Age-associated alterations in CD8α+ dendritic cells impair CD8 T-cell expansion in response to an intracellular bacterium

Age-associated decline in immunity to infection has been documented across multiple pathogens, yet the relative contributions of the aged priming environment and of lymphocyte-intrinsic defects remain unclear. To address the impact of the aging environment on T-cell priming, adult naïve OT-I TCR transgenic CD8 T cells, specific for the H-2Kb-restricted immunodominant OVA257-264 epitope, were transferred into adult or old recipient mice infected with the recombinant intracellular bacterium Listeria monocytogenes carrying the chicken ovalbumin protein (Lm-OVA). We consistently found that adult OT-I CD8 expansion was reduced in aged recipient mice, and this correlated with numeric, phenotypic, and functional defects selectively affecting CD8α+ dendritic cells (DC). Following Lm-OVA infection, aged mice failed to accumulate CD8α+ DC in the spleen, and these cells expressed much lower levels of critical costimulatory molecules in the first three days following infection. Further, aged CD8α+ DC showed impaired uptake of the bacteria at very early time points following infection. Treatment of aged mice with Flt3 ligand (Flt3L) improved the number of DC present in the spleen prior to Lm-OVA infection, and improved, but did not reconstitute, OT-I expansion to Lm-OVA infection. These results suggest that age-associated changes in antigen uptake, pathogen sensing, and/or antigen presentation contribute to impaired adaptive immune responses to microbial pathogens with aging.


Introduction
It is well established that older adults are more susceptible to many pathogens and respond poorly to vaccines. While various age-related impairments have been described in the innate and ⁄ or adaptive immune systems (reviewed in Haynes & Swain, 2012;Nikolich-Zugich et al., 2012;Solana et al., 2012), the relative importance of these defects remains incompletely understood and difficult to disentangle. Recently, we have shown that increased susceptibility to intracellular pathogens (i.e., West Nile virus, Listeria monocytogenes) in old mice correlates with numerically reduced and functionally impaired CD8 T-cell responses (Brien et al., 2009;Smithey et al., 2011). While these studies directly demonstrated the existence of a cell-autonomous age-related defect in T-cell immunity, contribution of defects in other cellular and molecular components of immunity that are important for T-cell triggering and differentiation was not examined and could not be excluded.
Understanding the totality of age-related changes in immunity, including delineation of the cell-intrinsic and environmental factors that contribute to suboptimal immune responses, is essential for developing optimal vaccines and immunomodulatory strategies for the aging population. In this study, we took advantage of the well-characterized L. monocytogenes (Lm) model system to study the effects of aging on both the early stages of DC infection by Lm, as well as the subsequent priming of CD8 T cells in vivo. The ability to mount robust CD8 T-cell responses against most pathogens depends upon efficient presentation of antigen by dendritic cells (DCs). In mice, up to six phenotypically distinct DC subsets have been categorized based on their expression of various surface markers (Shortman & Liu, 2002;Heath et al., 2004). Despite this heterogeneity, priming of CD8 T cells against intracellular viral and bacterial infections is mediated almost entirely by the CD8a+ dendritic cell subset Campisi et al., 2011). Further, this subset plays a unique role early following intravenous infection with Lm, facilitating bacterial entry, early survival, and replication within the predominant tissues of infection (Neuenhahn et al., 2006;Campisi et al., 2011;Edelson et al., 2011;Mitchell et al., 2011;Waite et al., 2011). While DCs recovered from adult and old mice appear to have equal capacity to prime T cells in vitro (Moretto et al., 2008;Jiang et al., 2009), few studies have directly examined these responses in infected animals in vivo, within the host microenvironments where they occur.
We found that the early establishment of Lm infection within splenic CD8a+ DCs was impaired in old mice and that this population failed to accumulate in old mice later, on days 3-5 post-infection. Although the bacterial burden eventually reached equivalent levels in adult and old animals, the alteration of the early stages of infection correlated with a reduced or bimodal expression of costimulatory molecules on the CD8a+ DC subset in old mice. Moreover, lack of accumulation of splenic CD8a+ DCs later in the infection correlated to reduced capacity of transferred adult CD8 T cells to accumulate in old recipients. Both the number of CD8a+ DC and the magnitude of CD8 T-cell expansion following Listeria infection in old mice could be partially restored by treatment with Flt3L treatment. These studies reveal that, in addition to the observed cell-intrinsic defect in expansion and effector differentiation in aged CD8 T cells (Brien et al., 2009), both the very early and later host-pathogen interactions at the level of old CD8a+ DCs can have a further negative impact on successful generation of adaptive immune responses in aging organisms.

