**4. Discussion**

Numerous murine models exist for the study of ZIKV infection. However most of these models use immunodeficient or immunosuppressed mice, many lacking an intact IFN pathway, which may influence the pathogenesis of infection (e.g., severe viremia, disease, frequent death) in a manner not applicable to that found in immunocompetent humans (e.g., asymptomatic infection, self-limiting illness, rare death) [40,67–70]. Several immunocompetent non-human primate models have been used to study the pathogenesis of ZIKV infection, however these models are limited due to their prohibitive cost, resulting small sample sizes, reduced statistical power, and requirements for specialized facilities and staff [44–48]. Due to concerns of sexual and vertical transmission and neurological sequelae of ZIKV infection, an immunocompetent small animal model is warranted.

A guinea pig model of ZIKV infection presents an attractive alternative to the above models due to the physiologic similarities between humans and guinea pigs, which include reproductive physiology and estrous cycle, and homology between immune systems (major histocompatibility molecules (MHC), complement systems, IFNγ pathways, IL-8/12 receptors, and CD8 sequences) [51]. Other attractive characteristics of the guinea pig include the ability to establish infection in an immunocompetent host, general ease of handling and maintenance, utilization of larger sample sizes, and availability of immunological assays and techniques [51]. Additionally, guinea pigs have been used as reliable models of flavivirus infection (e.g., Japanese encephalitis virus), as well as for sexual transmission studies (herpes simplex virus) and congenital syndrome caused by vertical transmission (cytomegalovirus) [54,55,71–74]. In this study, we demonstrate successful infection and persistence of ZIKV in immunocompetent female Hartley guinea pigs after a physiologically relevant inoculation (1 × 10<sup>6</sup> PFU) in clinically significant routes of transmission (SQ, VAG), which mimic mosquito-borne and sexual transmission.

ZIKV infection was established in all guinea pigs regardless of route of infection. Minimal clinical signs of infection were observed, although subtle signs were noted, such as ear sensitivity, vocalization, and hyperactivity. This is consistent with human infection, as the majority (~80%) of ZIKV infections in humans are asymptomatic [75–78]. The classic maculopapular rash observed in some humans with symptomatic ZIKV infection was not observed in our guinea pigs, and has not been reported in other guinea pig models. Dermatological manifestations have only been reported in non-human primates around injection sites and more recently in tree shrews, although the tree shrews did not demonstrate any other signs of ZIKV infection such as fever or weight loss [45]. The mechanism for the development of skin rash associated with ZIKV infection is not fully understood. More severe/overt signs of disease have been elicited in guinea pigs by inoculation with more contemporary ZIKV strains (e.g., PRVABC59) [56], suggesting that these animals may be useful for investigating pathogenicity di fferences between ZIKV strains. We are the first to observe a five-day cyclical/undulating nature of clinical signs of infection, although the clinical significance of this observation relative to humans is not clear. Also, our studies are the first to compare SQ and VAG routes of infection, including the observations of the increase in vaginal discharge in animals inoculated vaginally, which may be a factor in potential sexual transmission of ZIKV.

A low-level viremia was detected via RT-qPCR in the serum of both groups throughout the study. Most current models were unable to detect viremia beyond 5 dpi; however, one model reported detection of low-level viremia up to 14 dpi, and one reported no detection at any time point [56–59]. Our detection of a sustained and persistent low-level serum viremia in both SQ and VAG infected groups up to 37 dpi represents the longest detection of serum viremia in a guinea pig model. This finding is consistent with detection of ZIKV viremia in whole blood samples in humans from 14 to 100 days, for a median duration of 22 days, while another serosurvey showed viremia for up to 8 weeks in some patients [79,80]. Localization of ZIKV replication to secondary lymphoid organs (spleen, lymph nodes), genitourinary tract (uterus, ovary), brain (brainstem, cerebellum, midbrain, forebrain, pituitary gland), and eyes agree with results reported in murine models, non-human primate models, and human case reports [35,37,40,44,56,57,65,81]. However, the only statistically significant di fference we identified in tissue tropism between SQ and VAG routes of infection, based on tissue viral loads, was in spleens at convalescence (VAG > SQ), suggesting that ZIKV may be cleared more quickly after mosquito-borne transmission than after sexual transmission. This is also supported by the higher levels of ZIKV RNA we detected in serum at the convalescent time point in VAG infected compared to SQ infected animals. It is interesting to note the lower ZIKV RNA copy number in the cerebellums in both groups compared to any other region of the brain during acute infection, indicating a delay in entry or replication in the cerebellum for an unknown reason.

Infectious ZIKV was recovered from vaginal secretions from 1-21 dpi and from tears 7-17 dpi, with peak recovery at 3 dpi and 10 dpi, respectively. Our recovery of ZIKV from secretions is consistent with the recovery of ZIKV from vaginal secretions in humans from 10-14 days pso and in tears up to 30 days pso [28,63,82]. Interestingly, ZIKV was recovered from vaginal secretions in only VAG-infected guinea pigs and from tears in only SQ-infected guinea pigs, even though we detected ZIKV antigens in similar tissues in both groups and found no statistically significant di fferences in ZIKV RNA copy number in ocular or genitourinary tissues between the groups. Additionally, vaginal secretions supported a longer period of viral shedding and higher viral titers during peak shedding than tears, which may indicate a more robust and prolonged viral replication locally in VAG-infected animals. The increased vaginal discharge we noted in VAG-infected guinea pigs may contribute to higher shedding rates of infectious ZIKV in this group compared to SQ-infected animals, which may increase the risk of sexual transmission.

With respect to ZIKV shedding in tears, Deng et al. also isolated infectious ZIKV from the tears of SQ- and intranasally-infected guinea pigs, demonstrating contact transmission of ZIKV between SQ-infected guinea pigs and their naïve cage mates, potentially mediated through viral shedding in tears [57]. Our recovery of infectious ZIKV in tears from 7-17 dpi (albeit at low titers), combined with the detection of ZIKV RNA in tears of both index and contact animals by Deng et al., provides additional evidence to sugges<sup>t</sup> contact transmission can occur in animal models by viral shedding in tears. As an extension of these findings, it is worth noting a case report in which contact transmission of ZIKV is suspected to have occurred between an elderly patient with a fatal ZIKV infection with high serum viremia and an otherwise healthy family member participating in his care who came into contact with the patient's tears while not wearing personal protective equipment, subsequently developing a maculopapular facial rash and ZIKV antigenuria [83]. Taken together, these results sugges<sup>t</sup> contact transmission of ZIKV via tears can occur between humans in certain rare instances.

We set out to determine if there were differences in tissue tropism and sites of ZIKV persistence that could contribute to extended periods of viral shedding in genital secretions that could contribute to sexual transmission, as well as vertical transmission. We had previously speculated that ZIKV persistence in LS-DRG, which innervate the GUT, may provide an alternative reservoir for viral shedding in genital secretions, particularly after sexual transmission [32]. We did not find significant differences in viral load or viral antigen in DRG following SQ or VAG infection, nor did we find significant differences in the GUT. However, we determined that ZIKV persists in the DRG and uterus following both SQ and VAG infection. To our knowledge this is the first in vivo animal model to investigate the role of sensory and autonomic ganglia in the maintenance of ZIKV infection. ZIKV antigen was localized to satellite glial cells (SGCs) surrounding the sensory neurons within LS-DRG in both groups during acute infection by 7 dpi, and the ganglia remained infected for at least 37 dpi. SGCs wrap around sensory neurons within the ganglia, providing support for the neurons as well as protection, forming a barrier between capillary endothelial cells and neurons within the ganglia and thus preventing access of blood-borne pathogens to the neurons. Based on our previous in vitro studies, in which SGCs were destroyed and naked neurons became persistently infected with ZIKV, we had anticipated that ZIKV would lytically infect SGCs, gaining access to sensory neurons within the DRG. DRGs from infected guinea pigs showed altered morphology and loss of SGCs surrounding sensory neurons within the ganglia, consistent with the destruction of SGCs we observed previously in primary DRG cultures [32]. However, the neurons themselves were not infected in either group. Detection of ZIKV antigen in SGCs, but not neurons of LS-DRG (or C-DRGs), suggests that ZIKV gains access to sensory ganglia through hematogenous dissemination but SGCs effectively prevent the virus from reaching and infecting the sensory neurons in vivo. Infection and destruction of SGCs surrounding neurons could disrupt synaptic transmission and potentially contribute to peripheral neuropathies. SGCs have been implicated in protection of neurons from blood-borne pathogens, as well as exacerbation of infection by mediating a robust inflammatory response within the ganglion. Although additional studies are needed to assess SGC survival, our results support a pathogenic model in which SGCs protect the sensory neurons from viremic ZIKV infection but contribute to viral spread and persistence in non-neuronal cells within the DRG. Additionally, our studies demonstrated ZIKV persistence in both sympathetic and parasympathetic autonomic ganglia. As autonomic ganglia innervate secretory glands and regulate release of secretions, persistence within autonomic ganglia also represents a previously undefined reservoir of persistent ZIKV that may contribute to viral shedding in secretions.

In summary, we sought to determine if route of infection influences pathogenesis of disease, tissue tropism and persistent reservoirs of ZIKV that may contribute to viral shedding. Although we did not identify differences in viral load or tissue tropism, route of infection contributed to substantial differences in viral shedding in secretions. Our studies support a pathogenic model in which ZIKV replicates locally at the site of infection, and then spreads hematogenously throughout the host. Following subcutaneous infection, simulating mosquito-borne transmission, ZIKV is more effective at shedding from ocular sections, although the site of persistence and mechanism are not completely clear. Following vaginal infection, local replication in the genitourinary tract induces increased vaginal secretions, which carry infectious virus that could contribute to sexual transmission. Since the majority of the ZIKV antigen that we found in the GUT after vaginal infection was localized to the uterine walls, the possibility exists that ZIKV sexual transmission may increase risk for the developing fetus during pregnancy. Further studies are needed to address pathogenic mechanisms of ZIKV by different routes of infection and the guinea pig model is well-suited for these endeavors.

**Author Contributions:** Conceptualization, A.E.S., A.S.B.; methodology, A.S.B.; validation, A.E.S., A.S.B.; formal analysis, A.E.S., J.D.J.; investigation, A.E.S., S.A.C., J.D.J., A.S.B.; data curation, A.S.B.; writing—original draft preparation, A.E.S.; writing—review and editing, A.E.S., S.A.C., J.D.J., A.S.B.; visualization, A.E.S., S.A.C., J.D.J., A.S.B.; supervision, A.S.B.; project administration, A.E.S., A.S.B.; funding acquisition, A.S.B.

**Funding:** This research received no external funding. AS and SC were participants in the Summer Veterinary Student Research Program and were funded by Virginia-Maryland College of Veterinary Medicine. The article processing charge is partially funded by the Virginia Tech Subvention Fund.

**Acknowledgments:** Guinea pigs for this study were generously provided by Charles River Laboratories. Special thanks to Hannah Ivester for staining uninfected tissues for Figure 4.

**Conflicts of Interest:** The authors declare no conflict of interest.
