics of the rash as well as the severity of illness, ranging from a conspicuous, diffuse mildly pruritic maculopapular rash to cases with nearly universal erit hrodermia. In the middle of the disease spectrum morbiliform like rashes and exanthem like eruptions with predominance of macules, plaques and patches (like those seen in CHIKV infection) are often seen. As opposed to CHIKV and DENV where the rash occurs g enerally after the fourth day of onset of symptoms, in ZIKV, cutaneous manifestations occur commonly in the first 24 to 48 hours after the onset of symptoms in over 90% of the cases (X). Individual lesions are usually macules, papules, and plaques that can even appear as wheals. They are usually erythematous, round to oval, and arranged in combination patterns (maculo papular most commonly) (X) or exhibiting a reticular (linear and net shaped) appearance (X), and blanching on palpation (X). As for the distr ibution, lesions are usually generalized following a symmetrical pattern that commonly involves face, neck, trunk, palms and soles (X). In Venezuela however (data not published), a distinct pattern characterized by accentuation of the lesions in proximal a reas of lower and upper limbs, neckline and abdomen has been observed in the majority of patients. Painful periarticular edema of the joints is also a distinctive sign (PAHO), occurring most commonly in small joints (X) of wrists and ankles, often in a sym metrical fashion.
99 Atypical cutaneous manifestations have also been described in ZIKV infection. Karimi et al. recently reported a case of ZIKV immune mediated thrombocytopenia in a returning traveler from Surinam who developed generalized pruritus and a m aculopapular rash 11 days after, subsequently developing swelling of the hand and wrists with subcutaneous haematomas on both arms and legs without preceding trauma (X). Similarly, Sharp et al. have reported the occurrence of thrombocytopenia in two patien ts exhibiting petechial and ulcerous lesions of the tongue and oral mucosa as well as ecchymotic lesions of the upper limb (X). Although uncommon, the occurrence of jaundice during ZIKV infection was also reported during the initial description of human ca ses in Nigeria in 1952 (X). In a large study carried out on 72 ZIKV infected pregnant woman in Rio de Janeiro (Brazil), pruritus was present in 94% of the patients along with a maculopapular rash (X), highlighting the importance of cutaneous involvement in suggesting the possibility of ZIKV infection. Mucocutaneos and ophthalmologic involvement is not uncommon in ZIKV infected patients. Conjunctivitis is considered one of the most common signs (X), it is often non purulent and frequently described as a conj unctival hyperemia (X). On the other hand, the occurrence of petechial lesions in the hard palate has been described (X) and also often seen by us in our ZIKV infected patients. Aphtous ulcers have also been reported to occur (X). One interesting aspect th at is worth commenting from our field experience in Venezuela is that most of the patients who usually exhibited severe, extensive rash unequivocally recalled having a previous episode of flavivirus infection (mostly Dengue), which is endemic in our countr y. It can be speculated that such phenomena could reflect the presence of enhancing antibodies causing an anamnestic response due to a previous flavivirus infection.
100 Also, the occurrence of psoriatic like lesions in patients with no personal or family hist ory of Psoriasis, weeks after acute ZIKV infection, is an interesting finding that prompts further and more elaborated studies. It is possible that genetic changes in ZIKV as noted previously for other arboviruses could be responsible for phenotypic change s influencing virulence and clinical outcome of some of these patients. 5. Differential Diagnosis Considerations on the differential diagnostic approach, not only of ZIKV infection but other arbovirosis is challenging and must be based not only the type o f cutaneous lesions but also on systemic signs and symptoms. An in depth understanding on the natural history of arboviral infections is pivotal since incubation period and onset of symptoms may differ widely or even overlap in some circumstances. For exam ple, the incubation period for ZIKV, CHIKV and DENV may significantly superpose with a time range of 5 to10 days. Knowledge on the endemic epidemiology is also an essential tool for diagnosis, because geographic restriction, seasonal behavior and distribut ion of certain viruses may hold the clue to narrow the list of suspects. It is also important to consider that the presence of vectors is highly influenced by climate and that vectorial transmission is crucial in the life cycle of these viruses. A wide var iety of infectious and non infectious entities can course with fever and rash making diagnosis particularly challenging, with viral agents being the most common etiological source. Viral agents that should be considered in the list of differentials of ZIKV infection like rash include: Arboviruses (WNV, CHIKV, DENV and MAYV), parvovirus, human herpesviruses, enteroviruses, rubella and measles, amongst others.
