Production of the cytokine VEGF-A by infiltrating CD4⁺ T and myeloid cells disrupts the corneal nerve landscape and promotes herpes stromal keratitis

Summary

Herpes simplex virus type 1 (HSV-1) infected corneas can develop a blinding immunoinflammatory condition called herpes stromal keratitis (HSK), which involves the loss of corneal sensitivity due to retraction of sensory nerves and subsequent hyperinnervation with sympathetic nerves. Increased concentrations of the cytokine VEGF-A in the cornea are associated with HSK severity. Here, we examined the impact of VEGF-A on neurologic changes that underly HSK using a mouse model of HSV-1 corneal infection. Both CD4⁺ T cells and myeloid cells produced pathogenic levels of VEGF-A within HSV-1 infected corneas, and CD4+ cell depletion promoted reinnervation of HSK corneas with sensory nerves. In vitro, VEGF-A from infected corneas repressed sensory nerve growth and promoted sympathetic nerve growth. Neutralizing VEGF-A in vivo using bevacizumab inhibited sympathetic innervation, promoted sensory nerve regeneration and alleviated disease. Thus, VEGF-A can shape the sensory and sympathetic nerve landscape within the cornea, with implications for the treatment of blinding corneal disease.

Graphical Abstract

Introduction

The cornea is a clear avascular tissue in the front of the eye. Herpes simplex virus type 1 (HSV-1) infected corneas can develop an immunoinflammatory process called herpes stromal keratitis (HSK) that is characterized by neovascularization, leukocytic infiltration, edema and progressive scarring. HSK is potentially blinding and has a complex pathogenesis. In mouse models, HSK is promoted by CD4⁺ T cells through elaboration of a variety of Th1 and Th17 cytokines (Bouley et al., 1995; Doymaz and Rouse, 1992; Hendricks et al., 1992; Lepisto et al., 2006; Molesworth-Kenyon et al., 2008; Newell et al., 1989a; Newell et al., 1989b; Suryawanshi et al., 2011b; Tang et al., 1997; Thomas et al., 1997; Verjans et al., 1998). It is widely assumed that these cytokines are produced by CD4⁺ T cells in response to specific HSV antigens, though proof of this concept has remains elusive. A key feature of HSK in both humans and mouse models is loss of corneal sensation, which reduces blink reflex in response to tactile or mechanical stimulation (Chucair-Elliott et al., 2015; Gallar et al., 2010; Norn, 1970; Rosenberg et al., 2002; Yun et al., 2014).

In HSK, loss of sensation and blink reflex are associated with reduced or complete loss of corneal sensory nerve endings that innervate normal corneas from the trigeminal ganglion (TG) (Hamrah et al., 2010; Muller et al., 2003; Yun et al., 2014). Blinking, among other functions, spreads tear film over the surface of the cornea, protecting it from exposure and desiccation stress (Nishida and Yanai, 2009; Oliveira-Soto and Efron, 2001; Wang et al., 2018). Desiccation stress can then lead to dry eye phenotypes, which can compromise the corneal barrier integrity resulting in microtraumas and increased susceptibility to infection. As inflammation associated with desiccation builds, blood vessels form, which deliver tissue disrupting monocytes, neutrophils, and eventually innate and adaptive T cells (Reyes et al., 2018). Inflammatory processes associated with these infiltrating leukocytes cause disruption of the corneal collagen network that contributes to the clarity of normal corneas and promotes corneal scarring. Recurrent reactivation of HSV-1 from a latent state and shedding into the cornea can cause progressive scarring and permanent loss of visual acuity. The only recourse at this stage is corneal transplantation.

The desiccating effects of reduced blink reflex can be circumvented by a surgical procedure called tarsorrhaphy, forming adhesion between the upper and lower lid margins. Performing tarsorrhaphy on mice with HSV-infected corneas significantly reduces, but does not eliminate corneal inflammation and the accompanying opacity associated with HSK (Yun et al., 2014), showing that the loss of corneal sensation and resulting desiccation stress contributes to HSK in the murine model. Notably, tarsorrhaphy does not prevent the loss of corneal sensory nerves but prevents the exacerbating inflammation associated with corneal desiccation. Thus, infected corneas exhibit mild HSK when protected from desiccation by tarsorrhaphy, but rapidly develop severe HSK when the eyelids are re-opened (Yun et al., 2014).

The cornea is the most densely innervated tissue in the body (Bonini et al., 2003) and is hundreds of times more sensitive to sensory stimuli than other tissues, such as the skin (Zander and Weddell, 1951). Unlike other tissues, the cornea is primarily innervated with sensory nociceptor afferents that respond with sensations of pain to various stimuli (Acosta et al., 2001; Belmonte and Giraldez, 1981; Tanelian and Beuerman, 1984). A temporary or permanent loss of corneal sensory nerves is relatively common during corneal inflammation associated with infections (Bonini et al., 2003; Cruzat et al., 2010; Cruzat et al., 2011; Hamrah et al., 2010; Kurbanyan et al., 2012; Muller et al., 2015), as well as autoimmunity (Tuominen et al., 2003) and graft versus host disease (GvHD)(Royer et al., 2019). During HSV infection, sensory nerves completely retract from infected mouse corneas by 10 days post infection (dpi) (Chucair-Elliott et al., 2015; Yun et al., 2014). Recently, CD4⁺ T cell-dependent complement activation was reported to participate in the initial loss of sensory nerves in a model of corneal HSV-1 infection and ocular allergy (Royer et al., 2019). Rather than the cornea remaining de-innervated, sympathetic nerves eventually invade the numb cornea around 10 dpi and branch extensively to hyperinnervate the HSK cornea by 28 dpi (Yun et al., 2016). Superior cervical ganglion (SCG)-derived sympathetic hyperinnervation of the cornea represents an important mechanism of HSK pathogenesis. In fact, SCG excision prevents sympathetic hyperinnervation of infected corneas, promotes reinnervation with sensory nerves, and reduces the severity of HSK. The above findings demonstrate that change in the neuroimmune axis of the cornea is an important part of the underlying mechanism of the pathogenesis of HSK(Muller et al., 2003; Yun et al., 2016).

In addition to sensing stimuli, sensory afferents can be directly activated through multiple stimuli that are either host- or microbe- derived. In cutaneous tissue, direct sensing of microbes or cytokines like interleukin (IL)-1β, IL-17, or IL-4 can result in pain or itch (Chiu et al., 2016; Ferreira et al., 1988; McNamee et al., 2011; Oetjen et al., 2017; Pinto et al., 2010). Furthermore, activation of TRPV1⁺ neurons at specific sites in the skin can initiate a cascade of events resulting in the development of “anticipatory” immunity, which augments the host’s ability to resist fungal and bacterial infections (Cohen et al., 2019). In enteric tissue, microbe-induced inflammation and the activation of the NLRP6 inflammasome can lead to a loss of sensory enteric-associated neurons (EANs), which can impair proper intestinal function for weeks after infection (Matheis et al., 2020). Moreover, loss of EANs is coincidentally followed by an expansion of sympathetic nerves within the intestine, suggesting a dynamic relationship between the sensory and sympathetic nervous systems within the intestine. In lungs, sensory afferents, through production of calcitonin gene-related peptide (CGRP), suppress protective immunity during Staphylococcus aureus pneumonia (Baral et al., 2018a, b).

In this study, we sought to identify factors associated with the disrupted balance between sensory and sympathetic corneal nerves that contributes to corneal pathology, focusing on the context of HSK. Our findings point to a critical role for vascular endothelial growth factor-A (VEGF-A) in preventing the re-establishment of sensory nerves in the cornea after HSK induced de-innervation and in promoting the growth/innervation of non-sensing sympathetic nerves, with therapeutic implications.

Results

VEGF-A is produced by infiltrating leukocytes in HSV-1 infected corneas and levels correlate with HSK severity.

Murine corneas infected with HSV-1 were assessed for HSK severity (based on corneal opacity) at various dpi under conditions in which corneas were protected from desiccation by performing tarsorrhaphy at day 4 (closed eye) or not (open eye). Consistent with previously reported results (Yun et al., 2014), HSK severity measured at 7, 10, 14, and 28 dpi was reduced in corneas protected by tarsorrhaphy (Figure 1A). VEGF-A possesses both angiogenic and neurotrophic activity (Carmeliet and Tessier-Lavigne, 2005; Storkebaum et al., 2005; Tovar et al., 2007) and is produced in corneas with HSK, where it contributes to HSK severity by inducing vascularization of the normally avascular cornea (Ambati et al., 2006; Suryawanshi et al., 2011a; Suryawanshi et al., 2012; Zheng et al., 2002). We therefore measured VEGF-A in infected corneas with mild HSK due to tarsorrhaphy and in those with severe HSK that were not protected by tarsorrhaphy. There was a direct correlation between the levels of VEGF-A in HSV infected corneas and the degree of HSK severity at all times (Figure 1B). These findings were consistent with the notion that desiccation stress increased VEGF-A production in HSV-1 infected corneas, and that VEGF-A contributed to or regulated HSK severity.

