The two faces of heterologous immunity: protection or immunopathology

CD4 T cell responses

The CD4 T cell plays a crucial role in protecting against viral infection and in the development of memory B cells and CD8 T cells. CD4 T cell help for CD8 T cells is important for primary and/or memory responses to some bacterial and viral infections [19]. CD4 T cells can be protective or exacerbate the course of infection under conditions of heterologous immunity. For instance, when CD8 and CD4 subsets from LCMV immune mice were adoptively transferred into naive mice, they resulted in heterologous immunity upon subsequent infections with Pichinde virus or VV [20]. However, the specific functions of heterologous CD4 T cell immunity have not been well-characterized thus far.

CD4⁺ T cell responses, which were once thought to be pathogen-specific, clearly can protect against diverse challenges. In one study, mice immunized against the BCG strain of Mycobacterium bovis (a vaccine against tuberculosis) were also protected against challenge with VV, a poxvirus. This immunity was mediated by memory CD4⁺ T cells and in particular, by IFN-γ [3]. These observations were interpreted as true cross-reactive heterologous immunity and not bystander activation of noncross-reactive cells by the innate immune system; however, the study did not determine the cross-reactive specificities.

Tregs are an important subset of CD4⁺ T cells (Fig. 1C). During chronic infections, the outcome may be influenced by the activity of one or more types of Tregs.

Unlike conventional CD4 and CD8 T cells, the majority of expanded Tregs is not antigen-specific after viral infection. Not surprisingly then, although Tregs have not been evaluated extensively in coinfection models, it appears that induction of Tregs by one pathogen can lead to suppression of bystander responses to subsequent pathogen challenge. Tregs modulate pathogenic and protective immune responses to infection in the host and may be important mediators that influence the outcome of coinfections [21, 22]. Tregs also can inhibit inflammatory reactions associated with chronic viral infections [23], and similar responses can limit the magnitude of protective immunity to an acute viral infection or that induced by a vaccine [24, 25].

The activity of Tregs varies, depending on the virus and site of infection. For example, depleting natural Tregs using anti-CD25 prior to infection enhances antiviral responses with virtually no enhanced immunopathology in a footpad model of HSV-1 infection [24]. In contrast, Treg depletion prior to corneal HSV-1 infection results in severe T cell-mediated tissue lesions [26]. These reports indicate that tissues may require different levels of protection from immune-mediated pathology. Other anti-inflammatory mechanisms, such as galectin-9 and T cell Ig domain and mucin domain 3 signaling, can also potentially modulate the virus-specific responses [27] to subsequent infections. For example, following viral infection, the host up-regulates immunoinhibitory receptors on CD8 T cells to limit immunopathology [27]. Thus, subsequent infections at this stage may encounter an immunosuppressive state that may limit the host's ability to clear the second infection but may also protect the host from associated immunopathology [28], although these pathways have been investigted less thoroughly than those of Tregs.

Cross-reactive T cells

T cells are characterized by the presence of heterodimeric receptors. The number of antigenic peptides that can be perceived by the immune system far exceeds that actual number of unique TCRs in an individual at any given time [29]. TCR cross-reactivity (also known as alloreactivity) has been suggested as a primary means of increasing the effective size of the TCR repertoire. Several mechanisms have been proposed for TCR cross-reactivity. First, it may be achieved through changes in the conformation of the flexible loops within the CDR3 region. Second, TCRs may use two very different binding modes to recognize two unrelated ligands. Other proposed mechanisms of TCR cross-reactivity include molecular mimicry [30], and also, the same TCR can recognize different peptide-MHC complexes, resulting in cross-reactivity.

Although cross-reactivity can be caused by amino acid sequence similarity between epitopes, several exceptions to this have been observed. For example, the immunodominant HLA-A*0201-restricted IAV epitope M158-66 and the major EBV-specific epitope BMLF1280-288 demonstrate cross-reactivity to each other but share only 3 of 9 aa. Interestingly, in several studies of heterologous viral infections in C57BL/6 mice, which have two Class I MHC molecules—D^b and K^b—more cross-reactivity is observed with K^b-restricted epitopes than with D^b-restricted epitopes [1]. A likely reason for this is that K^b presents an 8-aa peptide epitope instead of the 9-aa peptide, usually presented by D^b. Shorter peptides are more likely to result in cross-reactivity, with an increased chance for relevant molecular mimicry [1]. Structural studies have shown that K^b molecules have a deep pocket, allowing a smaller portion of the peptide to protrude and be available to engage the TCR, reducing the specificity of those interactions. Thus, variation among the binding features of MHC molecules may also contribute to heterologous immunity.