101 West Nile virus infection can exhibit a maculopapular rash, however lesions tend to be punctate, les s confluent and more pronounced in the extremities (X). Neurological symptoms along with more severe systemic signs are important clues to its diagnosis. CHIKV on the other hand can reveal a myriad of cutaneous symptoms, being the morbiliform (maculopapula r) rash its most common skin manifestation (X). The rash most commonly affects trunk and limbs and to a lesser extent face, palms and soles (X). Pruritus and desquamation may occur, and recrudescence of the rash days after the onset of initial symptoms ha s been described (X). In addition, CHIKV may also course with vesiculobullous, ecchymotic, and vasculitic like purpuric lesions (X), as well as with transient nasal erythema, anogenital aphtous ulcers, generalized erythroderma and diffuse post inflammatory hyperpigmentation (X). Tenderness and edema of hands and feet (X) is a shared feature with ZIKV. The occurrence of severe constitutional DENV also exhibits a wide spectrum of s ymptoms that range from asymptomatic or mild undifferentiated febrile viral symptoms with or without morbiliform rash to sudden onset of fever, myalgias / arthralgias, retroorobital pain along with a diffuse maculopapular rash that may be pruritic and desq uamate (X). However, in severe cases, skin hemorrhages and petechiae are usually observed (X). Erythema infectiosum due to Parvovirus, also known as the fifth disease can present as a generalized confluent maculopapular eruption, however a key to its diagn osis is the intermittent nature of the rash which can last up to 3 weeks (X) and the accentuation of the eruption in cheeks (X). An important aspect to highlight is that in adults, Parvovirus B19 may cause arthralgias akin to those of caused by arthritogen ic alphaviruses like CHIKV (X).
102 Human herpesviruses should also be considered in the list of differential diagnosis. In particular, human herpesvirus type 6 which causes a very similar maculopapular eruption to ZIKV, known as exanthema subitum or roseola infantum (X) lasting similarly 24 to 72 hours (X). Other members of the herpesviridae family like Cytomegalovirus (CMV) and human herpesvirus type 7 can also cause a nonspecific morbiliform or urticariform eruption that may pose a challenge at the time of diagnosis (X). Epstein Barr virus (a Gammaherpesvirinae ) is often associated with a maculopapular or urticarial rash; however it commonly is accompanied by classic symptomatology of infectious mononucleosis (tonsillitis, pharyngitis, lymph node enlargement and visceromegalys) often occurring after the administration of ampicillin (X). In addition petechial lesions, and occasionally jaundice may occur as well (X). Petechial lesions of the hard palate are indistinguishable from those seen in ZIKV. In endemic areas ZIKV infection remains largely a childhood illness being often confused with other viral exanthems of childhood such as measles and rubella (X). Measles can be characteristically differentiated from a clinical perspective based on its well recognize d prodromic phase (the triad of cough, coryza and conjunctivitis) and the appearance of erythematous facial macules and patches early in the course of disease that later follow a cephalocaudal confluent spread to trunk and extremeties (X). Rubella (German measles) is of utmost importance being perhaps the most significant entity to consider in the differential when facing a pregnant patient proceeding from endo epidemic areas of ZIKV. This is due to the significant overlap of clinical symptoms and their s hared potential for vertical transmission and for causing congenital rubella and/or ZIKV syndrome (X).