Figure 1. VEGF is produced by infiltrating leukocytes in HSV-1 infected corneas and VEGF levels correlate with HSK severity.

Mice were infected with 1 × 10⁵ PFU of HSV-1 (strain RE) and either received tarsorrhaphy at 4 dpi or were left untreated. (A) Mice were examined at 7, 10, 14 and 28 dpi for the severity of HSK and the corneal opacity was monitored in a blinded manner and recorded. (B) At each time point, 5 mice from each group were sacrificed, and corneas were excised and dispersed into single-cell suspensions and cultured in Neural basal A medium (with NGF and GDNF) for 24 hours. Cultures were then harvested and analyzed by sandwich ELISA for VEGF. (C) At each time point, corneas were digested, dispersed into single-cell suspensions and were stained with fluorescently conjugated antibodies to detect corneal leukocytes (CD45⁺) by flow cytometry. Immune cell populations that produced VEGF-A were quantified as CD11b⁺F4/80⁺Ly6G⁻ myeloid cells, CD11b⁺F4/80⁻Ly6G⁺ myeloid cells CD11b⁺F4/80⁺Ly6G⁺ myeloid cells, and CD4⁺ T cells (CD4⁺). Flow plots are from 14 dpi. (D) At 14 dpi, HSK corneas were excised, flat mounted, and labeled with antibodies against CD4, F4/80, VEGF, and anti βIII Tubulin antibody. (Top row—high magnification) Arrows point to CD4⁺ cells in close proximity with VEGF producing F4/80⁺ cells. (Bottom row—low magnification) Arrows point to CD4⁺ cells and F4/80⁺ myeloid cells that were positive for anti-VEGF antibody staining. Data are pooled from 2–3 independent experiments and symbols represent individual mice. Bars represent the mean ± SEM. Significance was calculated using Student’s t test, *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001.

VEGF-A is produced by CD4⁺ T cells and myeloid cells in HSV-1 infected corneas

We next determined the source of VEGF-A in the corneas of infected mice. Using immunohistochemistry (IHC) and flow cytometric analysis of infected corneas, we established that CD4⁺ cells and CD11b⁺ monocytic cells produced VEGF-A (Figure 1 C&D). CD4⁺ T cells were in close proximity to F4/80⁺ cells throughout the cornea (Figure1D & Figure S1A). Knowing that dendritic cells and thymus-derived cells express CD4, we determined if VEGF⁺CD4⁺ cells were thymus-derived using FACS to isolate CD4⁺TCRβ⁺ cells from 14 dpi corneas and performed RT-PCR to assess their levels of VEGF mRNA. Notably, greater than 95% of CD4⁺ cornea cells co-stained with TCRβ. Furthermore, IHC illustrated the co-expression of CD4 and CD3 on VEGF⁺ cells in the 14 dpi cornea (Figure S1 B&C). Together, these data supported the conclusion that in addition to myeloid cells, CD4⁺ T produced VEGF during HSV-1 pathogenesis.

Among the CD11b⁺ cells, VEGF-A production was restricted to F4/80⁺Ly6G⁻ myeloid cells, with lesser contributions from F4/80⁺Ly6G⁺ myeloid cells and F4/80⁻Ly6G⁺ myeloid cells (Figure 1 C&D). By 14 dpi, nearing the peak of HSK severity, VEGF-A-producing myeloid cells were seen in close proximity to nerve fibers (FigureS1D), and in many cases, wrapped cellular projections around the nerve endings (Figure S1E). These findings suggested a possible role for VEGF-A⁺ myeloid cells in guiding sympathetic nerves into corneas that are rendered devoid of sensory nerve by 7 dpi (Chucair-Elliott et al., 2015; Yun et al., 2014).

Protecting the corneal surface by tarsorrhaphy reduces VEGF-A production by myeloid cells and inhibits sympathetic innervation of infected corneas.

Since protecting corneas from desiccation stress by tarsorrhaphy reduced overall VEGF-A production, it was of interest to determine if tarsorrhaphy uniformly reduced VEGF-A production by infiltrating leukocytes or selectively inhibited production by certain specific populations. In comparing eyes undergoing tarsorrhaphy to open eyes, we found fewer numbers of VEGF-A producing myeloid cells (Figure 2 A–C). Conversely, CD4⁺ T cells remained unchanged between the groups (Figure 2D). In CD4⁺ T cells, we found equivalent levels of VEGF mRNA (Figure S1B) and frequency of VEGF-A⁺ CD4⁺ T cells (Figure 2E) between the eyes open and tarsorrhaphy groups leading to the conclusion that VEGF-A is likely regulated at the level of transcription in this population. Together, these data suggested that desiccation stress is at least partially responsible for elevated VEGF-A due to the increased recruitment of VEGF-A producing myeloid cells in HSV-infected corneas. Moreover, the tarsorrhaphy-induced reduction of VEGF-A production by these infiltrating leukocytes was most apparent at 14 dpi, when sympathetic nerves were starting to invade the cornea. Indeed, corneas that were protected from desiccation stress by tarsorrhaphy showed a reduction in corneal innervation by sympathetic nerves, which was associated with reduced HSK severity (Figure 2F). Although sympathetic innervation was significantly reduced, sensory innervation was not re-established in the corneas that were protected by tarsorrhaphy.

Figure 2. Protecting the corneal surface by tarsorrhaphy reduces VEGF-A production by leukocytes and inhibits sympathetic innervation of infected corneas.

Mice were infected with 1 × 10⁵ PFU of HSV-1 (strain RE) and either received tarsorrhaphy at 4 dpi or were left untreated. Five mice from each group were sacrificed at 7, 14 and 28 dpi, and the corneas were excised and dispersed into single-cell suspensions that were stained with fluorescently conjugated antibodies to detect corneal immune cells. Numbers represent (A) CD11b⁺ cells, (B) VEGF-A⁺ CD11b⁺ Cells, (C) VEGF-A⁺CD11b⁺F4/80⁺Ly6G⁻ myeloid cells,VEGF-A⁺CD11b⁺F4/80⁺Ly6G⁺ myeloid cells, VEGF-A⁺CD11b⁺F4/80⁻ Ly6G⁺ myeloid cells, (D) CD4⁺ T cells, and (E) VEGF-A⁺ CD4⁺ T cells. (F) 28 dpi corneas with (Closed eye) or without (Open eye) tarsorrhaphy were excised, flat mounted and labeled with anti βIII Tubulin antibody which labels all the nerves, anti-Tyrosine Hydroxylase (TH) antibody which recognizes sympathetic nerves, anti-substance P (SP) antibody which recognizes sensory nerves, the arrows point to βIII ⁺TH⁺ sympathetic nerves in the corneas. Results in all panels are represented as mean number of cells ± SEM from 2–3 independent experiments. Symbols represent individual mice. Significance was calculated using Student’s t test, *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001.

Depletion of CD4⁺ T cells combined with tarsorrhaphy reduces local VEGF-A production and promotes reinnervation of HSK corneas with sensory nerves.

HSV-1 infected corneas that were protected by tarsorrhaphy still exhibited more VEGF-A production, sympathetic innervation, and inflammation than non-infected corneas. Moreover, infiltrating CD4⁺ T cell numbers (Figure 2 D&E) and the production of VEGF mRNA (Figure S1B) and protein (Figure 1B) were equivalent between the groups. In fact by 28 dpi, a higher frequency of CD4⁺ T cells were positive for VEGF by flow cytometry in the tarsorrhaphy group (Figure 2E). Therefore, we assessed whether depleting CD4⁺ T cells from infected corneas would further reduce VEGF-A production by infiltrating leukocytes in eyes that were protected by tarsorrhaphy. Our depletion protocol effectively depleted CD4⁺ T cells, which we showed were VEGF-A producers, from infected corneas (Figure 3A&B). CD4-depletion also significantly reduced the corneal infiltration of F4/80⁺Ly6G⁻ VEGF-A-producing myeloid cells (Figure 3C&D), but did not change the numbers of F4/80⁺Ly6G⁺ or F4/80⁻Ly6G⁺ VEGF-A-producing myeloid cells, possibly due to the low frequency in mock depleted corneas (Figure 3D). Indeed, CD4⁺ T cell depletion reduced overall VEGF-A levels in infected corneas to those observed in non-infected corneas (Figure 3E). Moreover, reducing VEGF-A levels in CD4⁺ T cell-depleted corneas almost completely eliminated ingrowth of sympathetic nerves, and promoted re-innervation by sensory nerves (Figure 3F&G). These changes were associated with virtual elimination of HSK in infected corneas (Figure 3H). It should be noted that CD4⁺ T cell depletion of eyes that did not undergo tarsorrhaphy had no reduction in VEGF-A nor in monocyte infiltration of the cornea, which illustrated the effect desiccation stress has on the corneal microenvironment (Figure S2). Therefore, in an HSK cornea without obvious desiccation stress, CD4⁺ T cells not only produced VEGF-A but also controlled VEGF-A production by myeloid cells. Moreover, when VEGF-A production by these two cell types was abrogated, there was an associated elimination of sympathetic innervation, reinnervation by sensory nerves, and reduction in HSK.

