Construction and characterization of a recombinant virus expressing GFP in HSV-1 strain McKrae (GFP-McKrae)
HSV-1 has previously been modified to express fluorescent reporter genes that distinguish viral infection and pathogenesis in HSV- strains 1 KOS, RE, H25, SC16, and 17 (
27–31). For efficient ocular infection, these viruses require corneal scarification that, by itself, stimulates immune cell trafficking into the eye, as we previously reported (
26). Among all known HSV-1 strains, McKrae is the only virulent ocular strain that does not require corneal scarification. Thus, to identify infected cell types in the corneas of ocularly infected mice during primary infection and to detect GFP in TG during latency, we constructed a recombinant virus expressing GFP under control of the CMV promoter into the unique short (Us) region between gJ (US5; 137712–138125) and gD (US6; 139762–142585) in HSV-1 strain McKrae (
Fig. 1A) (
33,
34). The region between gD and gJ is approximately 400 bp and does not code for any gene and does not affect gJ or gD expressions. The presence of GFP in GFP-McKrae virus was verified by PCR and next-generation sequencing (NGS) of the whole viral genome (Gene Bank, accession ID: Mutant Human alpha herpesvirus one clone recombinant GFP-McKrae genomic sequence; PP680711), confirming presence of the GFP gene and one unique PacI site in the virus. In contrast to GFP-McKrae virus, we expected sequencing of the parental McKrae virus (Gene Bank, accession id OL638991.1) to confirm the presence of GFP and two PacI sites.
To determine whether GFP-McKrae virus expresses GFP protein, RS cells were infected with the virus at 0.1 or 1 PFU/cell for 12 or 24 h. Using confocal microscopy, cells infected with 1 PFU of virus expressed more GFP than did those infected with 0.1 PFU, and cells infected for 24 h expressed more GFP than did those infected for 12 h (
Fig. 1B). Subsequently, RS cells were infected with 0.05 PFU/cell of GFP-McKrae and McKrae viruses or mock-infected for 24 h and stained with anti-gD antibody as described in the Materials and Methods. In McKrae-infected cells, 82% were gD
+ and, as expected, no GFP
+ cells were detected (
Fig. 1C, WT-McKrae, right lower quadrant), whereas 98% of GFP-McKrae cells were gD
+GFP
+ (
Fig. 1C, GFP-McKrae, right upper quadrant). No gD
+ or GFP
+ cells were detected in mock-infected controls (
Fig. 1C, mock control). Thus, using fluorescence imaging and flow cytometry, our results suggest that GFP-McKrae virus does express GFP protein.
To determine how the presence of GFP in GFP-McKrae virus affects virus replication
in vitro, RS cells were infected with 0.1, 1, or 10 PFU/cell of GFP-McKrae or WT-McKrae viruses for 12, 24, or 48 h. Virus titers in cells infected with GFP-McKrae or WT-McKrae viruses were determined by standard plaque assay and did not show statistically significant differences at any PFU or time point (
Fig. 2A through C,
P > 0.05). These results demonstrate that addition of GFP upstream of gD sequences did not affect virus replication
in vitro. To determine if the GFP sequence affects expression of other key HSV-1 transcripts, RS cells were infected with GFP-McKrae or WT-McKrae viruses at 0.1 PFU/cell, and the copy number of gB, gC, and gD transcripts were measured by qRT-PCR at 48 h PI (PI). The copy number of gB, gC, and gD in GFP-McKrae-infected cells were similar to those in WT-McKrae-infected cells (
Fig. 2D,
P > 0.05), suggesting that the presence of GFP did not affect levels of gB, gC, and gD expression in infected cells.
