T-Cell Infiltration Correlates with CXCL10 Expression in Ganglia of Cynomolgus Macaques with Reactivated Simian Varicella Virus

Abstract

Ganglia of monkeys with reactivated simian varicella virus (SVV) contained more CD8 than CD4 T cells around neurons. The abundance of CD8 T cells was greater less than 2 months after reactivation than that at later times and correlated with that of CXCL10 RNA but not with those of SVV protein or open reading frame 61 (ORF61) antisense RNA. CXCL10 RNA colocalized with T-cell clusters. After SVV reactivation, transient T-cell infiltration, possibly mediated by CXCL10, parallels varicella zoster virus (VZV) reactivation in humans.

TEXT

Varicella zoster virus (VZV) causes varicella (chickenpox) and becomes latent in ganglia, producing zoster (shingles) upon reactivation. Because VZV infects only humans, studies of virus latency and reactivation have been restricted to autopsy tissues. Analyses of ganglia obtained after death from individuals with recent zoster revealed lymphocytic infiltration (1, 2), possibly mediated by antigenic stimuli or chemokines, including CXCL10 (3). VZV-specific T cells have not been identified in human ganglia latently infected with VZV (4, 5). Simian varicella virus (SVV) infection in monkeys closely resembles VZV infection in humans (6). Earlier, we demonstrated reactivation of latent SVV in immunosuppressed monkeys (7). Herein, we extended those studies by examining ganglia containing reactivated SVV for infiltrating T cells. Four cynomolgus macaques (GP02, -04, -06, and -07) were naturally infected with SVV (7). Ten to 14 days later, all monkeys developed varicella (Fig. 1). At 4 months postinfection, monkeys were immunosuppressed with tacrolimus, resulting in a 34% reduction in mean white blood cell counts at 6 weeks posttreatment (7). Monkeys GP02, -06, and -07 developed zoster at 23, 3, and 10 days, respectively, after starting tacrolimus. Monkeys were euthanized at monthly intervals post-tacrolimus treatment (Fig. 1). Detection of SVV glycoproteins in lungs and multiple ganglia in monkey GP04, which did not develop skin rash, confirmed subclinical reactivation (7).

Fig 1.

Establishment of latent SVV infection in cynomolgus macaques and reactivation by tacrolimus-induced immunosuppression. Four cynomolgus macaques (GP02, GP04, GP06, and GP07) were exposed to monkeys inoculated intratracheally with 10⁴ PFU of SVV and developed varicella rash 10 to 14 days later. At 4 months (4m) postexposure, all 4 monkeys received tacrolimus for the remainder of their lives. Zoster rash developed 26, 3, and 10 days (closed triangles) after tacrolimus treatment in monkeys GP02, -06, and -07, respectively. All 4 monkeys were euthanized 1 to 4 months postimmunosuppression. Subclinical reactivation of SVV was confirmed in monkey GP04 by detection of SVV glycoproteins in ganglia removed at necropsy (open triangle) (6).

Immunohistochemical analysis of consecutive ganglion tissue sections from these monkeys and from an uninfected control monkey (CTRL) for CD3, CD4 and CD8 expression showed that SVV reactivation was associated with T-cell infiltration, mostly CD8 T cells, in ganglia along the entire neuraxis (Fig. 2). T cells were dispersed throughout ganglia. T-cell clusters, of both CD4 and CD8 T cells, were occasionally detected adjacent to neurons (Fig. 2A). Rare granzyme B⁺ (grB) cells, not restricted to T-cell clusters, were observed in ganglia from all monkeys (Fig. 2A), suggesting that ganglion-infiltrating T cells did not encounter their cognate antigen (8). Ganglion-infiltrating CD8 T cells in zoster patients are also predominantly grB negative (9).

Fig 2.