Results
Poor expansion of adult CD8 T cells in old recipients following Lm-OVA infection factors in the aged host. Our general approach was to seed adult and old mice with equal numbers of identical adult OT-I TCR transgenic CD8 T cells and to compare their ability to expand in age-disparate recipient environments in response to microbial challenge. To this end, low numbers (5 · 10 3 ) of congenically marked adult OT-I CD8 T cells (CD45.1+) specific for OVA-8p (SIINFEKL, the H-2K b -restricted ovalbumin epitope) were adoptively transferred into adult and old C57BL ⁄ 6 recipients (CD45.2+). Following systemic infection with recombinant L. monocytogenes expressing the ovalbumin antigen (Lm-OVA), three to fourfold more OT-I cells were recovered in adult recipients relative to old (Fig. 1A).
It was possible that Lm-OVA may not produce the same amount of the OVA-8p epitope in adult and old hosts owing to any number of steps involved in early bacterial replication, and ⁄ or antigenic uptake, processing, and presentation. To control for such differences, we performed identical transfers but immunized the recipients with wild-type Lm (to keep the inflammatory context constant) admixed with OVA protein (eliminating any possible differences in Ag abundance) or the immunodominant SIINFEKL peptide (OVA-8p; eliminating differences in Ag abundance, uptake, and processing). The same three to fourfold difference was observed in each case, except that the absolute level of CD8 T-cell expansion was 100-fold lower with the OVA protein ( Fig. 1B) and about 4-5· lower with the OVA-8p compared with Lm-OVA ( Fig. 1C; note the difference in Y axis values in A-C), likely due to the much lower abundance of the epitope present when delivered outside the replicating pathogen. Another important control was to ensure that similar levels of engraftment of the transferred OT-I CD8 T cells occurred in the spleens of old and adult recipients. As it is technically very difficult to observe transferred cells when only 5 · 10 3 cells are transferred (with an estimated take of 10-15%, this would yield 500-750 cells animal), we performed control experiments using the same cohort of recipients but with a transfer of 2 · 10 5 OT-I cells into adult and old recipients. Analysis of the congenic CD45.1 ⁄ CD45.2 marker system confirmed identical short-term OT-I cell engraftment in the spleen 24 h later (in the absence of infection), at the expected 10-15% level of the original inocula (Fig. 1D).
Diminished adult CD8 T-cell responses in old mice could be owing to various T-cell extrinsic factors in the host. We therefore designed experiments to uncouple the stages of the immune response that contribute to the optimal expansion of CD8 T cells. While only a brief period of activation (24-48 h) is required to initiate proliferation of CD8 T cells (Kaech & Ahmed, 2001;van Stipdonk et al., 2001), sustained antigen presentation and inflammatory signals (>48 h) are required for robust CD8 T-cell expansion in vivo (Prlic et al., 2006). To investigate this question, we next transferred 2 · 10 6 carboxyfluorescein methylester (CFSE)-labeled adult OT-I CD8 T cells into adult and old recipients at various time points following Lm-OVA infection, and examined dilution by these OT-I cells after 48 h in vivo. In this way, the CFSE-labeled OT-I CD8 T cells acted as indicator cells to examine the availability of antigen that can prime naïve T cells throughout the first 10 days following Lm-OVA infection in adult and old recipient environments. Both the total number of OT-I CD8 T cells ( Fig. 2A) as well as the number of OT-I cells that had undergone at least one division (CFSE LO , Fig. 2B, with representative CFSE profiles in Fig. 2C) increased two to fourfold when transferred into adult mice between 3 and 5 days post-infection, suggesting a window of time during which conditions were optimal for expansion of naïve OT-I cells. There was relatively little expansion ⁄ accumulation of OT-I CD8 T cells in old recipients throughout the course of the infection (Fig. 2), suggesting that this window for expansion ⁄ accumulation, if present, is greatly reduced in aged animals. Although these results differ from previous work that observed a peak in 'functional' antigen display on days 1-2 days after Listeria infection (Porter & Harty, 2006), these prior experiments were performed with a reduced dose of Listeria and in the BALB ⁄ c genetic background which is 10-fold more sensitive to Listeria (reviewed in Garifulin & Boyartchuk, 2005), perhaps explaining the differences in kinetics. Our data suggest that the most robust accumulation of expanded T cells can be observed between 3-7 days following infection with higher Listeria infectious doses in the adult C57BL ⁄ 6 mouse and that old mice have a reduced capacity to prime ⁄ expand ⁄ accumulate OT-I CD8 T cells during this critical window.
Old mice fail to accumulate the CD8a+ DC required for efficient CD8 T-cell priming to Listeria at early time points It has been well established that CD8 T-cell responses to Listeria infection are dependent on priming by CD11c+ DC (Jung et al., 2002). To confirm the importance of the CD11c+ DC population for antigen presentation following systemic Lm-OVA infection, we isolated total CD11c+ DC, B cells, and macrophages from Lm-OVA-infected adult mice on day 3 post-infection. Each purified antigen-presenting cell population was cocultured in vitro with CFSE-labeled adult OT-I cells for 60 h to determine the priming capacity. We confirmed that only the CD11c+ DCs were able to induce division and CFSE dilution by adult OT-I cells (Fig. S1). The CD11c+ DC population can be further discriminated into (at least) three major DC subsets in the spleen based on the expression of CD8a and CD45RA: CD8a+ DC (CD8a+ CD45RA)), double-negative DC (CD8a) CD45RA)), and plasmacytoid DC (CD8a) CD45RA+) (Liu & Nussenzweig, 2010). Subsequent studies in the Listeria infection model have identified the CD8a+ DC subset as the cells responsible for efficient CD8 T-cell activation (Campisi et al., 2011;Edelson et al., 2011;Mitchell et al., 2011), as well as serving an important role in establishing a productive infection (Edelson et al., 2011;Mitchell et al., 2011;Waite et al., 2011). Therefore, we next determined the distribution of the total CD11c+ DC population into each of these subsets throughout the 11 days following Lm-OVA infection in adult and old mice ( Fig. 3, with representative gating shown in Fig. S2). In both naïve mice, and animals evaluated one day after infection, there was no difference in either the frequency or total number of splenic CD8a+, double-negative (DN), or plasmacytoid (pDC) subsets. However, on days 3 and 5 following infection, there was a marked increase in the critical CD8a+ DC subset in adult animals that was notably absent in the old mice ( Fig. 3A,D). This increase in the CD8a+ DC population in adult mice correlates with the time frame in which adult recipients of OT-I cells showed a greater capacity for CD8 T-cell expansion (Fig. 2). As previously shown (Neuenhahn et al., 2006;Edelson et al., 2011), when purified DC subsets from adult mice were co-cultured ex vivo 3 days after Lm-OVA infection, CD8a+ DC were the only cells able to stimulate OT-I division (Fig. S3). This highlights the potential functional importance of the inability to mobilize ⁄ accumulate this DC subset in aged mice on days 3-5 postinfection ( Fig. 3A,D).