The
in vitro results described above suggested that the presence of GFP did not alter the infectivity of GFP-McKrae virus when compared with WT-McKrae virus. We next asked if the presence of GFP affected virus replication
in vivo. Mice were ocularly infected with 2 × 10
5 PFU/eye of GFP-McKrae or HSV-1 McKrae viruses. Virus titers determined in tears of infected mice on days 1–7 PI using standard plaque assays were similar in both groups on all days, suggesting that presence of GFP did not alter virus replication in the eyes of infected mice (
Fig. 3A,
P > 0.05). Survival of ocularly infected mice was recorded for 30 days. In GFP-McKrae-infected mice, 27/31 (87%) survived ocular infection, whereas only 19/32 (59%) of WT-McKrae-infected mice survived ocular infection (
Fig. 3B,
P = 0.02, Chi-squared test), suggesting that GFP had a protective role during ocular infection. CS and angiogenesis in surviving mice were scored in a blinded fashion on day 30 PI as described in the Materials and Methods. A significant reduction in CS was observed between mice infected with GFP-McKrae and WT-McKrae viruses (
Fig. 3C,
P = 0.01), indicating that the GFP sequence impacted eye disease. Conversely, the angiogenesis score showed no significant difference (
Fig. 3D,
P > 0.05). These data suggest that GFP expression driven by the CMV promoter exerts a milder effect on eye disease rather than a broader effect on other related physiological processes such as angiogenesis.
To determine the effect of GFP presence on latency and reactivation, mice were ocularly infected with GFP-McKrae or WT-McKrae viruses, and individual TG was harvested on day 30 PI. qRT-PCR of total TG RNA was used to determine latency-associated transcript (LAT) expression levels with no significant difference in LAT expression seen between the two groups of infected mice (
Fig. 4A,
P > 0.05). Time to reactivation was also determined using an
ex vivo explant assay. Similar numbers of TG reactivated in the two groups (25 of 25 for GFP-McKrae, 19 of 19 TG for McKrae,
Fig. 4B). The average time to reactivation between the two groups was also similar, 4.0 ± 0.2 days for GFP-McKrae-infected and 3.5 ± 0.2 days for McKrae-infected mice, suggesting that GFP sequences did not affect latency levels or reactivation in infected mice. The results of analyzing virus replication, eye disease, latency, and reactivation suggested that GFP-McKrae and WT-McKrae viruses were similar. We next asked whether the GFP sequence affected expression of the cellular genes CD8, CD4, IFNγ, IFNα2, IFNβ1, CD80, CD86, CD28, PD-L1, CTLA4, PD-1, TGFβ, and FoxP3 by using qRT-PCR to measure their expression in isolated TG RNA. No significant differences were detected in expression levels of all tested genes (
Fig. 4C,
P > 0.05). Thus, GFP expression did not alter the function of immune genes in TG of latently infected mice. Collectively, these observations showed that GFP expression did not affect the rate or pattern of GFP-McKrae infection, which were similar to those of the parental WT-McKrae virus.
Spatial profiling using IMC detects spatially adjacent GFP-positive immune cell clusters in vivo
A longitudinal analysis of murine corneas using imaging mass cytometry (IMC) was conducted with customized scripts and standard statistical tools, following the methodology of the Bodenmiller pipeline (
36,
37). Two distinct images of day three corneas were analyzed: one of the entire cornea, and another of a small region used for standardization. Single sections of both uninfected corneas and corneas at day 5 PI were similarly analyzed. As IMC has been used successfully to study the spatial arrangement of cells in virus infections (
38,
39), we used an IMC-based approach to study whole corneas from uninfected mice, as well as corneas isolated from mice infected with GFP-McKrae virus on days 3 and 5 PI (
Fig. 5). Single-cell stacked images of uninfected corneas showed the spatial arrangement of immune cells in the epithelium (
Fig. 5A). These cells were arranged in 11 clusters based on expression of immune and non-immune cell markers. Clusters 2, 6, 7, and 11 expressing CD4, E-cadherin and F4/80 showed the presence of CD4
+ T cells, F4/80
+ macrophages, CD11b
+ population of macrophages, neutrophils, and DCs on E-cadherin
+ epithelium (
Fig. 5A). Other markers showed very low or no expression (
Fig. 5B). As expected, no GFP expression was detected in any clusters in uninfected mouse corneas (
Fig. 5C), confirming that uninfected corneas were not exposed to GFP-McKrae virus, and there was no background GFP staining. An interaction map was created to show the proximity of clusters in the cornea. Notably, clusters 3, 6, 8, and 10 were closely associated (
Fig. 5D) and expressed CD11b, CD4, FoxP3, and E-cadherin, suggesting that macrophages (CD11b
+, F4/80
+), DCs (C11b
+), and T cells (CD4
+, FoxP3
+) were adjacent in uninfected corneal epithelia (E-cadherin
+). We also identified spatially associated clusters 0, 1, 4, and 9, composed of T cells that expressed CD4, FoxP3, and CD3. The presence of T cells, macrophages, and DCs in uninfected corneas is consistent with our previous reports (
26,
40).