Increased number of ganglion-infiltrating T cells and T-cell clusters early after SVV reactivation in monkeys. (A) Immunohistochemical staining of consecutive sections of lumbar ganglia from monkey GP02 at 4 days after zoster rash for CD3, CD4, CD8, and granzyme B (grB) antigens or IgG1 and IgG2a isotype control antibodies. Staining was enhanced in sections stained for CD8 and the corresponding IgG1 isotype control using tyramide signal amplification (TSA). Arrows indicate T-cell clusters. Arrowheads indicate occasional grB⁺ cells. Stainings were visualized using the substrate 3-amino-9-ethylcarbazole (AEC) (red), and nuclei were counterstained with hematoxylin (blue). Magnifications: ×200 for CD3, CD4, and CD8 cells and the corresponding isotype controls IgG1 and IgG1 plus TSA and ×400 for grB and the corresponding isotype control IgG2a. (B) Numbers of positive T cells and T-cell clusters (≥10 T cells clustered around a single neuron) compared to the number of neurons in the same section. Stainings were performed as described previously (5). Data are average ratios/monkey ± standard error of the mean. The Mann-Whitney test was used for statistical analysis.

We analyzed 11 to 19 ganglia from each monkey to determine the number of T cells per neuron. The number of ganglionic neurons counted in sections from each anatomical level of the neuraxis ranged from 657 to 3,991 (Table 1). Ganglia obtained from monkey GP02, euthanized at 4 days post-zoster rash, contained significantly greater numbers of CD3 (P = 0.0005), CD4 (P = 0.015), and CD8 (P < 0.0001) T cells/neuron compared to those in the control monkey (Fig. 2B) (Table 1). Monkey GP02 also contained significantly more CD3 T-cell clusters, defined as >10 T cells surrounding a single neuron, compared to the control monkey (P = 0.019). Similarly, ganglia from monkey GP04, euthanized at 2 months postimmunosuppression, contained significantly more CD8 T cells (P = 0.008) and had more CD3 cells (P = 0.075) and T-cell clusters (P = 0.095) than the control monkey. In contrast, T cells and T-cell clusters in monkeys euthanized at 3 and 4 months postimmunosuppression (monkeys GP06 and GP07) were similar to those of the control monkey (Fig. 2B). Overall, SVV reactivation led to a transient T-cell infiltration of ganglia, predominantly CD8 T cells, which clustered around neurons. Notably, T-cell numbers and clusters were not increased in ganglia corresponding to the site of zoster rash (Table 1).

Table 1.

CXCL10 transcript expression and numbers of infiltrating T cells in sensory ganglia from SVV-infected cynomolgus macaques