103 Non polio enteroviruses are also a cause of nonspecific exanthems, describing a variety of patterns such as scarlatiniform, urticarial, zosteiform and ve sicular forms (X). Echoviruses, in particular Echovirus type 6 and 9 have been associated with variable rubelliform or morbilliform eruptions and low grade fever that initially involve the face to later extend caudally to trunk and limbs (X). Echovirus 16 in particular causes a rubelliform like eruption with discrete pink red macules covering face and neck to the upper trunk and extremities, this characteristic roseola Other less likely diagnoses that may resemble the ZIKV maculopapular rash include: adenoviral infection, unilateral laterothoracic exanthema, scarlet fever, rickettsial diseases (endemic typhus and Rocky Mountain spotted fever), Q fever, erlichiosis (X), Reovirus infection (X), Barma h forest virus (X), Ross river virus (X) and more rarely the new and old world hemorrhagic fevers (X). The serologic approach to flavivirus diagnoses is often equivocal due to group specific, complex specific and subtype specific cross reactivities determ ined by the different domains of the envelope (E) protein. 6.1. Histopathology Arboviral infections in general are frequently associated with a spectrum of cutaneous symptoms that can range from mild to severe clinical manifestations. However, from a histo logical standpoint changes in this group of infections have been poorly documented and are usually non specific, frequently exhibiting a perivascular lymphocytic cell infiltrate (X). Nevertheless, in a recent experimental study performed to characterize th e biology of ZIKV infection in human skin cells, Hamel et al. described a number of distinct histopathological features which include: Cytoplasmic keratinocyte vacuolation with presence of pyknotic nuclei, usually confined to the stratum granulosum, as wel l as the sporadic occurrence of edema which
104 is usually limited to the subcorneal layer (X). More recently, our group analyzed biopsies from group findings, while p roviding further insights into pathological findings. The histological features vary according to the nature of the cutaneous lesion. Biopsies obtained from a classic erythematous maculo papular rash usually reveal a non specific lymphocytic dermal infiltr ate, often perivascular and superficial. Biopsies from macular and patch like lesions usually show slight acanthosis, focal spongiosis and in three cases we observed erythrocyte extravasation and slight papillary dermal edema. Variable degree of spongiosis focal exocytosis of lymphocytes into the epidermis and focal dyskeratosis were observed in confluent macular lesions (mimicking those observed in CHIV). Conclusions: Lessons learned from what has been observed in the recent outbreak in the Americas have been useful in guiding efforts to help recognize from a clinical standpoint ZIKV disease and its potential complications. Despite a significant clinical overlap with other viral exanthems, the skin still holds pathognomonic clues in recognizing ZIKV cutan eous disease.
105 LIST OF REFERENCES Alizon, S. and van Baalen, M. 2005. Emergence of a convex trade off between transmission and virulence. American Naturalist, 165: E155 -E167. Balm MN, Lee CK, Lee HK et al. A diagnostic polymerase chain reaction a ssay for Zika virus. J Med Virol 2012; 84:1501 5. Bangham, J., Kim, K.W., Webster, C.L. and Jiggins, F.M. 2008. Genetic variation affecting host parasite interactions: Different genes affect different aspects of sigma virus replication and transmission in Drosophila melanogaster Genetics, 178: 2191 2199. Bangham, J., Obbard, D.J., Kim, K.W., Haddrill, P.R. and Jiggins, F.M. 2007. The age and evolution of an antiviral resistance mutation in Drosophila melanogaster Proceedings of the Royal Society Biol ogical Sciences, 274: 2027 2034. Bjornstad, O.N., Finkenstadt, B.F. and Grenfell, B.T. 2002. Dynamics of measles epidemics: Estimating scaling of transmission rates using a Time series SIR model. Ecological Monographs, 72: 169 184. Borer, E.T, E.W. Seab loom, C.E. Mitchell, A.G. Power. 2010. Local context drives infection of grasses by vector borne generalist viruses. Ecology Letters 13(7): 810 818. Brun, G. and Plus, G. 1980. The Viruses of Drosophila (trans.) A. Ashburner and T.R.F. Wright. In The Ge netics and Biology of Drosophila (A. Ashburn er and T.R.F. Wright, eds), The Genetics and Biology of Drosophila, pp. 625 702. London: Academic Press. Buckner, E.A., B.W. Alto, L.P. Lounibos. 2013. Vertical Transmission of Key West Dengue 1 Virus by Aedes aegypti and Aedes albopictus (Diptera: Culicidae) Mosquitoes from Florida. Journal of Medical Entomology. 50(6): 1291 1297. Carpenter, J., Hutter, S., Baines, J.F., Roller, J., Saminadin Peter, S.S., Par sch, J. and Jiggins, F.M. 2009. The transcription al response of Drosophila melanogaster to infection with the sigma virus (rhabdoviridae). Plos One, 4. Sedlacek, W. E. (1999). Black students on White campuses: 20 years of research. Journal of College Student Development, 40 (5), 538 549. Carpenter, J.A., Obbard, D.J., Maside, X. and Jiggins, F.M. 2007. The recent spread of a vertically transmitted virus through populations of Drosophila melanogaster Mol. Ecol., 16: 3947 3954. Combes, C. 2001. Parasitism: The Ecology and Evolution of Intimate Interactio ns. Chicago: University of Chicago Press.