The presence of cross-reactive cells in infections has immunological consequences, resulting in substantial differences in the responses of a naive host compared with that of an immunologically experienced host [31,–33]. Cross-reactivity can narrow an otherwise oligoclonal TCR repertoire, changing the epitope-specific T cell hierarchies and alter immune outcomes [34]. Most likely resulting from a combination of these mechanisms, cross-reactive CD8 T cells have been detected in multiple heterologous viral infections [35]. As a result of their higher precursor numbers during heterologous challenge, cross-reactive memory cells are induced to expand and activate rapidly [36]. The role of these T cells varies from protective [37, 38] to pathogenic [39] (Fig. 1D and E). The outcome appears to depend, in part, on whether IFN-γ plays a protective or pathogenic role under the specific conditions in which the T cells are activated, as increased IFN-γ production is one of the major consequences of a cross-reactive response [31].

Narrowed TCR repertoire and viral escape

After an infection, a diverse repertoire of T cells is established to a variety of immunodominant epitopes [40] whose TCRs recognize antigens presented by MHC molecules [41]. CD8 T cell responses to a single epitope often use different Vβ families, and even within cells using the same dominant Vβ, many different clonotypes are present [42,–44]. A diverse TCR repertoire is important, as it may improve pathogen control and reduce the possibility of immune evasion by T cell escape variants [45, 46]. For example, in LCMV or PV infections of naive mice, the responding CD8 T cells constitute a very diverse repertoire with many distinct clonotypes [47]. When LCMV immune mice are subjected to homologous challenge (LCMV-LCMV), the repertoire of the responding CD8 T cells is largely conserved. However, when LCMV immune mice are challenged with the heterologous PV, the responding repertoire is narrowed significantly compared with that of naive mice infected with PV or the memory LCMV repertoire [47].

Additional examples of the narrowed TCR repertoire after heterologous challenge include secondary infections with HIV [46], HCV [45], and CMV [48]. Similarly, the Vβ TCR repertoire of some IAV epitope-specific memory CD8 T cells was altered after LCMV infection [2, 49]. Although the consequences are clear, we do not understand the mechanisms that drive this narrowing. A narrowed repertoire with a high-affinity clone in some circumstances may be beneficial during the early phase of infection [50], but it may also drive the increased generation of escape variants. Although repertoire narrowing might be expected to occur in most cases of cross-reactivity [47, 51], under conditions in which the baseline repertoire is already narrow, a heterologous challenge may force the recruitment of T cells that are not dominant clones but are expanded by the alternative epitope. Thus, heterologous infections not only have diverse effects on the magnitude of the cellular response but also at the level of the repertoire. These effects include: (1) the narrowing of the repertoire as a result of cross-reactivity, and (2) if the normal naive repertoire is narrow, heterologous infection can result in diversification by recruitment of nondominant clones. As an example of this, in a model with persistent CMV infection, followed by LM challenge, it was shown that persistent viral infection alters the naive CD8 T cell repertoire toward LM to an extent where there was no overlap between the LM responses in naive and persistently infected animals.

Altered T cell immunodominance hierarchies and diminished CD8 T cell responses to superinfecting viruses

Downstream of these effects on TCR repertoire, heterologous infection can alter the hierarchy of CD8-T cell responses. Most infectious challenges in the mouse model result in predictable hierarchies of CD8 T cell responses specific for the different peptide-MHC epitopes derived from the pathogen. These epitope-specific responses that form the immunodominance hierarchy may be dominant, codominant, or subdominant [52].

Cross-reactivity can alter the normal immunodominance hierarchy. In mice, a heterologous PV infection, followed by LCMV infection, results in the otherwise subdominant NP (205–212) epitope becoming dominant. This is most easily explained by cross-reactivity and could be the result of the presence of a higher precursor frequency of NP (205–212)-specific T cells and the fact that memory CD8 T cells respond more vigorously to antigen than do naive T cells. Thus, the original repertoire elicited by one infection can be altered drastically by a new, unrelated pathogen [34]. These examples demonstrate the variability that can result from challenging hosts with distinct immunological histories [53]. However, cross-reactivity by itself may not be the sole factor responsible for altered hierarchies.