Monkey	Ganglion	No. of copies of CXCL10/ng of GAPDH	Total no. of neurons counted	Ratio (mean ± SD)
				T cells/neurons	CD4+ T cells/neurons	CD8+ T cells/neurons	CD8+/CD4+ T cells
GP02	Cervical	NAa	NA	NA	NA	NA	NA
	Thoracicb	3.6	2,080	1.18 ± 0.18	0.24 ± 0.44	0.64 ± 0.09	2.85 ± 0.80
	Lumbar	12.0	1,265	1.48 ± 0.44	0.23 ± 0.11	0.74 ± 0.17	3.67 ± 1.46
	Sacral	11.6	1,766	1.41 ± 0.29	0.28 ± 0.09	0.83 ± 0.12	3.31 ± 1.06
GP04	Cervical	1.3	1,684	0.70 ± 0.17	0.11 ± 0.05	0.30 ± 0.08	2.91 ± 1.57
	Thoracic	1.5	1,862	0.97 ± 0.24	0.13 ± 0.02	0.43 ± 0.06	3.32 ± 0.84
	Lumbar	1.8	1,610	1.91 ± 1.66	0.03 ± 0.01	0.71 ± 0.37	31.29 ± 17.18
	Sacral	2.5	709	1.52 ± 0.19	0.08 ± 0.11	0.50 ± 0.15	9.93 ± 1.06
GP06	Cervical	0.4	3,521	0.58 ± 0.13	0.23 ± 0.11	0.23 ± 0.05	1.10 ± 0.27
	Thoracic	0.9	2,514	0.45 ± 0.05	0.11 ± 0.07	0.14 ± 0.00	1.56 ± 1.05
	Lumbar	0.8	3,991	0.70 ± 0.07	0.18 ± 0.11	0.29 ± 0.14	2.48 ± 1.66
	Sacralb	0.9	1,908	0.80 ± 0.14	0.13 ± 0.04	0.18 ± 0.03	1.70 ± 0.71
GP07	Cervical	0.3	704	0.80 ± 0.14	0.19 ± 0.09	0.31 ± 0.08	2.11 ± 1.25
	Thoracicb	0.4	2,636	0.82 ± 0.08	0.07 ± 0.02	0.19 ± 0.05	3.03 ± 0.99
	Lumbar	0.4	3,884	1.02 ± 0.25	0.23 ± 0.10	0.41 ± 0.18	1.88 ± 0.37
	Sacral	0.2	1,421	0.68 ± 0.28	0.12 ± 0.07	0.33 ± 0.16	2.93 ± 0.55
BI79	Cervical	NA	657	1.00 ± 0.42	0.30 ± 0.18	0.42 ± 0.10	1.59 ± 0.61
	Thoracic	NA	1,260	0.83 ± 0.11	0.13 ± 0.03	0.37 ± 0.07	3.05 ± 1.14
	Lumbar	NA	2,181	0.68 ± 0.22	0.10 ± 0.03	0.12 ± 0.09	1.09 ± 0.86
	Sacral	NA	1,854	0.75 ± 0.07	0.18 ± 0.01	0.30 ± 0.01	1.67 ± 0.02

^aNA, not available.

^bSite of zoster rash (7).

SVV open reading frame 61 (ORF61) antisense transcripts are abundant in latently infected ganglia of rhesus macaques (10, 11). Thus, we analyzed consecutive ganglionic sections after reactivation by in situ hybridization (ISH) for SVV ORF61 antisense RNA and by immunohistochemistry (IHC) for CD3 expression and SVV antigen to determine if SVV infection might mediate T-cell infiltration of ganglia. T cells did not cluster with neurons expressing SVV ORF61 antisense RNA and SVV antigens (Fig. 3A). Figure 3B is a positive control showing that SVV ORF61 antisense RNA and SVV antigens were readily detected in skin of an acutely infected monkey. The absence of SVV ORF61 antisense RNA in ganglia of our cynomolgus monkeys could reflect a difference in species studied or the well-documented variability of ganglionic infection within a single subject (7). T-cell infiltration in ganglia of monkeys after SVV reactivation did not correlate with local expression of SVV ORF61 antisense transcripts or SVV antigens.

Fig 3.

CD8 T-cell infiltration in ganglia of cynomolgus macaques after SVV reactivation did not correlate with local expression of SVV antigens or with the ORF61 antisense transcript. (A) Consecutive ganglionic sections from monkey GP02 at 4 days after zoster rash stained for the SVV ORF61 antisense transcript by in situ hybridization (ISH) or stained for CD3 and SVV antigens by immunohistochemistry. (B) Biopsied skin rash sections from an acutely infected African green monkey stained for the SVV ORF61 antisense transcript by ISH or for SVV antigens by IHC. Controls were biopsied skin samples stained for E. coli diaminopimelate B (DapB) and normal rabbit serum (NRS) by ISH and IHC, respectively. IHC stainings (5) were visualized using substrate 3-amino-9-ethylcarbazole (AEC) (red), and ISH staining was visualized using the substrate Fast Red (pink). The ISH probes were designed by Advanced Cell Diagnostics (Hayward, CA), and ISH was performed according to the manufacturer's instructions. Nuclei were counterstained with hematoxylin (blue). Magnification, ×200.