Aged CD8a+ DC show poor upregulation of costimulatory molecules at early time points following Lm-OVA infection
The ability to prime naïve CD8 T cells is critically dependent on the upregulation of various costimulatory molecules on the DC surface. Upregulation of CD40, CD80, and CD86 on CD8a+ DC has been shown to occur within 24 h of Lm infection in adult mice (Mitchell et al., 2011). To determine whether differences in costimulatory expression may be contributing to the impaired proliferation of adult OT-I cells in aged recipient mice (Figs. 1 and 2), the expression levels of MHC-II, CD40, and CD86 were measured on splenic DC subsets prior to and on days 1, 3, and 5 following infection with Lm-OVA. Higher overall basal expression of class II and CD40 was seen on resting CD8a+ DC in adult relative to old mice ( . Note that all the mice were infected at the same time and that transfers were performed on different days, probably accounting for some of the variation in the total cell content over time in O mice from day 0 to day 12 in overall yields; nonetheless, expansion in these mice was clearly much reduced compared with the adult counterparts. Data represent two independent experiments with 4 mice group )1 time )1 point. *P < 0.05, ****P < 0.0001 by two-way ANOVA with Bonferroni post-test.
Impact of DC aging on CD8 cell priming, G. Li et al.
ª 2012 The Authors Aging Cell ª 2012 Blackwell Publishing Ltd/Anatomical Society of Great Britain and Ireland by geometric mean fluorescent intensity of these molecules. The age-related impaired upregulation of costimulatory molecules became more pronounced on day 1-3 post-infection, with old CD8a+ DC failing to upregulate any of them ( Fig. 4A-C). Interestingly, when we evaluated the histograms for costimulatory expression, it was clear that adult CD8a+ DC appeared to almost uniformly upregulate these costimulatory molecules for the first 3 days post-infection. In contrast, old mice showed a bimodal expression pattern for both MHC-II and CD40, and failed to achieve uniform upregulation. For adult and old pDC and DN DC subsets (whose role in CD8 T-cell priming to Lm is less clear), differences were more modest. The pDC subset appeared to undergo greater maturation in old mice than young, while the response of DN DCs was relatively comparable (Fig. S4).
The question remained whether the differences in maturation seen between adult and old DC reflect an inherent defect in maturation and upregulation of costimulatory molecules, or whether this is a consequence of some breakdown or alteration of the in vivo infectious process in old animals. To address this, we next purified the total CD11c+ DC pool from adult and old nä ive mice and stimulated in vitro maturation with either Lm-OVA or LPS. As shown in Fig. 4D-F, incubation with viable Lm-OVA was mildly less effective at maturing old CD8a+ DC compared with adults in vitro. In contrast, there was no difference in the capacity of the CD8a+ DC to upregulate costimulatory molecules following in vitro stimulation with LPS. Further, when LPS-matured adult and old DC were pulsed with SIINFEKL peptide and co-cultured with adult OT-I cells, equivalent OT-I proliferation was observed regardless of T ⁄ DC ratio or peptide concentration (Fig. 4G). Collectively, these experiments show that old DC can undergo efficient maturation in response to an 'easy' maturation signal in vitro (free LPS), but may be less capable of detecting and responding to a maturation signal provided by an intact pathogen ⁄ infection. Thus, age-related defects in CD8a+ DC recognition of pathogen and the subsequent upregulation of costimulatory molecules may have downstream consequences on the priming and ⁄ or accumulation of naïve CD8 T cells in the aged environment.