Immunodominance results from the interplay of several complex factors, including the efficiency of peptide processing, the affinity of the peptide for the presenting MHC molecule, the overall number of presented peptide-MHC complexes, the availability of a TCR repertoire that recognizes the peptide-MHC complex, and immunodomination, where T cells specific for certain immunodominant epitopes suppress responses to other epitopes [31, 54]. Persistent infections may alter the inflammatory environment sufficiently to affect many of these features, which would result in altered dominance patterns. In latent murine CMV infection, a pronounced alteration in the T cell compartment occurs that is consistent with impaired, naive T cell function. This results in significantly weaker CD8 T cell responses to superinfection with influenza, human herpes virus I and West Nile virus, even 16 months after the original murine CMV infection [55].

Attrition of pre-existing CD8 T cells

Memory CD8 T cells are an important component of protective immunity to viral infections. Following the effector and death phases, a memory CD8 T cell population is established and maintained in the absence of antigen [56]. However, substantial evidence has shown that under some conditions of heterologous challenge, pre-existing immunity erodes, including the number of memory CD8 T cells [57]. Although cross-reactivity can induce a relative increase in the response to a challenge virus, when a secondary infection with strong Type 1 IFN-inducing virus occurs (Fig. 1F), the effect observed is attrition of previously existing memory T cell populations early during infection, in association with the IFN response. In this situation, heterologous immunity compromises the memory response against an antigen encountered previously.

This finding raises the disturbing possibility that vaccines that generate sufficient quantities of memory CD8⁺ T cells specific for an agent of interest could have catastrophic consequences for the host by displacing memory CD8⁺ T cells specific for previous infections [58] and vice versa, meaning a new infection could deplete memory cells generated by previous successful vaccination. However, these effects may be limited to specific infection combinations.

With the use of a heterologous prime-boost vaccination strategy [58], with LCMV and VSV, investigators reported contrasting results. In these experiments, the size of the memory CD8 T cell compartment doubled to accommodate the accumulation of newly generated EM CD8⁺ T cells. Importantly, the numbers of other cell types, such as CD4 T cells, B cells, and naive CD8 T cells, remained unchanged. One difference in these two contrasting reports [57, 58] is that the studies reporting attrition restricted their analyses to secondary lymphoid tissue after primary infections, which result predominantly in the generation of central memory CD8 T cells. Heterologous prime-boost vaccination, on the other hand, preferentially generates EM CD8 T cells that are present within the nonlymphoid compartment [58]. Furthermore, the concept of profound attrition of memory CD8 T cells on exposure to new pathogens may have depended on the ability of those infections to stimulate Type I IFNs, which can induce high levels of apoptosis in memory T cells early [57]. Thus, vaccination strategies that rely on the generation of Type I IFN responses should consider this issue.

Two additional studies provide insights into this question. In a model of LM infection and subsequent vaccination with modified VV Ankara, it was shown that there was only a slight loss of LM-specific memory CD8 T cells (consistent with findings in ref. [58]); however, pre-existing immunity to LM was lost as a result of alterations in the EM population [59]. Another group found that after infection with LM, adoptively transferred transgenic T cells that were not specific for LM antigens were depleted as a result of a lack of cognate antigen recognition [60]. These data support the observations made by Welsh and Selin [57].

The stable survival of memory CD8 T cells in mice can be disrupted by subsequent heterologous viral or bacterial infections in which the number of noncross-reactive memory CD8 T cells decreases via cytokine-dependent mechanisms [61,–66]. Broadly, there are two models of attrition: the passive-attrition model, where established memory T cells are replaced by newly formed memory T cells as a result of competition for limited space. In the active-attrition model, pre-existing memory cells undergo direct apoptotic attrition [31, 49, 60]. The IFN-mediated deletion of memory T cells mentioned above is an example of the active-attrition model. Additionally, the inhibitory molecule PD1, which was identified initially as a T cell exhaustion marker [67], also plays a role in the depletion of autoreactive CD8 T cells in mice [68]. It was also shown that PD1 is involved in the attrition of CMV-specific memory CD8 T cells after acute HBV infection (Fig. 1F).

The emerging field of predicting immune outcomes by mathematical modeling has suggested that if several infections occur simultaneously or within close intervals, then a pathogen that might normally be cleared could become persistent and held at low levels by a suboptimal immune response. In this scenario, infections are readily controlled initially, and the total size of the antigen-specific CTL memory pool is relatively small. When the host experiences new infections, viral load increases, as does the total CD8 T cell memory pool. However, viral control is degraded with each subsequent infection, so when the number of allowed infections is surpassed, the efficiency of viral control is lost. Thus, following each new infection, there is a progressive deterioration in the memory CD8 T cell compartment and an eventual failure to control new challenges.