On day 3 PI with GFP-McKrae virus, the single-cell map showed a marked increase in immune cell clusters 0–12, (
Fig. 5E and F). Markers expressed at highest levels on day 3 included B220, CD11b, CD11c, CD45, F4/80, Ly-6G, aSMA, iNOS, E-cadherin, and GFP. Markers expressed at moderate levels included CD163, CD4, CD68, CD8a, and FoxP3. Cluster 12 included B cells (B220
+); cluster one included M1 macrophages (CD11b
+, iNOS
+, CD68
+), cluster nine included myofibroblasts (aSMA
+), clusters 2 and 11 included neutrophils (CD11b
+, cluster 1 Ly6G
+); clusters 1 and 8 included T cells (CD3
+, CD4
+, FoxP3
+), and clusters 4, 6, 0, and 1 included epithelial cells (E-cadherin
+). GFP distribution was measured in each cluster, with the highest GFP expression in cluster 6, and formed epithelial cells, which confirms higher virus infectivity (
Fig. 5G). Epithelial cells, macrophages, and DCs from clusters 1 and 0 also had high GFP expression. Clusters 2, 4, 8, and 9, including epithelial cells, neutrophils, fibroblasts, and DCs (CD11b
+, CD11c
+), expressed lower levels of GFP. Neighborhood analysis showed that epithelial cells, macrophages, and DCs, with high GFP expression, were adjacently situated (
Fig. 5H). Macrophages and B cells were also closely situated. These findings were confirmed by examining a different sample where single ROI was analyzed (
Fig. S1). The combined data suggest that infection of mouse corneas with GFP-McKrae virus caused an influx of immune cells on day PI, which formed distinct, close clusters containing epithelial cells and GFP-McKrae virus.
On day 5 PI, the single-cell cornea map showed immune cells together with corneal cells (
Fig. 5I). Markers expressed at high levels on day 5 included B220, CD163, CD3, CD4, CD45, F4/80, FoxP3, and GFP (
Fig. 5J); and those expressed at moderate levels included CD11b, iNOS, E-cadherin, and CD11c. Clusters 2–8, 10, and 11 expressed these markers. GFP expression was reduced on day 5 and limited to clusters 3, 6, and 7 (
Fig. 5K). These clusters included closely associated T-regulatory cells (CD3
+, CD4
+, FoxP3
+) and M2 macrophages (CD11b
+, F4/80
+, CD163
+) (
Fig. 5L). Interestingly, the epithelial cells from clusters 0 and 2 did not express GFP, suggesting that virus was cleared from the corneal epithelium. T cells, B cells, and macrophages were also detected on day 5.
The above results indicate that pro-inflammatory immune cells, such as neutrophils, DCs, and M1 macrophages, play important roles in clearing the virus on day 3 PI. The proportion of immune cells increases in anti-inflammatory M2 macrophages, and T-regulatory cells on day 5 PI, which are important to establish memory of the virus infection. The imaging mass cytometry (IMC) has effectively provided the spatial signatures of immune cells on various days post-infection. However, unresolved questions remain regarding the virus–host microenvironment. GFP expression in immune cells indicates two potential scenarios: (a) the immune cells were directly infected by the virus and maintained GFP expression, or (b) the immune cells phagocytosed the infected cells and subsequently acquired GFP expression. Taken together, the IMC-based method provides an improved platform to detect and study immune cell dynamics in uninfected and GFP-tagged HSV-1-infected mouse corneas.
