Key role of T cell defects in age-related vulnerability to West Nile virus

We established a robust mouse model of the age-related susceptibility to WNV. In our hands, old mice exhibited increased susceptibility to WNV regardless of the infection route (i.p. or s.c.), the viral isolate of WNV Ia (NY-99, 31A, or 385–99; Table S1, Fig. 1; see Figs. 2 and 5; and not depicted), or the mouse strain (Fig. 1, C57BL/6; and see Figs. 2 and 5, C57BL/6; or Fig. S1, BALB/c) used, although, as expected, i.p infection produced lethal effects at a lower dose (1–20 PFU) compared with the more physiological s.c. infection (50–1,000 PFU). Overall, old mice were at least six times more susceptible to WNV as measured by survival rates over many viral concentrations (Fig. 1 A and not depicted). At low viral doses, that difference was drastically reduced and, in some experiments, disappeared (Fig. 1 A, Table S1, and not depicted), whereas at the high doses both old and adult animals succumbed to infection, suggesting that the viral dose is one of the principal determinants of selective mortality of old mice within a specific dose window.

In principle, this age-related susceptibility could be the result of a generalized inability of old mice to control the virus at the level of both the innate and adaptive immune systems, of the generalized inability to control the virus by one of these two components, or of their focused inability to control the virus in selected target organs. To investigate this issue, infectious viral titer was determined in different organs of adult and old mice and at different time points after infection. After infection with a viral dose that kills most old but not adult mice, we found no significant difference in viral titers between adult and old mice between days 2 and 5 (Fig. 1 B and Fig. S2). Likewise, early time points showed no difference in spleen viral titers (Fig. S2), and our survey of other organs (kidney, gut and liver) also failed to show viral titer differences between old and adult animals between days 3 and 10. We could not detect WNV in the blood past day 5 or in any other organs past day 10, and certain organs (lung and skin, except the site of injection) were negative throughout the course of infection, with the exception of the central nervous system. Indeed, we found significantly higher viral titers in the brains of old mice compared with the adult counterparts on days 8 and 10 after infection (actually between days 7 and 12 [not depicted]; Fig. 1 C, left and middle). Two additional lines of evidence showed that viral titers within the brain directly correlated to mortality. First, once the animals became moribund, they exhibited equivalent and high brain WNV titers regardless of age (Fig. 1 B, right). Moreover, when higher infection doses of WNV, lethal to both adult and old mice, were used, both adult and old mice again exhibited comparable and high brain viral titers (unpublished data). Therefore, in all subsequent experiments where we sought to dissect the immunological basis of vulnerability to WNV, we used the viral doses at which most old animals had high viral titers in brain and died but at which the majority of adult mice exhibited low brain WNV titers and survived.

We next asked whether the differences mentioned in the previous section could be explained by age-related defects in innate or adaptive immunity. As a model of pronounced innate immunity defect, we used IFN-αβ receptor (IFNAR)–deficient mice (IFNAR^−/−; Müller et al., 1994), which are unable to respond to type I IFNs (IFN-I) and are known to be highly susceptible to numerous infections (for review see Pestka et al., 2004), including WNV (Samuel and Diamond, 2005). Rag-1^−/− mice (Mombaerts et al., 1992), which have no T and B cells as a result of the lack of the recombinase essential for generation of T and B cell receptors and which are also susceptible to many infections, including WNV (Engle and Diamond, 2003), were used as a model of profoundly deficient adaptive immunity. Our results confirmed that both of these strains are highly susceptible to WNV (Fig. 2 A) but highlighted a difference in the mean survival times (MSTs; Fig. 2 A). Thus, IFNAR^−/− mice died rapidly after infection (MST, 5 d), which is consistent with the lack of innate defensive mechanisms, whereas the RAG1^−/− mice died within the same temporal window as old and adult mice (MST, 13 d). Because the MST of old mice was 13–14 d, we concluded that innate immunity in these mice, unlike that in IFNAR^−/− animals, was capable of containing the virus and fending off early WNV-mediated mortality. Therefore, any putative defects in innate immunity in old mice, if present, do not lead to early loss of viral control. To confirm this, we examined functional levels of IFN-I in the serum of adult and old animals. Although the results of this test cannot be taken as fully conclusive because of the low sensitivity of this assay (Fig. S3), we detected no major age-related differences within these confines. More importantly, any defect in innate immunity would have been expected to result in loss of viral control early after infection. However, this was not observed in any of the experiments designed to test early viremia and viral spread (Fig. 1 and Fig. S2).