Immunopathology

Whether a series of viral infections results in severe and prolonged lesions or can be resolved with minimal tissue damage depends on numerous factors. Two important contributors are the specific sequence of infections [39] and the route of infection. Consistent with this, when LCMV immune mice are infected with VV, they suffer a severe outcome [69], which neither of the viruses alone could induce [69, 70]. Interestingly, this pathology is not associated directly with viral load.

In naive animals infected with VV, there was robust infiltration of polymorphonuclear leukocytes and mononuclear leukocytes into the interstitium, peribronchial areas, and perivascular areas of the lung. However, VV-infected, LCMV immune mice demonstrate the induction of bronchus-associated lymphoid tissue [37] in the lungs, which is infiltrated with LCMV-specific CD8 T cells. The presence of these activated T cells contributes to the obstruction of the bronchiole (bronchiolitis obliterans). This immunopathology is dependent on a potent Th1 response, as the lesions in the fat and lungs were dependent on the production of IFN-γ [70].

Similarly, LCMV-infected, IAV immune mice have altered immune kinetics compared with that of naive animals, with increased numbers and activation states of CD8 T cells. Under specific infection conditions, IAV-induced memory CD8 and/or CD4 T cells play a role in enhancing lung immunopathology and can even contribute to increased viral titers [71].

In some instances, immunopathology is regulated by factors other than “traditional” immune mediators. For example, influenza infection triggers a generalized stress response that leads to a sustained increase in serum glucocorticoid levels, resulting in a systemic suppression of immune responses. This immunosuppressed state prevents immunopathology in mice coinfected with influenza and LM but at the same time, results in increased bacterial burdens [72]. Alterations in gut microbiota and mucosal immunity by previous viral infections can exacerbate subsequent pathology. For example, underlying CMV infection alters mucosal immunity, which is associated with an increased tendency of colitis to develop in CMV-infected hosts [73].

Immunopathology/cross-reactivity

CD8 T cell attrition in humans

The question of whether the development of memory cells to a new infection indeed impairs the pre-existing memory T cell population is of special interest in humans, as they encounter many different antigens during their lifespan. However, as a result of limitations in available reagents and the difficulty in tracking primary infections in humans, very little data addressing this question are available.

In a study of CMV-seronegative renal transplant recipients who experienced a primary CMV infection as a result of receiving a kidney transplant derived from a CMV-seropositive donor, the absolute numbers of pre-existing memory T cells in peripheral blood were not affected by the appearance of CD8⁺ T cells specific for the new CMV infection. Instead, the CD8⁺ T cell pool expanded to make space for the newly generated, virus-specific memory cells [83].

During acute EBV infection, pre-existing CMV- and influenza-specific memory CD8⁺ T cells showed signs of bystander activation. The numbers of CMV- and influenza-specific T cells were comparable before and after acute EBV infection [84]. These data are consistent with the idea that in humans, a robust CD8⁺ T cell response creates a new memory CD8⁺ T cell niche without substantially depleting pre-existing memory for heterologous infections. However, studies involving coinfections with CMV and hepatitis virus have produced opposing results: PD1 is important in the attrition of CMV-specific CD8⁺ T cells during an acute infection of HBV [85].

Similar to observations in the mouse model, some controversy persists about the extent of T cell attrition after heterologous challenge. Focusing on the stability of a single epitope-specific pool, one study reported a half-life of 8–16 years for VV-specific CD8⁺ T cells [86]. However, VV-specific CD8⁺ T cells were not detectable in 50% of the subjects who had been vaccinated >20 years earlier, suggesting a significant loss of human memory T cells in a subset of individuals.

Bystander activation

Bystander activation of T cells occurs in the absence of cognate antigen. Several hypotheses have been advanced to explain the underlying mechanisms of this kind of CD8 T cell activation. It is generally believed that most of these cells are activated by nonspecific mechanisms, including cytokine-driven activation. IFN and IL-15 can mediate bystander activation in mice [62, 87, 88]. However, the definitive mechanisms of bystander T cell activation in humans still remain to be determined.

Bystander activation of T cells appears to be a common occurrence during acute infection in humans, and not all activated CD8 T cells observed in peripheral blood during acute viral infection are virus-specific, as shown in HIV [89] and hepatitis B infections [90]. During acute EBV infection, pre-existing CMV- and IAV-specific memory CD8 T cells showed signs of bystander activation, including up-regulation of granzyme B [84].