Expression of GFP-McKrae and McKrae viruses during primary and latent infection in vivo
To examine GFP and gD protein expression in mouse corneas and TG during primary and latent infection, the mice were infected with GFP-McKrae or McKrae virus as described in the Materials and Methods. Eyes and TG were harvested on days 3, 5, and 30 PI, and tissues were sectioned and stained for GFP and gD proteins. Endogenous GFP signal was detected at 510 nM. Corneas were first immuno-stained to check for GFP and gD expression. Uninfected mice, used as controls, were processed in the same way as infected group. No GFP or gD expression was detected in uninfected corneas (
Fig. 6A). On day 3 PI, corneal images from mice infected with GFP-McKrae, had lesions with high GFP (endogenous GFP and anti-GFP) and gD expression (
Fig. 6B) as well as increased GFP and gD colocalization, suggesting that the virus has similar GFP and gD expression
in vivo. Corneas from mice infected with McKrae virus had only gD+ lesions (
Fig. 6C), with no GFP expression. We next examined GFP and gD expression in the eyes on day 5 PI with GFP-McKrae virus and, as expected, infected corneas had fewer regions with colocalized GFP and gD expression (
Fig. S3A), with reduced expression levels compared with expression on day 3. The expression of GFP and gD was observed in patches within the corneal stroma near conjunctiva, whereas the epithelial layer exhibited diffuse and low expression. McKrae virus-infected corneas also expressed gD, but not GFP, as expected (
Fig. S3B).
We next examined GFP and gD expression in TG sections. Uninfected TG did not express GFP or gD (
Fig. S4A), whereas on day 3 PI with GFP-McKrae virus, confocal images of TG were positive for GFP and gD expression (
Fig. S4B), with colocalized expression. On day 3 PI, McKrae-infected TG showed gD-positive regions (
Fig. S4C). Strikingly, on day 5 PI with GFP-McKrae virus, multiple TG regions were positive for GFP and gD expression (
Fig. 7A), with higher colocalization of GFP and gD. On day 5 PI, McKrae-infected TG continued to have multiple gD-positive regions (
Fig. 7B). To confirm our detection of gD expression in infected TG, TG were stained with anti-HSV-1 antibody. On day 5 PI with GFP-McKrae virus, multiple regions were positive for GFP and HSV-1, with colocalization of GFP and HSV1, suggesting a similar expression pattern for both viruses (
Fig. 7C). These results suggest that GFP insertion does not interfere with infection or with virus migration to sensory neurons of the TG. Taken together, these results demonstrate that GFP insertion into the virus does not affect virus establishment in the TG or gD expression.
To explore the effect of GFP on latency, the mice were infected with GFP-McKrae or McKrae viruses as described above. RNA isolated from TG of infected mice on day 30 PI were analyzed for gD, and GFP expression by qRT-PCR. Copy numbers were normalized to an endogenous GAPDH control. qRT-PCR analysis showed gD expression in 7 of 24 TG from GFP-McKrae-infected mice and in 3 of 19 TG from McKrae-infected mice, with no significant difference in gD expression between the two viruses (
Fig. 8A,
P > 0.05). The gD expression in TG of McKrae-infected mice varied from 0.5 × 10
7 to 2 × 10
7 copies and from 0.4 × 10
7 to 4 × 10
7 copies in GFP-McKrae-infected mice. The GFP copy number in latent TG of GFP-McKrae-infected mice (
Fig. 8B) was marginally increased in 7 of 24 TG, ranging from 100 to 800 copies in these mice. To determine whether viral proteins could be detected in latent TG sections from GFP-McKrae-infected mice, we examined sections for GFP, gD, or HSV-1. Reactivated explant TG were used to validate the positive GFP and gD signals. The explants displayed widespread GFP and gD expression in different regions of the TG (
Fig. 8C). Among GFP-McKrae-infected latent TG, one TG was positive for GFP and gD expression (
Fig. 8D), which we carefully examined to determine if the signal was truly positive for the virus. Serial sections were stained with HSV-1 antibody that detected whole virus (
Fig. 8E), and the region that was positive for GFP and gD expression (
Fig. 8D) was also positive for HSV-1, demonstrating that GFP-McKrae virus was maintained in a latent state in the TG. These results confirm and extend our previous studies that detected both gD RNA and gD protein in TG of latently infected mice (
41,
42).