Figure 3. Depletion of CD4⁺ T cells combined with tarsorrhaphy reduces local VEGF-A production and promotes reinnervation of HSK corneas with sensory nerves.

Starting two days prior to infection, mice were given a subconjunctival injection of either anti-CD4 antibody or an isotype control antibody every other day until 28 dpi. Mice were then infected with HSV-1 (RE) and eyes from both groups were protected from desiccation by tarsorrhaphy at 4 dpi. Mice were sacrificed at 7, 14, and 28 dpi and corneas from each group were excised, digested with collagenase, dispersed and analyzed by flow cytometry. Leukocyte (CD45⁺) subpopulations were quantified as (A) CD4⁺ T cells, (B) VEGF-A⁺CD4⁺ T cells, and (C) CD11b⁺ cells. (D) CD11b⁺ cells which were further characterized as VEGF-A⁺F4/80⁺Ly6G⁻ myeloid cells, VEGF-A⁺F4/80⁺Ly6G⁺ myeloid cells, and VEGF-A⁺F4/80⁻Ly6G⁺ myeloid cells. (E) In parallel, corneas from each group were excised and dispersed into single-cell suspensions and were then cultured in Neurobasal A medium (with NGF, GDNF) for 24 hours. Cells and supernatants were harvested and analyzed by sandwich ELISA for VEGF-A. For the apparent outlier, we performed a statistical outlier test offered by GraphPad; however, this point was not determined to be a statistical outlier. Omitting this point still achieves a p value of < 0.0001 for the cell number analysis and < 0.05 for the ELISA data analysis. (F) At 28 dpi, corneas from both groups were excised, flat mounted and labeled with anti βIII Tubulin antibody, anti-Tyrosine Hydroxylase (TH) antibody, and anti-substance P (SP) antibody. Arrows point to βIII ⁺TH⁺ sympathetic nerves or βIII ⁺SP⁺ sensory nerves. (G) Sympathetic nerve length in the corneas with mock or CD4 depletion at 28 dpi were quantified as the cumulative length of nerve fibers in the cornea. (H) The severity of HSK was monitored through 28dpi and the opacity scores (28dpi) were analyzed. Results in all panels are represented as mean ± SEM pooled from 2 independent experiments. Symbols represent individual mice. Significance was calculated using Student’s t test, *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001.

VEGF-A production by cells from corneas with HSK encouraged growth of sympathetic neurites while causing retraction of sensory neurites.

Knowing that VEGF-A levels correlated with disease and the degree of sympathetic innervation within the cornea, we asked if the VEGF-A produced in HSK corneas was able to induce sensory nerve retraction and promote the growth of sympathetic nerves. For this, we cultured sympathetic neurons from dissociated SCG or sensory neurons from dissociated TG in microfluidic chambers for 7 days. This allowed neurons to extend neurites from the neuron soma chamber up to the point of entering the second chamber (Figure 4A). Dissociated cells from infected 14 dpi corneas were then added to the second chamber in the presence or absence of either the clinically approved and widely used human anti-VEGF therapy, bevacizumab (Avastin), or anti-mouse VEGF (R&D Systems). Both bevacizumab and anti-mouse VEGF are antibodies that inhibit VEGF-A signaling in mice. We noted stark differences in axonal growth into the terminal chamber of different treatment groups, allowing us to assign a “yes” (growth) or “no” (no growth) to each well. Using this system we assessed the ability of VEGF-A from a single cornea to modulate the activities of sensory or sympathetic nerves. We then collated those responses to generate a frequency of retraction or extension for either sensory or sympathetic nerves. Corneal cell cultures in which VEGF-A was not neutralized caused retraction of sensory nerve termini in 8 of 9 of cultures (88.9%), while promoting extension of sympathetic neurites in 100% of cultures (Figure 4B–D). Conversely, neutralizing VEGF-A in corneal cell cultures with bevacizumab or anti-mouse VEGF permitted sensory neurite extension in 100% of cultures, while causing sympathetic neurite retraction in 85.7% and 83.3% of cultures, respectively (Figs. 4B–D). Addition of only anti-VEGF antibodies did not affect sensory or sympathetic neurites, suggesting that neither bevacizumab nor anti-mouse VEGF antibody was toxic to these neurons (Figure S3). In these experiments corneal cells were dissociated from 14 dpi corneas, a time when sensory nerves were retracted and sympathetic nerves were actively extending neurites. Since replicating virus is eliminated from the cornea by 6 dpi (Martin et al., 1991) and because the neutralizing antibodies used in vitro were specific for VEGF protein, we concluded from these data that VEGF produced in infected corneas throughout disease was necessary to both maintain sensory nerve retraction and promote sympathetic neurite extension.

Figure 4. VEGF-A production by cells from corneas with HSK promotes growth of sympathetic neurites and retraction of sensory neurites.

Sensory (TG-derived) or sympathetic (SCG-derived) neurons were cultured in a microfluidic chamber (soma chamber) and allowed to extend neurites to grow through the microtubes to the axon chamber. (A) Images represent neuronal growth prior to addition of stimuli. (B-D) Single-cell suspensions of corneas from 14 dpi were added into the axon chamber and incubated for three days in the presence or absence of anti-VEGF antibody or bevacizumab. Each chamber was used to measure the ability of a single cornea to promote or inhibit the growth of sensory or sympathetic nerves. Images in section (B) represents sympathetic nerve growth three days after corneal cells addition. Images in section (C) represent sensory nerve growth three days after corneal cells addition. (D) Data represent the pooled imaging results from two independent experiments. Retraction was determined by the fragmentation of the neurites (arrows) in the axonal chamber. Non-retractions was determined by the elongation of the neurites (arrows) in the axonal chamber. Due to the clear growth or absence of growth of axons in the microfluidic chambers, a “yes” (axonal growth) or “no” (absence of axonal growth) was given for each chamber (cornea). Each experiment consisted of assessing four or five separate corneas resulting in an n value of four or five. To generate the pie charts, we pooled yes/no results and calculated the defined frequency in the pie charts. Significance was calculated using Fisher’s exact test of the yes/no results, and n values were biological replicates of wells that had or had not retracted.

Neutralizing VEGF in vivo during HSK inhibits sympathetic innervation, promotes reinnervation by sensory nerves, and reduces HSK severity.

Our findings demonstrated that the change involving retraction of corneal sensory nerves and hyperinnervation of sympathetic nerves was a major contributor to HSK severity. We further showed in vitro that this neurologic change was regulated by VEGF-A. Therefore, we proposed that neutralizing VEGF-A in corneas with HSK would inhibit the change in the neurologic architecture and reduce HSK severity. To explore these potential therapeutic effects, we treated HSV-1 infected corneas with three different VEGF-A inhibitors.

Bevacizumab is a widely used and clinically approved drug for treating pathogenic vascularization in many diseases, including ocular diseases like age-related macular degeneration and uveitis (Group et al., 2011; Julian et al., 2011). While the assumption would be that bevacizumab would largely act on corneal blood vessels, here we assessed the effects of bevacizumab on corneal nerves and sensitivity. Our treatment strategy began at 9 dpi, after blood vessels had already infiltrated the central cornea (Suryawanshi et al., 2012), and continued every other day until 28 dpi. Unlike beginning treatment at the onset of infection, our treatment began after blood vessels were established, and there was only a marginal effect of bevacizumab on vessel ingrowth after treatment in a fraction of treated mice (Figure 5A & Figure S4). It is worth noting that the blood vessels in treated eyes did appear to be less patent. Despite a lack of effect on vascularization, bevacizumab significantly alleviated corneal opacity and HSK scores in mice that were exposed to desiccation stress and in mice that received tarsorrhaphy (Figure 5B). Next, we showed that bevacizumab prevented most sympathetic innervation of the cornea after HSV-1 infection (Figure 5C–E). Finally, we demonstrated a reinnervation by corneal sensory nerves, which correlated with a recovery of cornea blink reflex (Figure 5 C&F). To ensure that these effects were not limited to this specific VEGF-A inhibitor, we repeated the same experiments using anti-mouse VEGF and soluble VEGF-A receptor 1 (sVR1). Both treatments alleviated HSK in the same manner that was observed with bevacizumab treatment (Figure S5 & Figure 6). Together, these results supported our in vitro results that implicated VEGF as a critical factor in the establishment of pathogenic sympathetic nerves and prevention of sensory nerve regeneration in HSK corneas.

