*3.2. ZIKV Viral Shedding*

Guinea pigs infected SQ shed infectious ZIKV in tears from 7-17 dpi, with peak shedding at 10 dpi (up to 10 PFU), although VAG-infected guinea pigs did not have detectable infectious ZIKV in tears (Figure 2A). Shedding of infectious ZIKV in tears after SQ injection (footpad) in immunocompromised or immunosuppressed mice has been demonstrated by Miner et al., who also noted the development of pan-uveitis [62]. Guinea pigs infected VAG shed infectious ZIKV in vaginal secretions throughout the study period, from 1–21 dpi, with peak shedding at 3 dpi (up to 425 PFU) (Figure 2B). The early peak of viral shedding in vaginal secretions at 3 dpi is consistent with local replication of the virus in vaginal or cervical tissues, prior to hematogenous spread to secondary tissues. Similar results have been documented in humans where infectious ZIKV has been recovered from vaginal secretions up to 10–14 days post symptom onset (pso) [28,63], although vaginal swabs have not been assessed the first few days following infection in humans to determine if very early viral shedding occurs. In contrast, no infectious virus was detected in vaginal secretions from guinea pigs inoculated SQ (Figure 2A). Overgrowth of the microflora of the oral cavity prevented successful plaque assay of oral swab samples.

**Figure 2.** Viral shedding in guinea pigs inoculated subcutaneous (SQ) and intravaginal (VAG) with 1x10<sup>6</sup> PFU ZIKV MR766. Virus titer, determined by plaque assay on Vero E6 cells, detected in: (**A**) tears; (**B**) vaginal secretions. Inset shows the graph with the Y-axis shifted to show average titer in vaginal secretions through 21 dpi in VAG infected animals.

## *3.3. ZIKV in Guinea Pig Tissues*

A low-level serum viremia (1.09 × 103–3.47 × 10<sup>3</sup> ZIKV RNA copies/500 ng total RNA) was detected in both groups of guinea pigs during acute infection (Figure 3). This low-level viremia was still present in both groups of guinea pigs during convalescence (3.25 × 101–1.36 × 10<sup>3</sup> ZIKV RNA copies/500 ng total RNA), although reduced in SQ-infected guinea pigs. Wide variability was detected among all animals; thus, no statistically significant differences in viremia were found between the guinea pig groups during acute or convalescent time points. ZIKV RNA copy numbers in all tissues tested, other than the cerebellum of SQ-infected guinea pigs during the acute phase, were higher than serum viremia at both time points (7 dpi and 37 dpi) in both groups (Figure 3).

**Figure 3.** Viral load detected by RT-qPCR in serum and tissues of guinea pigs inoculated subcutaneously (SQ) or vaginally (VAG) with ZIKV at acute (7 dpi) and convalescent (37 dpi) time points.

Similar ZIKV RNA loads were found in tissues of guinea pigs following SQ and VAG infection (Figure 3). Viral load detected during acute infection (7 dpi) did not change substantially by the convalescent time point (37 dpi), suggesting an extended period of time is required for clearance of viral RNA even though no signs of disease were present at the end point of the study. However, an increase in ZIKV RNA copies was detected in the spleens of VAG-infected guinea pigs from acute (2.41 × 107) to convalescent (1.14 × 108) time points, and a one-log higher copy number of ZIKV was detected in the spleens of VAG compared to the SQ group (*p* = *0.0141*) at the convalescent time point. ZIKV RNA detected in lymph nodes was similar to that found in spleens (Figure 3), although no statistically significant differences were found.

The ovaries and uteri of both guinea pig groups contained ZIKV during both time points (Figure 3). Surprisingly, there was no statistically significant difference between the guinea pig groups at either time point due to inoculation route. Greater ZIKV copy numbers were noted in convalescent uteri (9.59 × 10<sup>9</sup> SQ, 5.75 × 10<sup>9</sup> VAG) than convalescent ovaries (7.65 × 10<sup>7</sup> SQ, 1.87 × 10<sup>8</sup> VAG) in both groups (p = 0.025 SQ; p = 0.021 VAG), suggesting a higher level of ZIKV persistent replication in the uterus during infection. The pituitary glands in each group supported robust ZIKV replication during acute infection (8.06 × 10<sup>9</sup> SQ and 4.07 × 10<sup>9</sup> VAG,), which continued into the convalescent period in both groups (5.37 × 10<sup>9</sup> SQ and 3.76 × 10<sup>9</sup> VAG), suggesting that ZIKV infection may affect glandular function.