Aged mice poorly establish productive Lm-OVA infection within CD8a+ DC
Within hours of Lm infection, those bacteria that escape neutrophil killing in the spleen are almost uniquely found within the CD8a+ DC subset (Neuenhahn & Busch, 2007;Waite et al., 2011). In mice lacking this subset, productive infection with Listeria is severely diminished (Edelson et al., 2011). We have previously reported an overall decrease in the bacterial burden within the spleen of old mice relative to adults at 24 h post-infection, although this difference was no longer apparent by day 3 (Smithey et al., 2011). Although we found no differences in the number of CD8a+ DC in the spleen of uninfected aged mice (Fig. 3D), we wondered whether their uptake of bacteria from the bloodstream, and therefore the early kinetics of establishment of infection, might be altered. To address this, the CD8a+, DN DC and pDC subsets were sorted from the spleens of adult and old mice infected 14, 24 or 48 h previously with Lm-OVA (Fig. 5). At 14-and 24-hr post-infection, the number of bacteria within the CD8a+ DC subset (Fig. 5A), as well as within the unfractionated total splenocyte population (Fig. 5B), was significantly reduced (10-fold at 12 h) in aged mice, whereas, by 48 h post-infection, bacterial burdens were similar, suggesting robust replication of Lm in both old and adult DC. There were very few bacteria recovered from other DC subsets at 14 h post-infection, suggesting that the pattern of infection remains the same in aged mice, with CD8a+ DC being the primary reservoir of viable intracellular bacteria (Fig. 5C).
Collectively, our data indicate that initial bacterial uptake by this specialized cell subset is much less efficient in aged animals. This change in the early dynamics of infection in aged mice, with fewer bacteria present during the first 24 h, correlates with decreased and ⁄ or bimodal upregulation of costimulatory molecules by the CD8a+ DC population in aged animals at early time points postinfection ( Fig. 4A-C), and with decreased CD8 T-cell priming and accumulation ( Figs. 1 and 2).