106 Crowder, DW, EA Dykstra, JM Brauner et al. (2013) West Nile Virus Prevalence across Landscapes Is Mediated by Local Effects of Agriculture on Vector and Host Communities. PLoS One. http://dx.doi.org/10.1371/journal.pone.0055006 Dahari H., R.M. Ribeiro, C.M. Rice, A.S. Perelson. 2007. Mathematical Modeling of Subgenomic Hepatitis C Virus Replication in Huh 7 Cells Journal of Virology 81(2): 750 76 0. Dejnirattisai, W, P. Supasa, W. Wongwiwat, A. Rouvinski et al. (2016) Dengue virus sero cross reactivity drives antibody dependent enhancement of infection with zika virus. Nature Immunology 17, 1102 1108 Ebert, D. 1999. The evolution and expression of parasite virulence. In Evolution in Health & Disease (S.C. Stearns, ed), pp. 161 172. New York: Oxford University Press. Ebert, D. and Bull, J.J. 2003. Challenging the trade off model for the evolution of virulence: is virulence management feasible? Tr ends in Microbiology, 11: 15 20. Flix, R., Guzmn, J. and de Garay Arellano, A. 1971a. Distribution of CO2 sensitivity (sigma virus) in an urban population of D. melanogaster from Mexico City. II. Low dispersal, a factor which explains differences among locations. Drosophila Information Service, 47: 105 109. Flix, R., Gzman, J. and de Garay Acellano, A. 1971b. Carbon dioxide sensitivity of Drosophilid flies from a location in the outskirts of Mexico City. Drosophila Information Service, 47: 110 112. Ferguson, J.M. and J.M. Ponciano. 2014. Predicting the process of extinction in experimental microcosms and accounting for interspecific interactions in single species time series. Ecology Letters 17: 251 259. Ferguson, J., Carvalho, F., Murillo, O. a nd J.M. Ponciano. 2015a. Population growth, density dependence and evironmental autocorrelation in animal populations. Theoretical Ecology, 8(2):doi 10.1007/s12080 015 0276 6 Ferguson, J.M. and J.M. Ponciano. 2015b. Evidence and implications of higher order scaling in the environmental variation of animal population growth. Proceedings of the National Academy of Sciences. 112(9): doi: 10.1073/pnas.1416538112. Fine, P.E.M. 1975. Vectors and vertical transmission: an epidemiological perspective. Annals of the New York Academy of Sciences, 266: 173 194. Fleuriet, A. 1976. Presence of the hereditary rhabdovirus sigma and polymorphism for a gene for resistance to this virus in natural populations of Drosophila melanogaster Evolution, 30: 735 739.