Figure 5. Neutralizing VEGF in vivo during HSK inhibits sympathetic innervation, promotes reinnervation by sensory nerves, and reduces HSK severity.

Four groups of mice were infected with HSV-1 and were given subconjunctival injections of 500μg bevacizumab or isotype every other day starting from 9 dpi through 28dpi. Two groups of mice (bevacizumab and isotype) were protected by tarsorrhaphy. The severity of HSK was monitored and (A) corneal neovascularization and (B) corneal opacity at 28 dpi were recorded and analyzed. (C) At 28dpi, mice were sacrificed, and corneas were excised, flat mounted and labeled with antibodies to βIII Tubulin, Tyrosine Hydroxylase (TH), and substance P (SP). Arrows point to βIII ⁺TH⁺ sympathetic nerves or βIII ⁺SP⁺ sensory nerves. (D & E) Sympathetic nerves in the corneas protected by tarsorrhaphy were traced using the Simple Neurite Tracer (Longair et al., 2011) in the segmentation package in FIJI and was followed by analysis using the 3D Skeletonize (Arganda-Carreras et al., 2010) FIJI plugin. The z projection of the traced neurons from one cornea is shown. (F) The mice without tarsorrhaphy were assessed for corneal reflex at 28dpi. Results in all panels are represented as mean ± SEM pooled from 2 to 3 independent experiments. Significance was calculated using Student’s t test, *p < 0.05, **p < 0.01, ****p < 0.0001.

Figure 6. Soluble VEGF receptor 1 treatment prevents sympathetic nerve innervation and regenerates corneal sensitivity.

Mice were infected with HSV-1 (RE) and beginning at 9 dpi were given subconjunctival injections of soluble VEGF receptor 1 or isotype antibody every other day through 23dpi. (A-D) HSK severity was monitored as corneal opacity, neovascularization, and corneal reflex. Arrows point to the blood vessels in the corneas. (E) The corneas were excised, flat mounted and labeled with antibodies to βIII Tubulin, Tyrosine Hydroxylase (TH), and substance P (SP). Arrows point to βIII ⁺TH⁺ sympathetic nerves or βIII ⁺SP⁺ sensory nerves. Images in all panels represented the nerve staining from 2 independent experiments. Results in all panels are represented as mean ± SEM pooled from 2 independent experiments. Significance was calculated using Student’s t test,**p < 0.01.

These data unveiled a role for VEGF-A in the neuro-immune interactions that mediated blinding pathology in HSV-1 infected corneas. We demonstrated that VEGF-A both promoted hyperinnervation of the cornea by sympathetic nerves and prevented reinnervation by sensory nerves. These changes in the neurologic architecture of the cornea were necessary components of pathology in corneas infected with HSV-1 and likely other infectious and non-infectious corneal diseases that are associated with a loss of corneal sensation. By showing that interrupting VEGF-A signaling during disease alleviated corneal opacity and restored sensitivity, we have identified a potential strategy for treating this blinding ocular surface disease.

Discussion

Given the significant contributions of sensory nerve retraction and sympathetic nerve ingrowth to HSK severity, we sought to identify factor(s) that mediate these neurologic changes in infected corneas. Our initial observation that levels of VEGF-A (a known neurotrophic factor) correlated with HSK severity suggested VEGF-A might be of interest. Therefore, we investigated the cellular sources of VEGF-A. Since HSK is associated with a heavy leukocytic infiltration, it was of particular interest to define leukocytic sources of VEGF-A. We found that VEGF-A was produced by a variety of leukocytes in corneas with HSK, including F4/80⁻Ly6G⁺, F4/80⁺Ly6G⁺, and F4/80⁺Ly6G⁻ myeloid cells as well as CD4⁺, TCRβ⁺, CD3⁺ T cells, with the latter two cell types being the most abundant VEGF-A producers. Notably, VEGF-A from these cells directly acted upon corneal nerves to retract sensory afferents while simultaneously encouraging the ingrowth of sympathetic nerves resulting in blinding ocular pathology.

Normal corneas are heavily innervated by sensory nerves with few if any sympathetic nerve fibers (Muller et al., 2003). In mice and humans, microbial infections are associated with loss of corneal sensory nerves (Bonini et al., 2003; Cruzat et al., 2010; Cruzat et al., 2011; Hamrah et al., 2010; Kurbanyan et al., 2012; Muller et al., 2015), and we have demonstrated that the resulting loss of corneal sensation and blink reflex contributes to HSK severity in mice. Humans, unlike mice, exhibit a phenomenon called consensual blink reflex in which both eyelids blink when one cornea is stimulated (Alexander G. Reeves; Kaplan and Kaplan, 1980; Yun et al., 2014). Therefore, in humans, even if the infected cornea is rendered numb by loss of sensory nerves, the eyelids of the infected eye can still blink when the contralateral cornea is stimulated, thus providing a level of protection from desiccation stress to the infected cornea. For this reason, we believe that reducing desiccation stress in HSV-1 infected mouse corneas by performing tarsorrhaphy created a better model of human HSK.

It is unclear if retraction of sensory nerve fibers during HSV-1 infection is advantageous to infected corneas. However, it is possible to posit that this mechanism might exist to reduce access of neurons harboring latent virus to the cornea, thus reducing viral shedding to the cornea following reactivation stimuli. In the C57BL/6 model used in this study, latent HSV-1 is harbored in TG but does not reliably reactivate despite being exposed to environmental stressors that have been linked to reactivation in humans (Huang et al., 2011). Thus, it is not possible to assess an effect of sensory nerve retraction on viral shedding at the cornea following a reactivation event. However, ongoing studies are aimed at refining our mouse model of viral reactivation from latency to address this important issue.

Previously, we and other have shown that HSK and desiccation stress begin to develop concurently around 7 dpi, which is after the clearance of live virus from the cornea (Jeon et al., 2018). Thus, by performing tarsorrhaphy to prevent corneal desiccation at 7 dpi we avoided any possible contribution of corneal live virus to HSK development. Protecting the infected cornea from desiccation stress reduced, but did not eliminate VEGF-A production, sympathetic innervation, and HSK, and these corneas did not exhibit re-innervation with sensory nerves. The sources of the residual VEGF-A in these corneas remained CD4⁺ T cells and F4/80⁺Ly6G⁻ myeloid cells. Depleting CD4⁺ T cells systemically not only eliminated VEGF-A-producing CD4⁺ T cells from infected corneas protected by tarsorrhaphy, but also eliminated VEGF-A production by F4/80⁺Ly6G⁻ myeloid cells, reducing overall VEGF-A production to levels observed in non-infected corneas. The reduced VEGF-A levels in infected corneas of CD4⁺ T cell depleted mice were associated with prevention of sympathetic hyperinnervation, recovery of lost sensory nerves, and virtual elimination of HSK.

It remains unclear exactly how CD4⁺ T cells are stimulated to produce VEGF-A within the context of infection. Recently, the activation of complement was linked to CD4⁺ T cell-dependent loss of corneal sensitivity in models of HSV-1 keratitis and allergic eye disease (Royer et al., 2019) suggesting a potential intrinsic (Arbore et al., 2016; Cardone et al., 2010) or extrinsic role of complement in CD4⁺ T cell-dependent VEGF-A production. Additional evidence that complement might regulate VEGF-A production by CD4⁺ T cells comes from the observation that complement factor H (CFH) can interfere with CD47 binding with thrombospondin-1 (TSP-1), and that CD47 deficiency in T cells results in elevated production of VEGF (Calippe et al., 2017).