Flt3 ligand treatment of old mice improves OT-I priming to Listeria
DC development is dependent on the growth factor Flt3L, and exogenous administration has been shown to both increase the number of CD8a+ DC in the spleen of young mice (Maraskovsky et al., 1996;O'Keeffe et al., 2003) and the bacterial burden in the spleen following infection with Lm (Alaniz et al., 2004). To evaluate whether old DC populations could be similarly expanded in vivo, old mice were treated for seven consecutive days with recombinant human Flt3L, and then the number of total CD11c+ DC as well as the CD8a+ DC subset in the spleen was determined. Although Flt3L treatment had a very modest effect on the number of total CD11c+ DC (Fig. 6A), it had a highly significant effect on the CD8a+ DC subset in old mice (Fig. 6B). To determine whether this improvement in CD8a+ DC numbers might improve the ability of the old environment to support OT-I proliferation after Lm-OVA infection, adult and old untreated mice as well as old mice treated with Flt3L underwent OT-I adoptive transfer followed by Lm-OVA infection. Seven days later, the number of OT-I cells recovered from Flt3L-treated old mice was significantly improved, albeit the number never approached that found in adults (Fig. 6C). These results suggest that increasing the overall DC population in old mice can improve their ability to effectively stimulate CD8 T cells. Limitations of this intervention and the implications of this result for our understanding of the relative importance of age-related defects in DC subsets are discussed below.