Increased expression of the chemokine CXCL10, which recruits activated T cells and NK cells by binding to their receptor, CXCR3 (12, 13), has been described in human ganglia after VZV infection in vitro and in situ in human ganglia of deceased zoster patients (3). To test the role of CXCL10 in the SVV monkey model, consecutive ganglionic sections were stained for CD3, CD4, and CD8 by IHC and for CXCL10 transcripts by ISH. CXCL10 transcripts were readily detected within T-cell clusters, but not within neurons (Fig. 4A). CXCL10-positive cells were more abundant in monkeys at 1 or 2 months than at 3 to 4 months postimmunosuppression (data not shown). The number of CXCR3-positive cells in ganglia at reactivation did not differ from that in uninfected ganglia (data not shown).

Fig 4.

T-cell infiltration correlates with CXCL10 mRNA expression in ganglia of cynomolgus macaques after SVV reactivation. (A) Consecutive tissue sections of ganglia from monkey GP02 stained for CD3, CD4, and CD8 antigens by immunohistochemistry (IHC) as described previously (5) and for CXCL10 transcript by in situ hybridization (ISH) or stained using the corresponding IgG1 isotype control antibodies by IHC (with or without tyramide signal amplification [TSA]) and the negative-control probe directed to the E. coli gene diaminopimelate B (DapB) by ISH. Stainings were TSA enhanced in sections tested for CD8 and the corresponding IgG1 isotype control. IHC stainings were visualized using the substrate 3-amino-9-ethylcarbazole (AEC) (red), and ISH staining was visualized using Fast Red (pink). The ISH probes were designed by Advanced Cell Diagnostics (Hayward, CA), and ISH was performed according to the manufacturer's instructions. Nuclei were counterstained with hematoxylin (blue). Magnification, ×400. (B) Average CXCL10 transcript levels (copies of CXCL10 mRNA/ng of GAPDH [glyceraldehyde-3-phosphate dehydrogenase] mRNA ± standard error of the mean) per monkey. (C) Scatter plots of CXCL10 transcript levels versus the average number of CD3⁺ T cells/neuron, CD4⁺ T cells/neuron, or CD8⁺ T cells/neuron and the number of T-cell clusters/1,000 neurons in the contralateral ganglia of the same monkey. Spearman's correlation test was used for statistical analysis. Real-time RT-PCR analysis was performed as described previously (11).

To determine whether CXCL10 transcript levels correlated with infiltrating T cells, RNA extracted from ganglia contralateral to the ones used for IHC were analyzed for CXCL10 transcription by quantitative RT-PCR (qPCR) using the following oligonucleotide primers and probe specific for human CXCL10: CXCL10 forward primeer, GCCAATTTTGTCCACGTGTTG; CXCL10 reverse primer, GGCCTTCGATTCTGGATTCA; and CXCL10 probe, TCATTGCTACAATGAAAAAGAAGGGTGAGAAGAG. Consistent with CXCL10 detection by ISH, CXCL10 transcript levels were greatest in monkeys GP02 and GP04 at 1 and 2 months after immunosuppression (Fig. 4B and Table 1) and correlated significantly with the number of T cells (P = 0.003) and T-cell clusters (P = 0.0046). Note that CD8 (P = 0.0005) but not CD4 (P = 0.47) T cells correlated with CXCL10 transcript levels (Fig. 4C).

Our findings demonstrate that SVV reactivation induced a transient CD8 T-cell infiltration in monkey ganglia, possibly mediated by the chemokine CXCL10. The magnitudes of CXCL10 expression and T-cell infiltration did not differ among ganglia, including those that corresponded to the dermatome associated with zoster rash. Our study parallels the previous findings that VZV reactivation induces CXCL10 expression and T-cell infiltration in human ganglia (3) and underscores the usefulness of equivalent findings in monkey ganglia containing reactivated SVV and SVV infection of primates as a useful model to study VZV pathogenesis (6). In both settings, however, it remains unclear whether CXCL10 is the primary cause of T-cell influx or whether the influx is secondarily due to gamma interferon secreted by activated infiltrating lymphocytes (14, 15). Future studies of latently infected ganglia and ganglia from monkeys with zoster may help to differentiate between the two possibilities.