VEGF-A is a known neurotrophic factor (Jin et al., 2006; Nishijima et al., 2007; Sondell et al., 2000), but we wished to provide a more direct demonstration of a potential role in selectively regulating extension and contraction of sympathetic and sensory nerves respectively in the infected cornea. Corneal sensory nerves derive from cell bodies in the TG, whereas sympathetic nerve fibers found in the peripheral area of normal corneas derive from neuronal cell bodies in the SCG. We cultured sensory neurons derived from dissociated TG or sympathetic neurons derived from dissociated SCG in microfluidic chambers comprised of two cell chambers separated by micro channels until neurites started to extend through the channels into the opposite chamber. Dissociated cells from infected corneas obtained at 14 dpi—more than a week after infectious virus is cleared from the cornea—were then added to the neurite chambers with or without VEGF-specific antibodies. Results showed that corneal cells caused retraction of sensory neurites and promoted the extension of sympathetic neurites into the corneal cell chamber. However, when VEGF was neutralized in corneal cultures by specific antibodies, the opposite effect was observed: sensory neurites extended into corneal chambers, while sympathetic neurites retracted. These findings suggested that VEGF created an environment within HSV-1 infected corneas that favors sympathetic hyperinnervation while discouraging reinnervation by sensory nerves.

Previous studies established that VEGF-A inhibition reduces HSK severity in association with reducing neovascularization of the infected cornea. Neovascularization of the normally avascular cornea places the cornea in direct contact with inflammatory mediators carried in the blood stream, and inhibiting that exposure would be expected to reduce inflammation within the infected cornea. It is possible that VEGF inhibitors could disrupt the patterns of HSV-1 entry into and exit from viral latency; however, the stability of viral latency in C57BL/6 mice and timing the administration of treatments after cessation of active viral replication would suggest that these inhibitors do not affect viral latency. Thus, VEGF-A has a continuing role in HSK beyond vascularization by promoting hyperinnervation by sympathetic nerves that contributes to HSK severity, and repressing sensory innervation that is required for normal corneal homeostasis (Labetoulle et al., 2019; Okada et al., 2019). While it is known from previous studies that VEGF-A can stimulate cellular processes in both sympathetic and sensory nerves, here, we illustrated that during the peak inflammatory response of HSK, sympathetic nerves are preferentially stimulated and sensory nerves are repelled or their growth inhibited. Whether the differential effect on sympathetic and sensory neurons reflects such variables as VEGF-A concentration or a supplemental contribution of other neurotrophic factors present in infected corneas remains to be clarified.

The cornea is uniquely innervated in that sensory nerves are the most prominent nerves found within the healthy tissue, while sympathetic and parasympathetic nerves are largely or completely absent. Here, our data revealed that VEGF-A was largely responsible for maintaining the disruption of the normal nerve architecture after viral infection where corneal sensory nerves were replaced by sympathetic nerves. We observed this not only through a quantifiable reduction in blink reflex but also a significant shift in the expression of SP in nerves to TH in nerves going from the healthy to diseased state, respectively. This shift in protein expression was not a simple change in phenotype of existing nerves, but a replacement of sensory nerves with sympathetic nerves. Data to support this comes from the observation that excising the superior cervical ganglion (SCG)—considered a sympathetic ganglion—does not affect corneal nerve architecture during steady state but does eliminate sympathetic innervation during the diseased state (Yun et al., 2016). The mechanisms governing why parasympathetic nerves are absent in the cornea during health and disease remain unknown. Together, these results may aid interpretation of disease in other tissues like the skin where sensory nerves and sympathetic nerves are intermingled.

A recent study investigated the therapeutic benefit of adding the isoform of VEGF, VEGF188, topically to the ocular surface to enhance corneal nerve regeneration and improve wound healing (Brash et al., 2019). While that study failed to distinguish sympathetic from sensory nerves, the results suggest that further resolution of the effects of different isoforms of VEGF might be warranted. Our study revealed a profound neuro-immune contribution to HSK and suggested that VEGF inhibitors, which are currently clinically approved to prevent pathogenic vascularization might be useful in preventing changes in the neural architecture that contribute to HSK severity and, potentially, other neuronal disorders. In VEGF-A, we have identified a potentially promising target where a neutralizing antibody, bevacizumab, is FDA-approved and has been used in humans for other diseases.

Limitations of Study

Despite the numerous similarities between mice and humans, our study uses a C57BL/6 mouse model of HSV-1 infection, which imperfectly models human disease. In this model, disease is severe and occurs relatively soon after primary infection; however, human disease is progressive and takes years to develop. This is likely due to the dramatic role desiccation stress plays in murine disease, which may not be recapitulated in humans due to consensual blink reflex mentioned above. Further, studies from our lab and others have demonstrated the lack of consistency in measuarable reactivation of HSV-1 in mouse models of disease. In collaboration with the Patrick Stuart Laboratory (St. Louis University), we are currently investigating the effects of measurable viral reactivation on the corneal nerve landscape using a more efficient mouse model of HSV-1 reactivation. An unexpected observation in our study was the maintenance of VEGF protein expression during tarsorrhaphy despite the dramatic loss of VEGF from the myeloid population. This would suggest that the stimulant that triggers CD4⁺ T cells to produce VEGF is not associated with desiccation stress. Because the specificities of pathogenic CD4⁺ T cells that infiltrate the cornea remain unknown, it has been difficult identify the mechanisms governing the induction of VEGF production within the cornea. In our study, we highlight the close proximity of CD4⁺ T cells, myeloid cells, and sympathetic nerves and suggest that these three cell types actively interact to promote VEGF production and pathology. Future studies are aimed at identifying how CD4⁺ T cells are stimulated and how VEGF production may be regulated within these cells and the myeloid cells that infiltrate the cornea during infection. Finally, it is likely that VEGF-A is not the only neurotrophic factor that regulates the neuronal landscape within the cornea, and ongoing studies are aimed at identifying other factors that may play a role.

STAR METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Anthony St. Leger (Anthony.stleger@pitt.edu).

Materials Availability

This study did not generate new unique reagents.

Data and Cod Availability

This study did not generate/analyze any datasets or code.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animal Studies

Wild-type female C57BL/6 mice purchased from Jackson Laboratories (Bar Harbor, ME, USA) were housed in the Animal Resource Facility at the University of Pittsburgh Medical Center (Pittsburgh, PA, USA) and used at 6–8 weeks of age in all experiments. Experimental procedures were reviewed and approved by the University of Pittsburgh Institutional Animal Care and Use Committee and the use of animals was in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Primary neurons

Sensory neurons of Trigeminal ganglia (TG) and sympathetic neurons of superior cervical ganglions (SCG) were isolated frm wild-type C57BL/6 mice.

HSV-1 strain RE

The RE strain of HSV-1 was grown to high titer in Vero cells after a low multiplicity of infection (0.001) and intact virions obtained from the infected cell surface were purified on OptiPrep density gradients (Accurate Chemical and Scientific Corp., Westbury, NY, USA) and stored at −80 °C, as previously described (Treat et al., 2020). Concentration of HSV-1 was determined in a standard virus plaque assay.

METHOD DETAILS

Mouse infections

Mice were anesthetized by intraperitoneal injection of ketamine hydrochloride (100 mg/kg body weight) and xylazine(0.1 mg/ kg body weight) in Hanks’ balanced salt solution. Topical corneal infection was performed by scarification of the central cornea with a sterile 30-gauge needle in a crisscross pattern and applying 3 μl RPMI containing 1×10⁵ pfu of HSV-1.

Scoring HSK severity

HSK severity was monitored using a Zeiss microscope (Stemin 305) and scored as described previously (Yun et al., 2014). Briefly, the corneal opacity was scored on a four-point scale protocol as follows: 0.5, any corneal imperfection; 1, mild corneal haze; 2, moderate opacity; 2.5, moderate opacity with regional dense opacity; 3, diffuse dense opacity obscuring the iris; 3.5, diffuse dense opacity with corneal ulcer; or 4, corneal perforation. The corneal neovascularization was scored by dividing each cornea into 4 quadrants and each quadrant was scored as 0 (no vessels visible), 2 (vessels extending into the paracentral cornea), or 4 (vessels extending to the central cornea). The total score of the 4 quadrants was then divided by 16 (360°vascularization to the central cornea) and multiplying by 100 which represents the percentage of each cornea that was vascularized.

Corneal blink reflex was tested by loosely holding the mouse and touching the cornea with the blunt tips of a surgical forceps without touching the eyelashes and whiskers. The cornea was divided into 5 areas (4 quadrants and center area), loss of blink reflex referred to the inability of the mouse to blink when an area was touched and was recorded as 0. Positive blink reflex referred to the ability to blink when an area of the cornea was touched and was recorded as 1. The total score of the 5 areas would be the final score of corneal blink reflex for a mouse. A score of 0 indicated a complete loss of corneal sensation such that the mouse failed to blink when any area of the cornea was touched. A score of 5 indicated retention of some degree of sensation such that the mouse blinked when any area of the cornea was touched.