Discussion
We have previously reported impaired T-cell priming in aged mice infected with either West Nile virus (WNV) or Lm-OVA (Brien et al., 2009;Smithey et al., 2011), and for WNV our experiments clearly documented the importance of T-cell-intrinsic age-related defects. This work sought to clarify the contributions of the aged environment over and above the cellintrinsic problems within the aged T cells. Our data are consistent with a model in which CD8a+ DC serve two inter-related roles in the immune response to systemic Listeria infection, in such that the ability to prime an effective CD8 T-cell population to mediate Lm clearance is dependent on early and efficient establishment of infection within the CD8a+ DC subset.
What is the relationship between the very early events following infection and the subsequent ability to generate a robust T-cell response to the pathogen? Numerous reports have demonstrated the importance of the CD8a+ DC subset in the establishment of Lm infection in the spleen following i.v. inoculation of mice (Neuenhahn et al., 2006;Neuenhahn & Busch, 2007;Campisi et al., 2011;Edelson et al., 2011;Mitchell et al., 2011;Waite et al., 2011). How does this relate to the dominant role for this same subset in priming the CD8 T-cell response required for clearance of the bacteria? We propose that the inefficient priming of adult OT-I CD8 T cells in aged recipient mice is a function of at least two processes: (i) early problems with the uptake of Lm by the CD8a+ DC subset; and (ii) the concurrent problem in mobilization of these cells from DC precursors, as evidenced by the failure of old mice to increase the number of CD8a+ DC cells following Lm-OVA infection.
In adult animals, Lm escapes the phagosome rapidly, with 50% of the bacteria within the CD8a+ DC subset located in the cytosol within 6 h following i.v. infection (Muraille et al., 2007). It is presumed that intracellular bacterial replication provides a supply of Listeria-derived antigens for the class I MHC antigen presentation pathway, while bacterial by-products trigger multiple innate pathogen-recognition pathways (reviewed in Eitel et al., 2011). It is required that Lm escapes the phagosome and enters into the cytosol for efficient CD8 T-cell priming (Muraille et al., 2007). In theory, injection of either heat-killed Lm or strains incapable of escaping the phagosome should activate pathogen-recognition receptors, while bacterially derived products would be delivered to CD8a+ DC for cross-presentation to T cells. Yet, neither of these vaccination regimens leads to effective priming of protective T-cell populations in vivo (Lauvau, 2001), and in fact, Lm strains unable to escape the phagosome actively suppress the development of T-cell responses to cytosolically located bacterial antigens (Bahjat et al., 2009).
Thus, Lm infection within the cytosol of the CD8a+ DC population appears to be a key component of successful CD8 T-cell priming to this pathogen. Further, infected DC are relatively inefficient at killing intruding pathogens (Nagl et al., 2002;Alaniz et al., 2004), suggesting that productive infection of these cells, and their presentation of antigen from actively replicating intracellular pathogens, may be an important process in facilitating adaptive immune responses in general. Similar to infection with Lm, it has been shown that ex vivo antigen presentation by DC harvested 24 or 48 h following either subcutaneous or intravenous infection with HSV-1, influenza, or vaccinia virus infection is also limited to the CD8a+ DC subset , although the roles of direct infection vs. cross-presentation in these studies were not separated. Xu et al. (2010) have elegantly demonstrated the importance of direct priming, as opposed to cross-presentation or cross-dressing, by vaccinia-infected antigen-presenting cells as critical for stimulating CD8 T-cell responses, although that work did not classify which DC subset was responsible. Whether direct targeting of pathogens to the CD8a+ DC subset is a com-mon mechanism for eliciting adaptive immunity to intracellular pathogens remains unclear.
The efficiency of pathogen-sensing processes within aged DC is in the early stages of investigation. Following murine infection with influenza, aged DC show impairments in NLRP3 inflammasome ⁄ caspase-1 activation, and downstream IL-1b and IL-18 production, suggesting age-related impairments in DC pathogen sensing and ⁄ or innate activation (Stout-Delgado et al., 2008). In humans, similar problems are seen in aged DC differentiated from PBMCs, including reduced expression of many tolllike receptors (TLR), impaired IFN-I production following TLR stimulation, and alterations to the downstream IFN-I signaling response following exposure to TLR ligands (Qian et al., 2011). Following oral infection of aged mice with Endozoon cuniculi microsporidia, Moretto et al. (2008) observed age-related T-cell impairments in conjunction with defects in DC within the mesenteric lymph nodes. These studies demonstrated that aged DC cultured with E. cuniculi were impaired in their ability to produce either IL-12 or IL-15, and showed poor upregulation of costimulatory molecules. Our preliminary studies found no difference in IL-12 production by DC when evaluated ex vivo at early time points following Lm-OVA infection of adult and old mice (data not shown); however, these may simply reflect differences in the specific pathogens investigated, in vivo (this study) vs. in vitro (Moretto et al., 2008) infection of DC, as well as the timing of the experimental evaluation.