Because the results suggested a key role for defects in adaptive immunity, we directly explored this possibility by performing adoptive transfers of old and adult spleen cells (Fig. 2 B, left) or T cells (CD4 and CD8; Fig. 2 B, right) into adult RAG1^−/− mice. In this series of experiments, we found that adult spleen cells, as well as the adult T cells, conferred significant protection to RAG1^−/− mice. In contrast, mice that received old cells were afforded weaker protection, which was not improved in the case of splenocytes and was significant for transferred old T cells (P < 0.04) compared with RAG1^−/− mice receiving no cell transfer. Direct comparison between the protective capacity of adult and old T cells revealed significantly better protection by adult T cells (P < 0.003; Fig. 2 B). Because antigen presentation and priming are not affected by either age or by the targeted deletion of Rag-1 in the adult RAG1^−/− mice, these results collectively strongly suggested that the process of aging impairs resistance to WNV at the level of generation of primary T cell responses.

IgM and the virus-specific CD8 and CD4 T cells were all implicated in affording protection against WNV in adult mice (Diamond et al., 2003; Shrestha and Diamond, 2004; Sitati and Diamond, 2006). Given the extensive literature on the decline of T cell immunity with aging and our results with T cell transfers, we initiated experiments to test possible defects in the T cell pool. We first examined signs of T cell activation using multicolor flow cytofluorometric analysis. We followed quantitative and qualitative aspects of antigen-specific responses using peptide MHC class I tetramers, as well as the functional response measured by the ability to produce IFN-γ, TNF, and granzyme B (GzB) and express CD43, a molecule involved in T cell activation, costimulation, and effector function (Onami et al., 2002), and to perform lytic function in response to recently identified immunodominant WNV peptide epitopes that stimulate CD8⁺(Brien et al., 2007) and CD4⁺ T cells (Brien et al., 2008; Figs. 3 and 4). At the peak of the response (days 7–8 after infection), we detected a strongly significant reduction in both percentages and numbers of CD8⁺ cells specific for the immunodominant class I–restricted epitope NS4b₂₄₈₈ by both tetramer staining (Fig. 3 A) and IFN-γ production after brief in vitro stimulation (Fig. 3 B). In addition to these quantitative defects, the ratio between IFN-γ–producing and NS4b₂₄₈₈ Tet⁺ cells was also significantly reduced in old CD8 cells so that less than one-half of Ag-specific cells were also making the cytokine (Fig. 3 C). That suggested the existence of superimposed qualitative defects in immunity of old mice against WNV.

To analyze the quality of the anti-WNV response in depth, we examined several aspects of the response, including the ability to produce polyfunctional cells, the intensity of cytokine expression/cell, the level of activation of CD8 cells as measured by CD43 expression, and the ability to lyse peptide-coated cells (Fig. 4). By each and every measure, old CD8 T cells were significantly inferior to the adult counterparts. Thus, although nearly 40% of adult CD8 cells expressed IFN-γ, TNF, and GzB (Fig. 4 A, 3 Fxn), only 21% of adult cells did so. Second, expression of each of these molecules per cell was significantly lower in old CD8 T cells (Fig. 4 B). Third, only about half of the WNV-specific (Tet⁺) CD8 T cells from old mice expressed high levels of CD43, an important activation and function marker (Fig. 4 C). Finally, direct ex vivo cytotoxicity correlated to GzB expression and was essentially nonexistent in CD8 T cells from old mice compared with adult counterparts (Fig. 4 D).