Tarsorrhaphy

Tarsorrhaphy is one of the safest and most effective surgical procedures to protect a neuropathic cornea from exposure (www.aao.org, American Academy of Ophthalmology) in which the closure of the eyelids is formed via an adhesion between the upper and lower lid margins.

Tarsorrhaphy was performed on mice 4 days post HSV-1 infection under a surgical microscope. First, the upper and low eyelids of a mouse were mildly cut with scissors so that the margins were rough, and then the two margins were stitched together with 9–0 Black Nylon monofilaments through the interstitial part of the eyelids. The sutures were removed 7 days after the surgery. Tarsorrhaphy was removed at the end points of the experiments to permit examination of the corneas.

CD4 depletion

Mice were depleted of CD4⁺ T cells with an intraperitoneal injection of 0.15 mg rat anti-mouse CD4 Ab clone GK1.5 (InVivoMAb anti-mouse CD4, BioXcell, West Lebanon, NH) or mock depleted with an intraperitoneal injection of 0.15mg anti-keyhole limpet hemocyanin (LTF, InVivoMAb rat IgG2b isotype control). Injections were given 2 days before infection, one day after infection and then every 6 days through 28 days post infection (dpi). The efficacy of CD4 depletion was confirmed by a lack of CD4⁺T cells in corneas of depleted mice as assessed by both confocal microscopy and flow cytometry (not shown).

Anti VEGF treatments to the mice

Mice received anti VEGF treatment or mock treatment at 9 dpi. To block VEGF pathway by anti VEGF antibodies, mice were given 500μg bevacizumab (Avastin) or 10μg rat anti mouse VEGF antibody by subconjunctival injection every two days through 28 days. To sequester VEGF by soluble VEGF receptor 1 (sVEGFR1, sVR1), mice were given 5μg (Suryawanshi et al., 2011a) of soluble VEGF receptor 1 by subconjunctival injections every two days through 28 days. The same amount of Recombinant Mouse IgG2A Fc Protein was given as mock treatment.

Primary culture of neurons in microfluidic chambers

Microfluidic chambers were manufactured with two compartments connected by micro channels. The silicon-based master was fabricated in cleanroom conditions. In brief, silicon wafers (University Wafer; Boston, MA) were spin-coated with SU-8 2002 Photoresist (Microchem; Westborough, MA) in a 2-step process and soft baked. An array of microchannels (10 μm width; 450 μm length; 2–2.5 μm height) was defined by UV light exposure and development of SU-8. Standard soft lithography was performed using Sylgard 184 polydimethylsiloxane (PDMS) (Dow Corning). Because of the restrictive height (2.5 μm) in the channels, cells are unable to migrate through the channels, but diffusion of small molecules is not impeded. PDMS microfluidics devices were permanently bonded to glass substrates using a PE-25 plasma cleaner (Plasma Etch). PDMS and glass substrate were exposed to plasma for 2 minutes at a flow rate of 5 cc/min under 200 mTorr vacuum pressure and then the plasma-exposed surfaces immediately brought together. The chambers of the devices were coated with poly-D lysine (PDL) and laminin before the addition of cells. One chamber of the device was designated as the “Soma” side and the other “Axon”.

Sensory neurons of Trigeminal ganglia (TG) and sympathetic neurons of superior cervical ganglions (SCG) were cultured according to a protocol obtained from Dr. Todd Margolis, University of Washington at St Louis MO (Bertke et al., 2011). Briefly, TGs and SCGs were removed after 6–8 week old female C57BL/6 mice were euthanized and incubated at 37°C for 20 minutes in papain (5mg/ml in Neurobasal A medium), followed by an additional incubation at 37°C for 20 minutes in Hanks balanced salt solution containing dispase (4.67 mg/ml) and collagenase (4 mg/ml), and then were mechanically dissociated by triturating with a 1000-μl pipette. Cell suspension of TG was layered on a 5-step OptiPrep (Sigma-Aldrich) gradient and centrifuged for 20 minutes at 800xg. The lower end of the centrifuged gradient was then transferred to a new tube and washed twice with Neurobasal A medium supplemented with 2% B27 supplement and 1% penicillin streptomycin. The neurons were cultured in complete neuronal medium, consisting of Neurobasal A medium supplemented with 2% B27 supplement, 1% penicillin streptomycin, L-glutamine (500μM), nerve growth factor (NGF, 50ng/ml), glial-cell-derived neurotrophic factor(GDNF, 50ng/ml). Fluorodeoxyuridine (40μM) was added for the first day. The neurons were then maintained in complete neuronal medium without mitotic inhibitor.

Sympathetic neurons were cultured using the same methods without the step of gradient centrifugation.

Suspended HSK corneal cells (14 dpi) were isolated using the same methods without the step of gradient centrifugation.

Each Soma chamber were seeded with sensory neurons isolated from one TG or sympathetic neurons isolated from two SCGs. The same neuronal culture medium was added to the axon chamber to a higher level that creates a hydrostatic pressure. When axonal projections were observed to extend through the micro channels and into the Axon chamber (usually in 7 days), the suspended HSK corneal cells in complete neuronal medium with or without anti VEGF antibodies (rat anti mouse VEGF antibody, 10μg/ml; bevacizumab, 1.25mg/ml) were added into the axon chamber side.

The growth of neurites in the axon chamber was observed and imaged using a Zeiss Axi0 microscope. The images were analyzed using ZEN 2 (blue edition) software (ZEISS, USA).

ELISA

For each experiment, suspended corneal cells were obtained as described in the method of “Microfluidics Devices and neuron culture”, and then were cultured in complete neuronal culture medium for one day. The supernatants and the cells were harvested and analyzed by sandwich ELISA for VEGF (Mouse VEGF DuoSet ELISA, R&D SYSTEMS, DY493) using DuoSet Ancillary Reagent Kit 2 (R&D SYSTEMS, DY008), and the samples were processed according to the Research and Development protocol. Absorbance was read at 450 nm using a SpectraMax M3 plate reader and analyzed using SoftMax Pro 7.0 software (Molecular Devices).

Flow Cytometry

HSK was scored at the end point of each experiment (28 dpi or 21 dpi) and the mice were euthanized, and eyes were collected for flow cytometry. Individual corneas were excised, rinsed and quartered for preparation of single cell suspensions, made by treating the corneas with collagenase type I (84 U/cornea; Sigma-Aldrich, St. Louis, MO) for 60 min at 37°C, and triturated until no apparent tissue fragments remained. The single-cell suspension of each cornea was then filtered through a 40μm cell strainer cap (BD Labware, Bedford, MA) and washed with FACS buffer. Corneal cells were treated with anti-mouse CD16/CD32 (Fc III/II receptor; 2.4G2; BD Pharmingen, San Diego, CA) to prevent nonspecific antibody binding and then stained for various leukocyte surface markers and Ghost UV 450 Viability Dye for 30 min at 4°C. The cells were then washed with FACS buffer and incubated in BD Cytofix/Cytoperm (Fixation and permeabilization solution) for 20 min at 4°C followed by wash with BD Perm/Wash buffer twice. The cells were then stained with goat anti mouse VEGF antibody and followed by a Alexa Fluor 633 conjugated-donkey anti-goat IgG(H+L) antibody staining. After two washes with BD Perm/Wash buffer, the cells weretransferred into FACS buffer for the assessment with a CytoFLEX LX Cytometer (Beckman Coulter) and were analyzed using FlowJo software (TreeStarInc., Ashland, OR). Gates were set based on staining with the appropriate single antibody and a cocktail lacking that particular antibody. Data are listed as total numbers of cells per cornea.

CD4⁺TCRβ⁺ cells and CD11b+ myeloid cells Sorting

Mice were infected with 1 × 10⁵ PFU of HSV-1 (strain RE) and either received tarsorrhaphy at 4 dpi or were left untreated. At 14 dpi,10 mouse eyes from each group were collected, and corneas were excised, digested with collagenase, and dispersed into single-cell suspensions, and then cells from 5 eyes were pooled into one tube for antibodies staining. Corneal cells were treated with anti-mouse CD16/CD32 (1:500, Fc III/II receptor; 2.4G2; BD Pharmingen, San Diego, CA) to prevent nonspecific antibody binding. The final dilution of various leukocyte surface markers was 1:500, and the Aqua viability dye was added at 1:1000 final concentration. FACS was used to isolate CD4⁺TCRβ⁺ cells and CD11b⁺ myeloid cells for the assessment of their levels of VEGF mRNA by performing RT-PCR. Two independent experiments were performed and 500 CD4⁺TCRβ⁺ cells and CD11b⁺ myeloid cells were collected from each sample.