Herpes virus-specific CD8 T cells also display an increased production of the antiviral cytokine IFN-γ during the acute phase of heterologous viral infection. In a clinical study of herpes viruses, such as EBV and CMV, the number of activated virus-specific T cells and total T cells was analyzed in patients with an infection of acute hepatitis B, DENV, influenza, or adenovirus. All acute viral infections triggered the activation and expansion of herpes virus–specific cells, which in turn, contributed to the heterologous, antiviral T cell response. This finding led to the question: what is the biologic significance of herpes virus-specific CD8⁺ T cell activation during acute heterologous viral infections? One possibility is that the activation/proliferation state of herpes virus-specific CD8⁺ T cells counteracts the attrition exerted by the expansion of the CD8⁺ T cells specific to the acute infection [90]. Therefore, the activation of the herpes virus-specific CD8⁺ T cells might prevent the reactivation of the herpes viruses.

Therapeutics

TNF is required for naive mice to control VV infection. However, with prior LCMV infection, the host does not need TNF to mediate efficient VV clearance. Thus, the use of anti-TNF therapies to treat various diseases (e.g., Crohn's disease and rheumatoid arthritis) may be relatively safe, perhaps, in large part, as a result of heterologous immunity. Similarly, humans are exposed to many pathogens throughout their lives; thus, we have a large, complex pool of memory T cells that have the potential to cross-react with any new pathogen and mediate heterologous, protective immunity [98].

Elevated CMV-specific humoral immunity (an indication of failed virus control) has been correlated with increased mortality of elderly individuals and impaired vaccination responses [99]. Treatment with the antiviral drug valaciclovir, for as long as 12 months in mice with established murine CMV infection, reduced the magnitude of the CMV-specific CD8⁺ T-lymphocyte response by 80% and reduced the IAV load following challenge [100].

In humans and primate models, narrow oligoclonal responses to infection have been associated with poor prognosis and the generation of epitope-escape viral variants. These results can facilitate clinical decisionmaking by providing the information necessary to select the appropriate treatment.

Transplantation

Heterologous immunity is a major concern in tolerance induction. A transplant recipient's prior infection history can govern his/her responsiveness to alloantigens, and this history is an important factor in the design of novel methods to achieve optimal graft survival, as humans have such large, complex memory T cell populations. Sequential infections with heterologous viruses can increase the frequency of T cells specific for alloantigens, thereby generating a large pool of memory cells that need to be tolerized before engraftment [53]. A critical threshold of memory T cells is needed to promote graft rejection, and CD8⁺ “central” memory T cells are primarily responsible for this transplantation-associated complication.

Anti-CD25, rapamycin, and blocking mAb against the common γ-chain have all been reported to act synergistically with a costimulation blockade to inhibit allograft rejection [101,–103]. However, one possible consequence of these efforts to suppress T cell activation and induce tolerance to the allograft is that antiviral immune responses could be compromised, leading to undesirable consequences for the transplant recipient. Conversely, ongoing viral infections could be a barrier to the successful induction of allograft tolerance. For example, persistent infections, such as HCV, EBV, and CMV, could be a threat to the clinical application of costimulation blockade-based, tolerance-induction regimens. Even more importantly, in the absence of antiviral therapy, such regimens also have the potential to hamper protective antiviral immunity.

Treatment with deoxyspergualin, an inhibitor of NF-κB translocation, and costimulation blockade synergistically impairs memory T cell activation and promotes antigen-specific tolerance of memory [104]. Furthermore, in transplant recipients, the pathogenicity of viruses may be enhanced by concurrent infection with virus. This may be a result of virus–virus interactions or virus–host interactions, resulting from modulation of the host cell functions, production of suppressive cytokines, or secondary effects on host immune responses. Thus, additional therapeutic modalities are needed to address viral infections in transplant recipients [105].

This work was funded by U.S. National Institutes of Health Grant AI091938; the National Institute of Allergy and Infectious Diseases, U.S. National Institutes of Health, under Contract Numbers HHSN266200700005C (“St. Jude Center of Excellence for Influenza Research and Surveillance”) and HHSN272200800058C (“A Systems Biology Approach to Infectious Disease Research”); and American Lebanese Syrian Associated Charities (ALSAC) at St. Jude Children's Research Hospital.