Our study describes a role for the acute interaction of HSV-2 gD with HVEM in modifying the antiviral recall immune response after intravaginal infection of mice. Our principal observation was that engagement of HVEM during acute infection (priming) increased the HSV-specific CD8+ T-cell response at the mucosa during rechallenge. We investigated several possible contributing factors for this observation, but were unable to clearly demonstrate roles for CD4+ T-cells, Foxp3+CD4+ Tregs, or expression of the chemokine receptor CXCR3 on responding cells in relation to whether the priming virus was able to engage HVEM.
We used the well-established murine intravaginal challenge model for these studies [21
] to investigate the antiviral cellular recall response at the mucosa. This model is limited for studying HSV-2 pathogenesis in mice by mortality in productively infected animals when wild-type virus is used, requiring us to generate initial antiviral immune responses by priming with an attenuated virus. Mice can be protected against disease after intravaginal challenge with wild-type virus by a variety of means, including passive antibody transfer [30
] and by generation of HSV-specific cellular responses [23
], and from preliminary studies we knew that our attenuated viruses could provide protection against subsequent wild-type HSV challenge. We did not formally demonstrate the mechanism for this protection, but we expect that priming with either virus leads to both humoral and cellular immunity capable of mediating protection. Since our prior work [9
] and current data () also demonstrate that mutation of gD to abrogate interaction with HVEM has only minor effects on local viral replication in the intravaginal model, we believe our results reflect an effect that is primarily attributable to the relative ability of the priming viruses to engage HVEM. Memory T cell responses are generated from a minor population within the effector pool of CD8+
T cells [33
], and antiviral T cell activation and expansion are barely underway 24 hours after infection [34
]. Also, the numbers of memory T cells generated following resolution of an acute infection are thought to depend at least in part on the peak response during the effector phase of the cellular immune response [34
]; our prior data did not identify differences in the peak HSV-specific CD8+
T cell response on the basis of viral engagement of HVEM [9
]. Therefore, the small replication difference between the two viruses one day after inoculation (0.5
log PFU/mL) would be unlikely to explain the more than twofold difference in recall response at the mucosa.
A second potential alternative explanation for our findings is that the gB-specific recall response is specifically altered by engagement of HVEM during priming, leading to a recall response that is perhaps HSV-specific but involves subdominant epitopes. Consistent with this possibility is our observation that overall CD8+
T cell frequencies in mucosal tissue were unaltered in the recall response. However, our prior findings of no differences in the acute CD8+
T cell response to gB496–503
after challenge with virulent HSV-2(333)/gD compared to HSV-2(333)/Δ7-15, combined with the understanding that memory responses are partly dependent on peak responses [34
], suggest against this possibility. Nevertheless, we are unable to completely rule out this alternative explanation with our data.
HVEM is a member of the TNF receptor family, which is broadly expressed in hematopoietic cells [35
]. Signaling through HVEM results in different responses in immune cells depending on the context in which it is engaged [8
]. Engagement of HVEM by LIGHT or lymphotoxin-α
increases T-cell activation [35
], while BTLA [37
] and CD160 [39
] attenuate T-cell activation and proliferation upon interaction with HVEM. Although lymphocytes are not generally considered to be targets of infection with HSV in vivo
, the interaction of HSV gD on the viral envelope and on infected cells may modulate lymphocyte activity through an interaction with HVEM. HSV gD binds to HVEM in the same membrane-distal cysteine-rich domain (CRD1) [40
] as both BTLA [41
] and CD160 [39
], and soluble gD competitively inhibits the BTLA-HVEM interaction [38
]. The interaction of gD with HVEM can itself trigger NF-κ
B activation [42
]. Also, gD might competitively inhibit the binding of HVEM to BTLA or CD160, with consequences that depend on the effects of the HVEM-BTLA/CD160 interactions. The HSV gD interactions with HVEM may also alter responses of other (nonlymphocyte) immune cells which express HVEM or its ligands, such as dendritic cells, whose homeostasis is dependent on HVEM and BTLA signaling [43
], and NK cells, which may be activated by engagement of CD160 [44
Given the multiple combinations for binding between HVEM and its multiple ligands, between LIGHT and its binding partners HVEM and the lymphotoxin-β
R), and the differential regulation of expression of these molecules on different cell types during an inflammatory response, a mechanism by which any of the above interactions would be altered by gD to affect memory T-cell responses is not immediately obvious. Prior studies in BTLA-deficient mice show increased differentiation of naïve CD8+
T cells into central memory cells in the absence of BTLA [14
], suggesting that our results could be explained by interference of HVEM-BTLA signaling by gD during acute infection. Optimal Treg responses are also dependent on the HVEM-BTLA signaling pathway [17
]; upregulation of HVEM by Tregs and BTLA by effector T cells after TCR stimulation suggests the hypothesis that the level of Treg signaling to effector T cells through HVEM-BTLA during the acute response may program subsequent memory cell differentiation, controlling either the numbers or other characteristics of the memory cell population (e.g., migratory characteristics might be altered by effects on chemokine receptor expression). However, as other cell types also express HVEM and BTLA, including DCs and NK cells, a role for HVEM signaling by these cells in the shaping of the memory immune response is also possible.