107 Fle uriet, A. 1980. Polymorphism of the hereditary sigma virus in natural populations of Drosophila melanogaster Genetics, 95: 459 465. Fleuriet, A. 1981. Comparison of various physiological traits in flies ( Drosophila melanogaster ) of wild origin, infecte d or uninfected by the hereditary Rhabdovirus sigma. Arch. Virol., 69: 261 272. Fleuriet, A. 1996. Polymorphism of the Drosophila melanogaster Sigma virus system. J ournal of Evolutionary Biology 9: 471 484. Gay, P. 1978. Les genes de la drosophile qu i interviennent dans la multiplication du virus sigma. Mol. Gen. Genet., 159: 269 283. Goddard, L.B., Roth, A.E., Reisen, W.K. and Scott, T.W. 2003 Vertical transmission of West Nile virus by three California Culex (Diptera : Culicidae) species. Journa l Of Medical Entomology, 40: 743 746. Goldstein, L. 1949. A mutant genoid in Drosophila that is sensitive to Carbon Dioxide. Nature, 164: 369 369. Hotez P.J., M.G. Basez, A. Acosta Serrano, M.E. Grillet. 2017. Venezuela and its rising vector borne neglected diseases. PLoS Negl Trop Dis 11(6): e0005423. https://doi.org/10.1371/journal.pntd.0005423 Inouye, B.D. 2005. The importance of the variance around the mean effect size of ecological processes: Comment. Ecology, 86: 262 265. Jones, K.E., N.G. Patel, M.A. Levy, A. Storeygard, D. Balk, J.L. Gittleman, P. Daszak. 2008. Global trends in emerging infectious diseases. Nature 451 (7181): 990 993 Lafferty, K. D., A.M. Kuris 2009. Parasitic castration: The evolution and ecology of body snatchers. T rends in Parasitology, 25, 564 572 Lee, KP, SJ Simpson, K Wilson. Dietary protein quality influences melanization and immune function in an insect. Functional Ecology 2008, 22, 1052 1061 Lloyd Smith, J.O., George, D., Pepin, K.M., Pitzer, V.E., Pulliam J.R.C., Dobson, A.P., Hudson, P.J., Grenfell, B.T. 2009. Epidemic dynamics at the human animal interface. Science 326(5958): 1362 1367. L'Hritier, P. 1958. The hereditary virus of Drosophila. Advances in Virus Research, 5: 195 245. L'Hritier, P. 197 0. Drosophila viruses and their role as evolutionary factors. Evol. Biol., 4: 185 209.
108 L'Hritier, P. and Teissier, G. 1945. Transmission hrditaire de la sensibilit au gaz carbonique chez la Drosophile. Publ. lab. ENS Biologie, 1: 35 76. Longdon, B ., Obbard, D.J. and Jiggins, F.M. 2010. Sigma viruses from three species of Drosophila form a major new clade in the rhabdovirus phylogeny. Proceedings of the Royal Society Biological Sciences, 277: 35 44. Lord, C. C., B.W. Alto, S.L Anderson, C.R. Conne lly, J.F. Day, S.L Richards, C.T Smartt, W.J. Tabachnik. 2014. Can Horton Hear the Whos? The Importance of Scale in Mosquito Borne Disease. Journal of Medical Entomology, 51(2): 297 313. Mangin, K.L., Lipsitch, M. and Ebert, D. 1995. Virulence And transm ission modes of 2 microsporidia in Daphni a Magna. Parasitology, 111: 133 142. Mayer, S.V, R.B. Tesh, N. Vasilakis. 2017. The emergence of arthropod borne viral diseases: A global prospective on dengue, chikungunya and zika fevers. Acta Tropica 166: 155 1 63 Mishra, A.C. and Mourya, D.T. 2001. Transovarial transmission of West Nile virus in Culex vishnui mosquito. Indian Journal of Medical Research, 114: 212 214. Morris, W.F. 1997. Disentangling effects of induced plant defenses and food quantity on her bivores by fitting nonlinear models. American Naturalist, 150: 299 327. Murray, K.O., Mertens, E. and Despres, P. 2010. West Nile virus and its emergence in the United States of America. Veterinary Research, 41. Padmanabhan P, NM Dixit. 2011. Mathematic al Model of Viral Kinetics In Vitro Estimates the Number of E2 CD81 Complexes Necessary for Hepatitis C Virus Entry. PLoS Computational Biology 7(12): e1002307. https://doi.org/10.1371/journal.pcbi.1002307 Patterson, J., M. Sammon, M. Garg 2016. Dengue, Zika and Chikungunya: Emerging Arboviruses in the New World. West J Emerg Med. 17(6): 671 679. Peck, GW and WE Walton (2005) Effect of Different Assemblages of Larval Foods on Culex quinquefasciatus and C ulex tarsalis (Diptera: Culicidae) Growth and Whole Body Stoichiometry Environmental Entomology 34(4):767 774. 2005 Perera, R, M Khaliq, RJ Kuhn. (2008) Closing the door on flaviviruses: Entry as a target for antiviral drug design. Antiviral Research 80 :11 22 Read A. F. (1994). The evolution of virulence. Trends in Microbiology 2:73 76 Reisen, WK. (2013) Ecology of West Nile Virus in North America. Viruses, 5, 2079 210 Romero Brey, I. and R. Bartenschlager 2016. Endoplasmic Reticulum: The Favorite Int racellular Niche for Viral Replication and Assembly. Viruses 8(6): 160.