These findings were consistent in every experiment performed. Sometimes not all the parameters would reach statistical significance when smaller animal groups were used, but at least three out of four observations were always significant. Similarly, these trends and significant differences were confirmed with the CD8 epitope ENV₃₄₇ and also in CD4⁺ cells using the immunodominant CD4⁺ T cell epitopes (ENV₆₄₁ and NS3_1616+2066) by measuring frequency and magnitude of the IFN-γ (Fig. S4) and other responses (not depicted). Based on the prior descriptions of signaling defects in CD4 T cells (Garcia and Miller, 2003; for review see Miller, 1996), we expected that some of the stated problems in old T cells may trace to inferior ability of old T cells to process antigenic stimulus. However, experiments examining peptide sensitivity of WNV-specific CD8 and CD4 T cells showed superimposable sensitivity of old and adult cells at the peak of the response (Fig. S5). We conclude that old mice exhibit profound quantitative and qualitative defects in mobilizing fully developed effector T cells but that these defects do not extend to all aspects of antigen recognition and activation.

We next wanted to determine whether the observed reduction in IFN-γ secretion and/or GzB expression were relevant for protection against WNV in vivo. To that effect, we performed two types of adoptive transfer experiments using RAG1^−/− mice as recipients. First, we transferred total T cells from WT, IFN-γ^−/−, or perforin^−/− donors into RAG1^−/− recipients, which showed that T cells defective in IFN-γ or perforin provide negligible, if any, protection as compared with animals which received no cells at all (Fig. 5 A). This confirms and extends prior results on the importance of these molecules in anti-WNV protection (Engle and Diamond, 2003; Wang et al., 2003) and stresses the critical role of their expression in T cells. However, because old T cells have a reduction but not a complete absence of these molecules, we tested their antiviral ability in vivo by separately transferring highly purified (< 0.5% cross-contamination) CD4 and CD8 T cells from old or adult naive donors into adult RAG1^−/− recipients. These results convincingly showed that either adult CD4 or adult CD8 T cells were sufficient to confer significant protection to RAG1^−/− mice against primary WNV infection in the absence of other components of the adaptive immune system (Fig. 5 B). More importantly, neither the CD4⁺ nor the CD8⁺ T cells from the old mice were able to confer any protection upon RAG1^−/− mice over the level seen in the absence of transfer (Fig. 5 B), although the combination of the two old T cell subsets did show some synergy, affording a low level of protection (Fig. 2 B, bottom).

If our conclusions were correct, we would expect that they would hold at the level of key organs targeted by WNV. Alternatively, it was also possible that there may be defects in migration and localization of virus-specific T cells to the old brain. To address this issue, we examined the localization and composition of leukocyte infiltrate in the brains of WNV-infected old and adult mice. It is of interest that immunohistochemical examination revealed that in old mice, like in the adult animals, CD3⁺ cells readily infiltrated the brain parenchyma on days 7 (not depicted) and 12 (Fig. 6 A), suggesting that there was no major difference in migration and homing. In contrast, the composition of the infiltrate was both qualitatively and quantitatively different (Fig. 6, B and C). The large (activated) mononuclear cell gate (Fig. 6 B) contained nearly 3× fewer cells in old mice compared with adult counterparts (3.1 vs. 9.1%, P < 0.0005). Moreover, within that gate, the percentage of CD4 and CD8 cells was also about twofold lower with old age (5.6 vs. 11.4%, P < 0.013, and 14.1 vs. 24.3%, P < 0.037 for CD4 and CD8 cells, respectively; Fig. 6 C), leading to a >5× reduction in total CD8 and CD4 cells in old WNV-infected brains. Among these already diminished numbers, the content and quality of virus-specific cells was further reduced. Therefore, the percentage of NS4b-specific tetramer⁺ CD8 cells specific for WNV was reduced by another twofold (18.6 vs. 38.8%, P < 0.0012; Fig. 6 D), with another reduction in GzB-producing cells (trend but not significant at 3.3 vs. 6.7%, P < 0.06; not depicted). Most importantly, when all the reductions were taken into account, compared with the levels found in adult mice (taken as 100%), our analysis suggests that old brains contain at least 12.5× fewer and up to 20× fewer Tet⁺ GzB⁺ CD8 T cells (Fig. 6 D) compared with adult brains at the peak of infection.

Two important conclusions can be drawn from these results. First, migration to the site of major virus-induced tissue damage is not drastically affected in T cells of old animals. Second, although old T cells apparently arrived to the site of infection, they were not able to differentiate into effector T cells either systemically or locally in numbers sufficient to ensure control of neurovirulence. Therefore, our results identify defects in generation of sufficient number and quality of effector antiviral T cells as the key phenomenon underlying age-related susceptibility to WNV in this model, providing targets for potential therapeutic manipulation.

Our results show that specific age-related defects in T cell immunity, affecting both CD8 and CD4 T cells, underlie the susceptibility of old mice to WNV. This is in contrast to some of the other viral infections, such as Vaccinia, where no defects in GzB expression directly ex vivo can be detected (unpublished data), or influenza, where there is no reproducible age related decrease in survival in mice (Bender et al., 1995; Gardner, 2005) and where the loss of CD8 T cells was not accompanied by impaired viral resistance (Eichelberger et al., 1991), but it is similar to impaired lytic function of CD8 T cells against HSV-1 in old mice (Messaoudi et al., 2004). Numerous innate (IFN-I, complement, and innate IgM antibody) and adaptive (B cells and CD4 and CD8 T cells) mechanisms were implicated in anti-WNV resistance in adult mice (Diamond, 2005) but their relative importance and primary roles still remain incompletely mapped. Our results show that both T cell subsets play important and independent antiviral roles in adult mice, although there is evidence that they could also synergize (e.g., Fig. 2) against the virus. Adult CD4 and CD8 T cells were both able to secrete IFN-γ (results presented in this study) and to kill infected target cells (Brien et al., 2007, 2008; Purtha et al., 2007), which are critically important anti-WNV effector mechanisms (Shrestha et al., 2006; Sitati and Diamond, 2006; Fig. 5 A), and both functions were impaired with aging (Figs. 3 and 4). Moreover, adult T cells were polyfunctional and exhibited significantly more robust per cell responses at the level of effector molecule mobilization but did not exhibit higher TCR sensitivity (Fig. 4, Fig. S4, and Fig. S5). Of major importance are the results from adoptive transfers into RAG1^−/− mice, where adult, but not old, CD4 or CD8 T cells could confer significant anti-WNV protection. Because these animals possess functional (and young) innate and reticuloendothelial system components and only lack B, T, and NK-T cells, our results reveal that the defects in transferred aged T cells are cell autonomous in nature and are not precipitated by the aging of accessory and/or antigen-presenting cells. Consistent with this finding, examination of brain T cell infiltrates revealed normal migration in old mice but insufficient accumulation of differentiated effector T cells, which were 10–20× less numerous in the brains of old mice. This would suggest that the use of T cell rejuvenation therapies to remedy this condition may offer the best chance for therapeutic intervention. In this case, thymic reawakening and production of additional cohorts of Ag-specific T cells by the use of a combination of stem cell intervention to circumvent the block in lymphopoiesis and the stromal rejuvenation using KGF, androgen blockade, or IL-7 treatment (for reviews see Linton and Dorshkind, 2004; Nikolich-Žugich, 2005) would appear to provide the best approach.

However, it will also be important to understand other, perhaps more subtle, potential defects in innate and adaptive immunity. Thus, although our results strongly suggest that there are no major age-related defects in early virus control by the innate immune system, they do not exclude the possibility of correction of T cell defects by providing enhanced activation of innate immunity by cytokine manipulation (Haynes et al., 1999, 2004) and/or antigen presentation. Indeed, such approaches may provide more practical means for therapeutic intervention compared with T cell rejuvenation, which still remains to be effectively customized for mass utilization. However, one could imagine providing key cytokines or costimulatory molecules in repeated rounds of antigen-driven T cell and B cell expansion to enhance and improve effector T cell generation. It is our belief that such intervention may be most effective to remedy the defects described in this paper. Overall, given the multitude of defects that affect an aging immune system, it seems prudent to elucidate specific and key defects affecting resistance to each pathogen and then rationally target the identified defects for intervention.

Old (18–22 mo) and adult (4–6 mo) C57BL/6 (B6) mice were purchased from the National Institute of Aging breeding colony (Harlan). C57BL/6 RAG1^−/−, 129, C57BL/6 perforin^−/−, and C57BL/6 IFN-γ^−/− mice were purchased from The Jackson Laboratory and bred at the Oregon Health & Science University (OHSU) West Campus vivarium. IFNAR^−/− mice were a gift of A. Hill (OHSU, Beaverton, OR). All animals were housed and bred under specific pathogen-free conditions at OHSU and experiments conducted under the Institutional Animal Care and Use Committee and the Institutional Biosafety Committee approvals in accordance with all applicable federal, state, and local regulations. All WNV experiments were completed within a United States Department of Agriculture (USDA)–inspected Biosafety Level three facility.

WNV strains NY99, 31A, and 385–99 were used and all virus strains yielded similar results. WNV strains NY99 and 385–99 were gifts of R. Tesh (University of Texas Medical Branch, Austin, TX). Strain 31A was provided by the USDA. Peptides NS3 ₁₆₁₆, NS3 ₂₀₆₆, and NS4b _2488-2496 (21st Century Biochemicals) were diluted in 10% H₂O and 90% DMSO and stored at −80°C. Vero, MC57g, and EL-4 cell lines were mycoplasma negative and were cultured under aseptic conditions with DME (VERO and MC57g) or RPMI (EL-4) supplemented with antibiotics and 5% fetal calf serum. MC57gs were infected using an MOI of 10 for 30 h before use for ⁵¹Cr assays. MC57gs were infected using an MOI of 10 for 30 h before use for ⁵¹Cr assays.

Cytokine-producing T cells were detected using the Cytofix-Cytoperm kit (BD). Single-cell splenocyte suspension was depleted of red blood cells and was incubated with 1 µM of peptide or infected with WNV in the presence of 5 µg/ml brefeldin A (Sigma-Aldrich) for 6 h at 37°C. After 6 h, the cells were washed and blocked with Fc block (anti–mouse FcγRI/III; BD) and incubated overnight in the presence of a saturating dose of surface antibodies against CD8, CD3, CD4, CD11a, CD43 (clone 1B11), CD44, and CD62L (BD, eBioscience, and BioLegend). After washing, the cells were fixed and permeabilized, and intracellular antibodies (anti–IFN-γ or anti–IL-2; eBioscience) were added for 30 min. The samples were then washed and analyzed using either a FACSCalibur or LSR II cytometer (BD) instrument. GzB (clone gb12; Invitrogen) intracellular staining was completed directly ex vivo with no stimulation.

Serum levels of IFN-I were measured using a bioassay. Type I IFN standards (National Institute of Allergy and Infectious Diseases international standard) and experimental sample mouse serum were serially diluted twofold down a flat-bottom 96-well tissue culture–treated plate in complete media, 10% FBS, Pen/Strep, and DME. IFN-responsive L929 cells were plated at 5 × 10⁴ cells/well, incubated overnight with the serum, and media aspirated, and media containing 5 PFU VSV Indiana was added to each well, except for control wells. 24 h later, media was aspirated, the plate washed twice with PBS, and the monolayer fixed with 5% formaldehyde, incubated for 10 min, and stained with 0.05% crystal violet for 10 min. Washed monolayers were allowed to dry, and 100% methanol was used to elute the dye. Absorbance is measured at 595 nm on an ELISA plate reader (Invitrogen). Experimental samples are compared with the standard curve to determine endpoint values.

Mice were infected i.p. with 0.16–20 PFU of WNV virus or s.c. with 20–1,200 PFU per mouse as denoted in the figure legends. At indicated times after infection, lymphocytes were isolated and used for direct flow cytometry analysis, direct ex vivo restimulation for cytokine production, and direct ex vivo ⁵¹Cr assay. Percentage of specific lysis was calculated as [(E−S)/(M−S)] × 100, where E equals the counts per minute released from targets incubated with lymphocytes, S equals the counts per minute released from target cells incubated with no lymphocytes, and M equals the counts per minute released from cells after lysis with 1% Nonidet P40 (Affymetrix).

Old and adult T cells from corresponding mouse populations were enriched by positive selection of CD4 and/or CD8 T cells using MACS separation (Miltenyi Biotec) in accordance with manufacturer's protocol. Purity of obtained cells was 90–95% CD8⁺/CD4⁺ as determined by flow cytometry, and the opposite subset and B cells were present at <0.5% in all transferred populations. 2–10 × 10⁶ cells were injected i.v. and engraftment success was evaluated by flow cytometry 24 h later at the time of infection.

Animals were challenged with no other manipulations or 24 h after cell transfer with indicated doses and routes of WNV. Survival was scored on a daily basis. Death occurred between days 10 and 18, and all animals surviving this period remained disease free for >60 d, at which point the experiment was discontinued. Data are shown as percentage of survival at the termination of the experiment.

Animals were sacrificed and liver, spleen, brain, and kidney were removed and homogenized in RPMI using a beadbeater-96 (BioSpec). Samples were spun at 2,000 rpm for 10 min at 4°C and aliquots of each sample were stored at −80°C. Viral titer was determined by plaque assay by serially diluting sample onto Vero cells. After co-culture of the virus with the cells for 2 h, agarose overlay was added. 2 d after the initial overlay, cells were overlayed with additional agarose containing Neutral red (0.2%). Plaques where then counted to determine viral load.

Fisher's exact test and Log-rank test were used to analyze results from survival experiments. Statistical significance of viral titer observed between groups was analyzed using a Mann Whitney U test. Other tests are as indicated. All calculations were performed using Prism software (GraphPad Software, Inc.).

Fig. S1 shows survival of adult and old BALB/c mice infected with WNV. Fig. S2 shows viral titer in the organs of adult and old mice at indicated days after administration of a WNV dose lethal to old but not adult mice. Fig. S3 shows IFN-I activity in the serum of old and adult mice. Fig. S4 shows the response of adult and old CD4 T cells to class II MHC-restricted WNV peptide epitopes. Fig. S5 shows that peptide sensitivity of 8-d-old and adult CD8 and CD4 T cells is largely superimposable. Table S1 shows that old mice exhibit increased WNV mortality regardless of route of injection or viral isolate.

The authors would like to thank Drs. Robert Tesh (U. Texas Medical Branch, Galveston, TX) and Ian Lipkin (Columbia University) for providing WNV virus strains, Dr. Michael Diamond (Washington University) for constructive input, the members of the Nikolich laboratory for support and help, and the National Institutes of Health tetramer facility for expert assistance.

This work was supported by the United States Public Health Service awards N01 50027 and AI 81860 (to J. Nikolich-Žugich), T32 AI007472 (to J.D. Brien), and RR0163 (to the Oregon National Primate Research Center) from the National Institute of Allergy and Infectious Diseases and the National Institute for Research Resources, National Institutes of Health. J. Nikolich-Žugich is an Elizabeth Bowman Professor in Medical Research.

The authors declare no competing financial interests or commercial affiliations.