Several lines of evidence suggest possible ways that the chemokine and cytokine environment within which the acute immune response is developing may influence the generation and persistence of memory cells. Antigen-specific CD4+
T cells have recently been shown to require expression of both LIGHT and HVEM to persist as memory cells [45
]. Expression of the chemokine receptor CXCR3 on T cells in DLNs has also been implicated in induction of T-cell memory [29
]. Any modulation of HVEM-LIGHT signaling by gD could affect either or both of these pathways. It is also possible that memory cell generation and persistence is not affected by the initial gD-HVEM interaction, but subsequent chemokine production or chemokine receptor expression by memory cells is programmed in some manner by the initial context of HVEM signaling, leading to changes in chemokine gradients during the recall response, which change the relative infiltration of different memory lymphocytes. We did not measure the chemokine response in the challenge phase in these experiments, and CXCR3 measurements did not reveal a role for expression of this receptor in the response we observed. Further work on defining the underlying mechanism for our observations is ongoing, including evaluation of different time points and experiments using adoptive transfer and HVEM knockout mice.
To our knowledge, this study is the first to show an influence of the HSV gD-HVEM interaction on recall immune responses in an HSV infection model. However, an influence of the gD-HVEM interaction on memory immunity is not entirely unexpected based on prior vaccine studies [10
]. In these investigations, candidate vaccines were constructed to express different viral antigens fused to the C-terminal domain of HSV-1 gD and delivered intramuscularly to mice. These constructs induced stronger immune responses than those which lacked either the gD fusion or in which gD was unable to interact with HVEM. The authors attributed this observation primarily to interference by gD with coinhibitory signaling by the natural BTLA-HVEM interaction.
Among the many questions left unanswered by our work is whether any advantage is conferred to the virus by manipulation of HVEM signaling pathways. It seems counterintuitive that HSV evolved to use a receptor that ultimately leads to a stronger recall response at the site of initial infection than if a different entry receptor had been used. One possibility is that initial establishment of infection is favored by the use of gD to disrupt HVEM signaling [9
]. If virus is able to reactivate and shed even in the presence of a strong recall response, it is possible that there is no significant selection pressure against this effect. Further investigation into the pleiotropic functions of HVEM in immunity may shed light on this question.
Finally, it is worth commenting further on the implications of this observation on pathogenesis of HSV disease and possible therapeutics, including vaccination. Studies of human trigeminal ganglia and skin biopsy samples strongly support the concept that memory CD8+
T-cell responses are critical for the control of recurrent infection [46
]. Manipulation of HVEM signaling to properly direct these responses to relevant tissues could benefit therapeutic vaccine strategies [11
]. There may also be implications for disease recurrence. Although a prior study of individuals with HSV-specific cellular immunity but no serologic or clinical evidence of infection failed to identify HVEM polymorphisms which altered viral entry into cells [48
], it is possible that HVEM variants may lead to signaling differences that either promote or diminish effective mucosal cellular immune responses during viral reactivation. A similar survey of HVEM variants in patients with frequent recurrences has not been completed.