109 Root, J.J. 2013. West Nile Virus associations in wild mammals: a synthesis. Arch ives of Virol ogy 158:735 752 Sall, A.A., Faye, O., Diallo, M., Firth, C., Kitchen, A. and Holmes, E. C. 2010. Yellow fever Virus exhibits slower evolutionary dynamics than dengue virus. Journal of Virology, 84: 765 772. Sterner, R. W., & Elser, J. J. (2002). Ecological stoichiometry: The biology of elements from Molecules to the biosphere. Princeton: P rinceton University Press. Turell MJ, Dohm DJ, Sardelis MR, Oguinn ML, Andreadis TG, Blow JA. An update on the potential of north American mosquitoes (Diptera: Culicidae) to transmit West Nile Virus. J Med Entomol. 2005;42(1):57 62. Vieira de Morais Bro nzoni, R, FG Baleotti, RM Ribeiro No gueira, M Nunes, L Tadeu Moraes Figueiredo (2005) Duplex Reverse Transcription PC R Followed by Nested PCR Assays for Detection and Identification of Brazilian Alphaviruses and Flaviviruses Journal of Clinical Microbiol ogy: 43(2); 696 702 WHO. Zika virus, Microcephaly, and Guillain Barr Syndrome Situation Report. 25 August 2016.WebLink Wayne, M.L., Contamine, D. and Kreitman, M. 1996. Molecular population genetics of ref(2)P, a locus which confers viral resistance i n Drosophila. Mol. Biol. Evol., 13: 191 199. Wayne, M.L., Telonis Scott, M., Bono, L.M., Harshman, L., Kopp, A., Nuzhdin, S.V. and McIntyre, L.M. 2007. Simpler mode of inheritance of transcriptional variation in male Drosophila melanogaster Proceedings of the National Academy of Sciences of the United States of America, 104: 18577 18582. Weaver, S.C., W.K. Reisen 2010 Present and Future Arboviral Threats. Antiviral Research 85(2): 328. Wilfert, L. and Jiggins, F.M. 2010a. Disease association mapping in Drosophila can be replicated in the wild. Biology Letters, 6: 666 668. Wilfert, L. and Jiggins, F.M. 2010b. Host parasite coevolution: genetic variation in a virus population and the interaction with a host gene. Journal of Evolutionary Biology, 23: 1447 1455. Williamson, D.L. 1961. Carbon dioxide sensitivity in Drosophila affinis and D. athabasca. Genetics, 46:1053 1060.
110 Yampolsky, L.Y., Webb, C.T., Shabalina, S.A. and Kondrashov, A.S. 1999. Rapid accumulation of a vertically transmitted parasi te triggered by relaxation of natural selection among hosts. Evolutionary Ecology Research, 1: 581 589 Zambrano Y, A Chiarello, A Soca, I Villalobos, M Marrero, M Soler, J Laferte, M Alvarez. Use Of polymerase chain reaction for the diagnosis of central nervous system infections. Invest Clin 2006; 47(4):337 47 Zlotnick, A, S. Mukhopadhyay 2011. Virus assembly, allostery, and antivirals. Trends in Microbiology. 19(1): 14 23.
111 BIOGRAPHICAL SKETCH Ga briela Blohm earned her BSc in wildlife ecology, and MS c in z oology at the University of Florida. She has studied a broad range of ecological systems and will be directing her research efforts towards the control of vector borne diseases in her home country. She hopes to continue her career as a research biolo gist with a focus on public health and wildlife disease surveillance in Venezuela. Her long term goal is to work with biologists, epidemiologists and health care practitioners to develop strategies for improving public health in Venezuela, with a focus on reducing the spread of mosquito borne viruses in areas where water quality control measures are not in place and where fertilizer use is unregulated. She also plans to continue working with her family on the management of a cattle ranch and research stati on in central Venezuela Researchers and staff at Masaguaral, the ranch, have continued a 42 year long research program that has resulted in more than 500 peer reviewed scientific publications and 30 years of hands on education for students in veterinary m edicine and crocodilian conservation.