RT-PCR

TaqMam Gene Expression Cells-to-Ct Kit was used to isolate mRNA from the cells according to the manufacturer’s protocol. Briefly, the isolated CD4⁺ cells and CD11b⁺ myeloid cells were washed with 1XPBS, mixed with Lysis Solution, and incubated at room temperature for 5 min, and then Stop Solution was mixed into the lysate to inactivate the lysis reagents. After that, cell lysates were reverse transcribed to synthesize cDNA.

As the quantities of cDNA from each sample were small, TaqMan PreAmp Kit was used for the preamplification of cDNA samples. Equal volume of 20X mouse vegfa and beta-actin TaqMan Gene Expression Assay were combined and diluted using 1X TE Buffer to make each assay for a final concentration of 0.2X. The amplification reaction was performed in 50ul volume according to the protocol for 14 preamplification cycles.

The preamplification products were 1:20 diluted using 1XTE Buffer and prepareed real-time PCR reaction in a 20ul volume system. Three replicates of each PCR reaction were performed. The data was analyzed using StepOne Software v2.3 and absolute CT of vegfa was used to indicate the existence of gene expression. The expression of beta-actin was used as the quality control for each sample.

Immunohistochemistry

Corneas were dissected and fixed at room temperature for 1 hour in 1.3% paraformaldehyde in PBS, and radial incisions were made to facilitate flat-mounting of the corneal tissues. Corneas were washed in PBS five times, permeabilized in 1% Triton-X100 in PBS at room temperature for 60 minutes and blocked with either 20% goat serum or donkey serum (Cedarlane, Burlington, NC, USA) in blocking buffer (0.3% Triton-X-100/0.1% Tween-20 in PBS) for 1 hour. The corneas were then incubated in a 100μl cocktail of primary antibodies at room temperature for 2 hours, followed by an additional incubation overnight at 4 °C. After five 5-minute washes in wash buffer (0.1% Tween-20 in PBS), the corneas were incubated in a 100μl cocktail of secondary antibodies in blocking buffer at room temperature for 2 hours. Following five 10-minute washes with wash buffer, the corneas were mounted on slides and dried at 4°C for at least 12 hours before imaging.

Confocal Microscopy

Stitched Z-stacks spanning entire corneal whole mounts were acquired with an OLYMPUS BX61 motorized upright Fluoview 1200 laser scanning confocal microscope equipped with a x20 or a 60x oil (numerical aperture, 0.85) objective lens and an automated stage. The Z-stack images were saved in the native Olympus Image Binary (OIB) format and stitched together using FV10-ASW 2.0 software (Olympus Life Science, Tokyo, Japan). Brightness levels in the figures were adjusted for display.

The evaluation of sympathetic nerves in each cornea was processed using Simple Neurite Tracer (Longair et al., 2011) in the segmentation package in FIJI programs and then analyzed by the 3D Skeletonize (Arganda-Carreras et al., 2010) FIJI plugin. The total length of sympathetic nerves in each cornea was calculated from the data provided.

QUANTIFICATION AND STATISTICAL ANALYSIS

All values are presented as mean ± SEM. The normality of each set of data were checked by D’Agostino & Pearson normality test with GraphPad Prism 7.03. Outliers were detected by Outlier calculator in GraphPad (https://www.graphpad.com/quickcalcs/grubbs1/). If the data were normally distributed, the significance of differences between groups were determined by unpaired parametric t-test; If the data were not normally distributed, the significance of differences between groups were determined by unpaired nonparametric t-test (Mann-Whitney test). Fisher’s exact test was used to analyze in vitro neuron culture data. Differences were considered to be statistically significant at P < 0.05.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Brilliant Violet 785 anti-muse CD4	BioLegend	Cat#100551; Lot# B259957; RRID AB_11218992
Alexa Fluor 700 anti-mouse CD11b	BioLegend	Cat#101222; Lot#B259438; RRID AB_493705
PerCP Rat Anti-Mouse CD45	BD Pharmingen	Cat#557235; Lot# 8179968; RRID AB_396609
Brilliant Violet 650 anti-mouse Ly-6G	BioLegend	Cat#127641; Lot#B268950; RRID AB_2565881
PE/Cy7 anti-mouse F4/80	BioLegend	Cat#123114; Lot#b265636; RRID AB_893478
Anti-mouse-VEGF	R&D Systems	Cat#AF-493-NA; Lot#YU1417041; RRID AB_354506
Donkey anti-goat IgG (H+L) Alexa Fluor 633	Molecular Probes	Cat#99E1-1
Ghost UV 450 Viability Dye	TONBO Biosciences	Cat#13-0868-T100; Lot#D0868083018133
PE Rat Anti-Mouse CD4	BD Pharmingen	Cat#553049; Lot#7194731; RRID AB_394585
APC Rat-Mouse CD31	BD Pharmingen	Cat#561814; Lot#6168667; RRID AB_10893351
eFluor 450 Anti-Mouse CD45	eBioscience	Cat#48-0451-82; Lot#4295770; RRID AB_1518806
APC Rat-Mouse F4/80	eBioscience	Cat#17-4801-82; Lot#E07285-1631; RRID AB_2784648
Goat anti-chicken IgG(H+L) Alexa Fluor 546	Life Technologies	Cat#A11040, Lot#1618409; RRID AB_1500590
Goat anti-rat IgG(H+L) Alexa Fluor 633	Invitrogen	Cat#1889312; Lot#1889312
Goat anti-rabbit IgG(H+L) Alexa Fluor 488	Abcam	Cat#Ab15007; Lot#GR23372; RRID AB_301568
Rat Anti-Substance P	BD Pharmingen	Cat#556312; Lot#7032604; RRID AB_396357
Anti-Beta III Tubulin	Abcam	Cat#Ab18207; Lot#GR220660-1; RRID AB_444319
Anti-Tyrosine Hydroxylase	Abam	Cat#Ab76442; Lot#GR214411-3; RRID AB_1524535
InVivoMAb anti-mouse CD4	BioXCell	Cat#BE0003-1; Lot#699918O1; RRID AB_1107636
InVivoMAb rat IgG2b isotype Control	BioXCell	Cat#BE0090; Lot#695618J2; RRID AB_1107780
Bevacizumab	Genentech	Cat#NDC50242-060-01
Recombinant Mouse VEGFR1/Flt-1 FC Chimera Protein	R&D Systems	Cat# 471-F1-100; Lot#DCPJ041
Recombinant Mouse IgG2A	R&D Systems	Cat#4460-MG, Lot#PBV0416101; RRID AB_884569
FITC Rat Anti-Mouse Ly6C	BD Pharmingen	Cat#553104; RRID AB_394628
PE Rat Anti-Mouse CD3	BD Pharmingen	Cat#555275; RRID AB_395699
Aqua Viability Kit	Thermo Fisher Scientific	Cat#L34966; Lot#2069643
APC-Cy7 Rat Anti-Mouse CD4	BD Pharmingen	Cat#100526; Lot#B282355; RRID AB_312727
BV421 Hamster anti-mouse TCR β Chain	BD Pharmingen	Cat#562839; Lot# 8311947; RRID AB_2737830
Bacterial and Virus Strains
Herpes Simplex Virus Type 1 (Strain RE)	In-house Virus Stock	PMID: 1401927
Chemicals, Peptides, and Recombinant Proteins
4’,6-Diamidino-2-phenyindole,dilactate	Sigma-Aldrich	Cat#08168-100EA
Recombinant Human β-NGF	Shenandoah Biotechnology	Cat#100-38; Lot#116-100-38
Recombinant Rat GDNF	R&D Systems	Cat#512-GF-010; Lot#UH3116051
Laminin	Sigma-Aldrich	Cat#11495-81-9; Lot#127M4092V
Poly-D-lysine hydrobromide	Sigma-Aldrich	Cat#27964-99-4; Lot#SLBX7897
Papain	Worthington Biochemical Corporation	Cat#38D18191; Lot#38D18191
Neurobasal-A Medium	Thermo Fisher Scientific	Cat#10888022
Optiprep Density Gradient Medium	Sigma-Aldrich	Cat#D1556 SIGMA; Lot#MKCG1091
Collagenase Type I	Stemcell Technologies	Cat#7416
Gibco Dispase II	Fisher Scientific	Cat#17-105-041; Lot#1970805
GlutaMAX	Life Technologies	Cat#35050-061; Lot#2085465
B-27 Supplement	Life Technologies	Cat#17504-044; Lot#2077007
Penicillin Streptomycin	Life Technologies	Cat#15140-122; Lot#1924797
Ketamine Hydrochloride Injection	Henry Schein	Cat#NDC11695-0702-1; Lot#AH02P6J
Xylazine Injection	Henry Scein	Cat#NDC11695-4022-1; Lot#8L022
Antisedan	Orion Pharma	Cat#107204-6; Lot#1776061
Hanks Balanced Salt Solution (HBSS)	Life Technologies	Cat#24020; Lot#2048597
RPMI	BioWhittaker	Cat#12-167Q; Lot#696504
Critical Commercial Assays
BD Cytofix/Cytoperm	BD Pharmingen	Cat#554722; Lot#9064590
BD Perm/Wash	BD Pharmingen	Cat#554723; Lot#8263982
Mouse VEGF DuoSet ELISA	R&D Systems	Cat#DY493; Lot#P210213
DuoSet Ancillary Reagent Kit 2	R&D Systems	Cat#DY008
TaqMan® Gene Expression Cells-to-CT™ Kit	Ambion by Life Technology	Cat#4399002; Lot#00905842
TaqMan® PreAmp Master Mix	Applied Biosystems	Cat#4384266; Lot#00885329
TaqMan Gene Assay-Vegfa	Thermo Fisher	Cat#4331182 Mm00437306_m1
TaqMan Gene Assay-Actb	Thermo Fisher	Cat#4387430
Experimental Models: Organisms/Strains
C57BL/6 mouse	The Jackson Laboratory	JAX: 000664; RRID IMSR_JAX:000664
Software and Algorithms
FV10-ASW 4.1	Olympus Life Sciences	https://www.olympus-lifescience.com/en/support/downloads/ RRID SCR_014215
SoftMax Pro 7.0	Molecular Devices	https://www.moleculardevices.com/ RRID SCR_014240
Flowjo V.10	BD	https://www.flowjo.com/ RRID SCR_008520
Other
Axio Microscope	Zeiss	www.zeiss.com/us/microscopy
Stemin 305 Microscope	Zeiss	www.zeiss.com/us/microscopy
BX61 motorized upright microscope	Olympus	https://www.olympus-ims.com/en/microscope/bx61-2/
9-0 Black Nylon Monofilament	AROSurgical Instruments Corporation	Cat#TK-091038; Lot#PG467301

TABLE WITH EXAMPLES FOR AUTHOR REFERENCE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-Snail	Cell Signaling Technology	Cat#3879S; RRID: AB_2255011
Mouse monoclonal anti-Tubulin (clone DM1A)	Sigma-Aldrich	Cat#T9026; RRID: AB_477593
Rabbit polyclonal anti-BMAL1	This paper	N/A
Bacterial and Virus Strains
pAAV-hSyn-DIO-hM3D(Gq)-mCherry	Krashes et al., 2011	Addgene AAV5; 44361-AAV5
AAV5-EF1a-DIO-hChR2(H134R)-EYFP	Hope Center Viral Vectors Core	N/A
Cowpox virus Brighton Red	BEI Resources	NR-88
Zika-SMGC-1, GENBANK: KX266255	Isolated from patient (Wang et al., 2016)	N/A
Staphylococcus aureus	ATCC	ATCC 29213
Streptococcus pyogenes: M1 serotype strain: strain SF370; M1 GAS	ATCC	ATCC 700294
Biological Samples
Healthy adult BA9 brain tissue	University of Maryland Brain & Tissue Bank; http://medschool.umaryland.edu/btbank/	Cat#UMB1455
Human hippocampal brain blocks	New York Brain Bank	http://nybb.hs.columbia.edu/
Patient-derived xenografts (PDX)	Children’s Oncology Group Cell Culture and Xenograft Repository	http://cogcell.org/
Chemicals, Peptides, and Recombinant Proteins
MK-2206 AKT inhibitor	Selleck Chemicals	S1078; CAS: 1032350-13-2
SB-505124	Sigma-Aldrich	S4696; CAS: 694433-59-5 (free base)
Picrotoxin	Sigma-Aldrich	P1675; CAS: 124-87-8
Human TGF-β	R&D	240-B; GenPept: P01137
Activated S6K1	Millipore	Cat#14-486
GST-BMAL1	Novus	Cat#H00000406-P01
Critical Commercial Assays
EasyTag EXPRESS 35S Protein Labeling Kit	Perkin-Elmer	NEG772014MC
CaspaseGlo 3/7	Promega	G8090
TruSeq ChIP Sample Prep Kit	Illumina	IP-202-1012
Deposited Data
Raw and analyzed data	This paper	GEO: GSE63473
B-RAF RBD (apo) structure	This paper	PDB: 5J17
Human reference genome NCBI build 37, GRCh37	Genome Reference Consortium	http://www.ncbi.nlm.nih.gov/projects/genome/assembly/grc/human/
Nanog STILT inference	This paper; Mendeley Data	http://dx.doi.org/10.17632/wx6s4mj7s8.2
Affinity-based mass spectrometry performed with 57 genes	This paper; and Mendeley Data	Table S8; http://dx.doi.org/10.17632/5hvpvspw82.1
Experimental Models: Cell Lines
Hamster: CHO cells	ATCC	CRL-11268
D. melanogaster: Cell line S2: S2-DRSC	Laboratory of Norbert Perrimon	FlyBase: FBtc0000181
Human: Passage 40 H9 ES cells	MSKCC stem cell core facility	N/A
Human: HUES 8 hESC line (NIH approval number NIHhESC-09-0021)	HSCI iPS Core	hES Cell Line: HUES-8
Experimental Models: Organisms/Strains
C. elegans: Strain BC4011: srl-1(s2500) II; dpy-18(e364) III; unc-46(e177)rol-3(s1040) V.	Caenorhabditis Genetics Center	WB Strain: BC4011; WormBase: WBVar00241916
D. melanogaster: RNAi of Sxl: y[1] sc[*] v[1]; P{TRiP.HMS00609}attP2	Bloomington Drosophila Stock Center	BDSC:34393; FlyBase: FBtp0064874
S. cerevisiae: Strain background: W303	ATCC	ATTC: 208353
Mouse: R6/2: B6CBA-Tg(HDexon1)62Gpb/3J	The Jackson Laboratory	JAX: 006494
Mouse: OXTRfl/fl: B6.129(SJL)-Oxtrtm1.1Wsy/J	The Jackson Laboratory	RRID: IMSR_JAX:008471
Zebrafish: Tg(Shha:GFP)t10: t10Tg	Neumann and Nuesslein-Volhard, 2000	ZFIN: ZDB-GENO-060207-1
Arabidopsis: 35S::PIF4-YFP, BZR1-CFP	Wang et al., 2012	N/A
Arabidopsis: JYB1021.2: pS24(AT5G58010)::cS24:GFP(-G):NOS #1	NASC	NASC ID: N70450
Oligonucleotides
siRNA targeting sequence: PIP5K I alpha #1: ACACAGUACUCAGUUGAUA	This paper	N/A
Primers for XX, see Table SX	This paper	N/A
Primer: GFP/YFP/CFP Forward: GCACGACTTCTTCAAGTCCGCCATGCC	This paper	N/A
Morpholino: MO-pax2a GGTCTGCTTTGCAGTGAATATCCAT	Gene Tools	ZFIN: ZDB-MRPHLNO-061106-5
ACTB (hs01060665_g1)	Life Technologies	Cat#4331182
RNA sequence: hnRNPA1_ligand: UAGGGACUUAGGGUUCUCUCUAGGGACUUAGGGUUCUCUCUAGGGA	This paper	N/A
Recombinant DNA
pLVX-Tight-Puro (TetOn)	Clonetech	Cat#632162
Plasmid: GFP-Nito	This paper	N/A
cDNA GH111110	Drosophila Genomics Resource Center	DGRC:5666; FlyBase:FBcl0130415
AAV2/1-hsyn-GCaMP6- WPRE	Chen et al., 2013	N/A
Mouse raptor: pLKO mouse shRNA 1 raptor	Thoreen et al., 2009	Addgene Plasmid #21339
Software and Algorithms
ImageJ	Schneider et al., 2012	https://imagej.nih.gov/ij/
Bowtie2	Langmead and Salzberg, 2012	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Samtools	Li et al., 2009	http://samtools.sourceforge.net/
Weighted Maximal Information Component Analysis v0.9	Rau et al., 2013	https://github.com/ChristophRau/wMICA
ICS algorithm	This paper; Mendeley Data	http://dx.doi.org/10.17632/5hvpvspw82.1
Other
Sequence data, analyses, and resources related to the ultra-deep sequencing of the AML31 tumor, relapse, and matched normal.	This paper	http://aml31.genome.wustl.edu
Resource website for the AML31 publication	This paper	https://github.com/chrisamiller/aml31SuppSite