Pathogen sensing may not be the only problem in aged mice, and it is likely that efficient delivery of Lm into the CD8a+ DC subset as a result of age-related changes within the splenic environment may also compromise the establishment of infection. Brown et al. (2011) have demonstrated that changes to the splenic marginal zone architecture of aged mice have profound consequences for the development of prion disease after administration of scrapie. In this model system, i.v. injection of scrapie leads to uptake of complement-coated complexes within the marginal zone, and then transfer to follicular DC, where accumulation precedes neuroinvasion. Aged mice do not develop disease at same incidence as young animals, and this correlates with a failure to accumulate scrapie agent in the folicular DC areas of the splenic architecture. Immunohistochemical analysis revealed that the architecture of the marginal zone (where particulate antigens are filtered from the blood) is disrupted in aged mice, and the ability to trap pre-formed, IV injected complexes was compromised.
We are aware of one report of impaired phagocytosis by aged monocyte-derived DC isolated from blood samples from elderly human subjects (Agrawal et al., 2007), but to our knowledge ours is the first to suggest in vivo age-related phagocytosis defects in lymphoid DC. Whether our data truly reflect specific defects in Lm uptake by aged CD8a+ DC or rather are a consequence of age-related changes to the splenic marginal zone architecture that compromise filtration of particulate antigens from the bloodstream will require further clarification. The former explanation is supported by the fact that a similar strategy of Flt3L pre-treatment prior to Lm infection of adult mice was found to both increase the number of DC and cause greater bacterial burdens and mortality, even in the presence of a robust, protective CD8 T-cell response (Alaniz et al., 2004). Unfortunately, these experiments did not clarify whether treatment with Flt3L led to more DC being infected by Lm, or to higher bacterial numbers within individual cells. As the mechanism underlying Lm targeting into the CD8a+ DC has been recently demonstrated to depend on serum complement proteins and platelets (Edelson et al., 2011), such mechanistic studies are underway. Experiments to distinguish between these possibilities and to elucidate whether similar DC defects also further disadvantage T-cell responses to other intracellular infections in old organisms are currently under investigation. Our data suggest that poor establishment of Lm-OVA infection within the CD8a+ DC subset of aged mice within the first 24 h of infection (Fig. 5), and the lower costimulatory molecule expression by this subset over the first 3 days of infection (Fig. 4), may set up a less effective T-cell priming environment in old mice. This can be partially overcome by increasing the DC population in old mice by Flt3L treatment prior to infection (Fig. 6). However, it is also clear that this numeric increase in the DC available to take up Lm cannot truly compensate to increase T-cell priming, likely because the efficiency of bacterial uptake into individual CD8a+ DC remains compromised. Alternatively, there may be issues of DC recruitment that we have not been able to clearly evaluate owing to the lack of commercially available aged congenic mice for bone marrow chimera studies. The relationship between the nearly immediate problem of targeting Lm delivery to the splenic CD8a DC subset in old mice (Fig. 5), then problems in recruiting additional CD8a+ DC to the spleen in the first few days post-infection (Fig. 3), and the inability to upregulate costimulatory molecules on those CD8a+ DC are not entirely reconciled. Our OT-I transfer experiments indicate that the ideal conditions for T-cell priming are found somewhere between days 3-7, whereas the various defects we observed within the old CD8a+ DC subset occurred between 14 h and 3 days post-infection. One interpretation is that the use of adult OT-I transfers as a tool to detect 'optimal' antigen presentation conditions is somewhat misleading in that we may be measuring an environment that is above and beyond the conditions that are required for T-cell priming in an unperturbed system. Alternatively, these results may signal larger age-related problems affecting multiple arms of the DC populations that have not been previously appreciated.
The reduced capacity of the aged environment to prime adult OT-I CD8 T cells is reminiscent of truncated T-cell expansion seen when Lm infection is aborted early using either toxin-mediated depletion of CD11c+ DC (Prlic et al., 2006) or via antibiotic treatment (Porter & Harty, 2006). In contrast, antigen hyper-secreting strains of Lm appear to stimulate normal CD8 T-cell responses following early antibiotic treatment, suggesting both early events within the CD8a+ DC population as well as antigen density contribute to the magnitude of the response that develops (Smithey et al., 2008). Old mice do eventually reach the same CFU burden as adults as the Lm infection progresses (Fig. 5), and (presumably) antigen availability eventually catches up. Nonetheless, this is insufficient to overcome diminished early T-cell stimulation observed in the aged priming environment.
Defects in DC uptake of Lm, in reduced MHC and costimulatory molecule expression, and in accumulation of the critical CD8a+ DC subset at discrete times post-infection, provide another piece of the puzzle explaining why aged mammals fail to mount robust CD8 T-cell responses to Listeria. We conclude that both CD8 T-cell-intrinsic and environmental ⁄ antigen presentation issues contribute to impaired adaptive immune responses against intracellular bacterial pathogens in aged mice and suggest targeted manipulations to correct this defect. How alterations of CD8a+ DC function in aged organisms may influence adaptive immunity to other pathogens still remain to be fully elucidated.