ZIKV antigen was detected by immunofluorescence (IF) staining in the spleen and cervical lymph nodes at both acute and convalescent time points in SQ- and VAG-infected guinea pigs, with minimal visible differences between the groups (Figure 4). At the convalescent time point, ZIKV antigen was detected in diffuse vibrant clusters of cells, as opposed to distinct individual cells at the acute time point, suggesting possible cell-to-cell spread within lymph nodes (Figure 4).


**Figure 4.** Representative immunofluorescence images of spleens, lymph nodes (LN), genitourinary tracts (GUT), and eyes from guinea pigs inoculated SQ or VAG with ZIKV at acute (7 dpi) and convalescent (37 dpi) time points. GUT images are from the uterus, which was the only part of the GUT we identified as positive for ZIKV antigen. Eye images show ciliary bodies, which were positive for ZIKV antigen.

The genitourinary tract (GUT) of both groups were IF stained for ZIKV to determine if ZIKV persisted in tissues through the convalescent time point. ZIKV antigen was detected in the GUT, specifically in the uterine wall, with virus persisting into the convalescent time point (Figure 4). At 7 dpi, ZIKV antigen was found localized to the uterus after VAG infection. ZIKV antigen was noted to be heaviest in VAG-infected guinea pigs at both time points compared to SQ-infected guinea pigs.

ZIKV antigen was observed in the eye of both SQ and VAG infected animals at both time points, but most prominently in the ciliary body of SQ infected guinea pigs during acute infection (Figure 4). Although present in VAG-infected animals, ZIKV antigen was noted to be heaviest in SQ infected guinea pigs compared to VAG-infected guinea pigs. This comports with our finding that guinea pigs infected SQ shed infectious ZIKV in ocular secretions. Similar immunofluorescent localization of ZIKV in the ciliary body has been reported in immunocompromised mouse models [64]. Although we did not detect substantial histopathology consistent with inflammation in the eyes of our adult guinea pigs, histopathological signs of inflammation have been localized to the ciliary bodies of fetal rhesus macaques (vertical transmission) and in a case series assessing congenital ZIKV syndrome in humans [65]. It is worth noting that no infectious virus was recovered from ocular secretions at acute or convalescent time points in VAG-infected guinea pigs, even though ZIKV viral RNA was detected in the eyes of this group, although less than after SQ infection. This suggests the possibility that a more robust viral replication is occurring in tissues that provide ocular secretions, such as the lacrimal glands.

#### *3.4. ZIKV in Guinea Pig Nervous Systems*

Sensory ganglia (LS-DRG, C-DRG, and TGs) of the peripheral nervous system of both groups supported stable and persistent ZIKV RNA at both time points, with no significant di fferences between routes of infection. A slightly lower ZIKV copy number was detected in TGs from VAG- infected guinea pigs at the convalescent time point, although the results did not reach statistical significance (Figure 5).

**Figure 5.** Viral load detected by RT-qPCR in nervous system of guinea pigs inoculated subcutaneously (SQ) or vaginally (VAG) with ZIKV at acute (7 dpi) and convalescent (37 dpi) time points. LS-DRG (lumbosacral dorsal root ganglia, sensory), C-DRG (cervical dorsal root ganglia, sensory), TG (trigeminal ganglia, sensory), SCG (superior cervical ganglia, sympathetic), CG (ciliary ganglia, parasympathetic).

Autonomic ganglia (SCG and CG) also contained similar ZIKV copy numbers following SQ and VAG infection, detectable during both acute and convalescent time points (Figure 3). These ganglia, in both groups, at both time points, supported ZIKV copy numbers slightly higher than those detected in sensory ganglia (1.24 × 109–5.59 × 10<sup>9</sup> RNA copies/500ng total RNA). There was no significant di fference between ZIKV copy numbers in sympathetic (SCG) versus parasympathetic (CG) ganglia.

Structures throughout the brains of both SQ and VAG infected guinea pigs had high levels of detectable ZIKV RNA at acute and convalescent time points (Figure 5). ZIKV RNA copies decreased approximately 10-fold from acute to convalescent time points in brainstem (1.1–3.3 × 10<sup>9</sup> acute to 2.1-7.5 × 10<sup>8</sup> convalescent), midbrain (3.9 × 108–2.0 × 10<sup>9</sup> acute to 2.0 × 107–2.2 × 10<sup>8</sup> convalescent), and forebrain (1.6–2.6 × 10<sup>9</sup> acute to 23.2 × 107–1.3 × 10<sup>8</sup> convalescent). In contrast, the cerebellum had fewer ZIKV RNA copies during acute infection (7.2 × 10<sup>5</sup> SQ and 8.6 × 10<sup>6</sup> VAG) compared to the convalescent time point (2.4 × 10<sup>8</sup> SQ and 1.7 × 10<sup>9</sup> VAG). In fact, the cerebellum contained lower copy numbers of ZIKV RNA than any other region of the nervous system during acute infection.

We previously demonstrated that cultured primary adult mouse sensory DRGs became infected with ZIKV and persistently released infectious virus for at least five days without dying [32]. We also determined that satellite glial cells (SGCs) in those cultures became infected and were killed by the virus within 24 h post inoculation. Thus, we hypothesized that the LS-DRG may be an alternative reservoir of persistent virus that could be shed in genital secretions, since the LS-DRG innervates the GUT. Thus, we assessed the presence of ZIKV in the LS-DRG to determine if neurons or SGCs became persistently infected with ZIKV following SQ or VAG infection. We detected ZIKV antigen by IF in the LS-DRG in both groups (Figure 6). During acute infection (7 dpi), ZIKV was detected by immunofluorescence surrounding sensory neurons within the ganglia and co-localizing with satellite glial cell (SGC) marker glutamine synthetase (GS). However, ZIKV was not detected within the neurons themselves, which were visualized by immunofluorescence for sensory neuronal marker PGP9.5. ZIKV persisted within the ganglia through the convalescent time point, particularly in the VAG-infected animals. SGCs normally wrap around sensory neurons within the ganglion (see LS-DRG Uninfected GS in Figure 6) but in ZIKV infected animals, regardless of route of inoculation or time point, the morphology of SGCs was substantially altered compared to uninfected animals, suggesting viral destruction of the SGCs. In contrast to the findings of our in vitro study, the sensory neurons were not infected in LS-DRG. ZIKV antigen was also detected in the SGCs of the C-DRG following SQ infection, but not in neurons. ZIKV antigen was found in C-DRG in only one of the VAG-infected guinea pigs during the acute time point.

Within the central nervous system, ZIKV antigen was observed in the brains of both groups at both acute and convalescent time points. Infection was di ffuse throughout the brains of both groups, with fluorescence most notably in cortical tissues in the frontal lobe and in the hippocampus (Figure 7). Within each of these regions of the brain, isolated neurons identified as positive for ZIKV showed an altered morphology, substantially larger than nearby uninfected neurons and surrounded by a "halo". This localization of ZIKV has been observed in a rhesus macaque model of infection (pregnant mother and fetus), as well as in human fetuses [17,66]. Histopathology, suggesting ZIKV localization to cortical tissues, has also been demonstrated in numerous murine models [36,38,62].

**Figure 6.** Representative immunofluorescence images of lumbosacral dorsal root ganglia (LS-DRG) from uninfected guinea pigs, and guinea pigs inoculated subcutaneously (SQ) or vaginally (VAG) with ZIKV at acute (7 dpi) and convalescent (37 dpi) time points. Neurons were immunostained for the sensory neuronal marker PGP9.5 (green) or the satellite glial cell marker glutamine synthetase (GS, green), and ZIKV (red). Merged images are shown enlarged to show co-localization of ZIKV with satellite glial cells, not sensory neurons. Note the morphological changes of satellite glial cells between uninfected (upper right) and ZIKV infected guinea pigs.


**Figure 7.** Representative immunofluorescence images of brain, including cortex in the frontal lobe (Frontal) and hippocampus (Hippo) from uninfected guinea pigs and guinea pigs inoculated subcutaneously (SQ) or vaginally (VAG) with ZIKV at acute (7 dpi) and convalescent (37 dpi) time points. Neurons were immunostained for the neuronal marker NeuN (green) and ZIKV (red). Insets are 200% enlargements of infected neurons, depicted by the white arrow, showing altered morphology.