Experimental procedures Ethics statement
All animal research was conducted in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The studies were approved by the University of Arizona IACUC under the protocols #09-005 and subsequently #08-102. No human research was performed in this study.

Mice
Old (18 and 22 months) C57BL ⁄ 6 (B6, H-2b) mice were obtained from the National Institute of Aging breeding colony (Harlan). Adult (2-3 months) B6 mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). OT-I TCR transgenic CD45.1+ mice were bred in our existing specific pathogen-free barrier colony. All mice were maintained under specific pathogen-free conditions in the animal facility at the University of Arizona and experiments were conducted under guidelines set by the University of Arizona Institutional Animal Care and Use Committee. Unless otherwise stated, mice received 1-3 · 10 4 colony-forming units (CFU) of Listeria. For all experiments, the number of inoculated bacteria was determined retrospectively by plating serial dilutions of the injected bacterial suspension onto brain-heart infusion (BHI) agar and counting colonies the next day.

Isolation of DC and other cell populations from spleens
Spleens were digested with Accutase (eBioscience, San Diego, CA, USA) for 30 min at 37°C, and then cells were passed through a 40-lm mesh screen to prepare single-cell suspensions for analysis. CD11c + cells were enriched by AutoMACS (Magnetically Activated Cell Sorter; Miltenyi Biotec, Bergisch Gladbach, Germany) before sorting into specific subsets by fluorescence-activated cell sorting (FACSAria, Becton Dickinson). Macrophage and B cells were isolated with CD19 or CD11b microbeads by AutoMACS. CD8 T cells were enriched by AutoMACS without digestion of spleen.

CFSE labeling
OT-I splenocytes were resuspended at 10 8 ⁄ mL in PBS +0.1% BSA, and then mixed 1:1 with a 10-lM working solution of CFSE. Cells were incubated in the dark for 10 min at 37°C, and then 5· the staining volume of ice-cold RPMI + 5% fetal calf serum (FCS) was added. Cells were incubated 5 min on ice, and then washed 3· more with RPMI + 5% FCS before used. In vitro dendritic cells maturation and OT-I proliferation CD11c+ DCs were isolated from adult and old mice as above, and graded numbers were seeded in 100 lL RPMI 1640 + 10% FCS in 96-well U-bottom plates. DC were stimulated overnight with either 0.1 lg mL )1 LPS (Sigma, St. Louis, MO, USA) or Lm-OVA at an MOI of 1. Costimulatory molecule expression on DC subsets was determined by flow cytometry. For OT-I proliferation, DC stimulated overnight with LPS were then pulsed for 5 h with SIINFEKL peptide. DC were washed twice; then, 5 · 10 4 enriched CFSE-labeled CD45.1+ CD8+ OT-I TCR transgenic cells were added a T:DC ratios of 2:1 to 20:1. Cultures were analyzed for proliferation after 60 h.

Analysis of antigen presentation in vivo
Mice were inoculated with Lm-OVA, and then at various time points post-infection; 2 · 10 6 CFSE-labeled OT-I TCR transgenic cells were adoptively transferred by intravenous injection into Listeria-infected or control recipients. At indicated days post-transfer, the proliferation of CFSE-labeled OT-I cells in the spleen was determined by flow cytometry. The transferred donor OT-I and endogenous recipient T cells were distinguished by flow cytometric detection of the CD45.2 molecule.

Ex vivo sorting of Listeria-infected dentric cells subsets
To enumerate intracellular bacteria within different cell subsets at early time points post-infection, spleens were processed through CD11c+ enrichment as above with all steps performed in the presence of 5 lg mL )1 gentamycin to kill extracellular bacteria. Cells were incubated for 45 min on ice with anti-CD11c, CD11b, CD8a, CD45RA ⁄ B220 antibodies to discriminate cell subsets, and CD3, CD19, and DX5 antibodies in a 'dump gate' to exclude contaminating B, T, and NK cells. Cell subsets were then sorted on a FACSAria cell sorter. 3000-300 00 cells of each DC subset were sorted into pure FCS. Cell purity was confirmed by flow cytometry and found to be > 95%. Cells were washed in sterile PBS, and then lysed in sterile water before plating serial dilutions onto BHI agar. CFU were enumerated after incubation at 37°C for 24 h.

Treatment of old mice with Flt3 ligand
Old mice were injected intraperitoneally for seven consecutive days with 10 lg recombinant human Flt3 ligand (eBioscience) in 100 lL LPS-free PBS. The next day, mice were infected with Lm-OVA.

Statistical analysis
Data are expressed as the mean ± SEM, and a representative experiment (of 2-4 repeats) is shown for each figure. All statistical analyses were performed by either Student's t-test or two-way ANOVA with Bonferroni posttests using Prism software (GraphPad Software Inc., San Diego, CA, USA). Probability values of P < 0.05 were considered to be significant. The following notations have been used to denote p values in all figures: *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001.

Supporting Information
Additional supporting information may be found in the online version of this article at the publisher's web-site: