The route of priming influences the ability of respiratory virus–specific memory CD8⁺ T cells to be activated by residual antigen

To investigate the factors that regulate the recruitment of memory CD8⁺ T cells to the lung airways, we needed an experimental approach that would generate memory T cells at a distal site to the lung. We speculated that an i.p. infection offered a means to generate substantial numbers of effector and memory CD8⁺ T cells in the absence of a pulmonary infection. To confirm that virus is not transferred to the lung in the course of an i.p. infection, we analyzed the immune response and viral loads after i.p. infection with Sendai virus (Fig. 1 A). Even in the absence of a productive lung infection, i.p. infection elicited a robust Sendai-specific CD8⁺ T cell response in the spleen and lung parenchyma (Fig. 1 B). Notably, the numbers of Sendai-specific CD8⁺ T cells in the spleen and lung parenchyma were similar between i.p.- and i.n.-infected mice during the memory stages, indicating no requirement for i.n. virus infection in the steady-state recruitment of antigen-specific memory CD8⁺ T cells to the lung parenchyma (Klonowski et al., 2004). The i.p. infection protocol also elicited memory CD8⁺ T cells that were phenotypically similar to those elicited by i.n. infection (Fig. 1 C).

In contrast to the lung parenchyma, the ability of i.p.- and i.n.-primed memory CD8⁺ T cells to localize to the lung airways was vastly different. As shown in Fig. 1 B, after i.n. infection a robust peak of recruitment to the lung airways was observed at day 10 after infection, followed by sustained continual recruitment of antigen-specific memory CD8⁺ T cells at least 2 mo after infection. After i.p. infection, however, very few Sendai-specific CD8⁺ T cells were present in the lung airways during both the acute and memory phases. This limited recruitment of i.p.-primed cells to the lung airways is not a unique feature of Sendai virus infection or peritoneal infection because an identical migratory pattern was also observed after i.p. infection with influenza virus or after subcutaneous Sendai virus infection and peptide vaccination (unpublished data). Collectively, we conclude that antigen-specific memory CD8⁺ T cells originally primed via nonpulmonary routes show a severe defect in recruitment to the lung airways under steady-state conditions.

Based on previously published studies, we hypothesized that the defective recruitment of i.p.-primed memory CD8⁺ T cells to the lung airways was the result of a lack of residual antigen in the lung-draining LNs (Zammit et al., 2006). In support of this hypothesis, Sendai-specific memory CD8⁺ T cells in the MLN after i.n. infection were mostly CD69⁺, suggesting recent antigen stimulation, whereas memory CD8⁺ T cells in the MLN after i.p. infection were mostly CD69⁻ (Fig. 2 A). In addition, after i.n. infection the majority of CD69⁺ memory CD8⁺ T cells in the MLN also expressed high levels of the integrins CD49a and CD103, which are associated with retention in the lung and extravasation through mucosal epithelium, respectively (Ray et al., 2004; Sung et al., 2006). In contrast to the MLN, both the frequency of cells expressing these integrins and the relative intensity of expression were low on i.n.-primed memory CD8⁺ T cells in the spleen and CD69⁻ memory CD8⁺ T cells in the MLN (Fig. 2, A and B, top). In the case of i.p. infection, up-regulation of these integrins was not observed on antigen-specific cells regardless of CD69 expression (Fig. 2, A and B, bottom). However, residual antigen is present in the MLN of i.p.-primed mice (Fig. S1), suggesting that the limited activation of i.p.-primed memory CD8⁺ T cells was a consequence of the infection route and not a lack of residual antigen.

The rapid migration of antigen-bearing dendritic cells from the lung to the MLN is a well characterized process after respiratory virus infection (Legge and Braciale, 2003; Belz et al., 2004). Although it has been previously shown that the peritoneal cavity can also drain to the MLN, it was possible that the differences observed between i.n.-primed and i.p.-primed memory CD8⁺ T cells in this tissue were a result of the differential localization or altered activation status of APCs within the LN (Marco et al., 1992). Thus, we used a dual transfer approach to directly compare the ability of i.n.-primed and i.p.-primed memory CD8⁺ T cells to be reactivated by residual antigen under identical conditions in vivo. Surprisingly, after transfer into the same congenic host only i.n.-primed memory CD8⁺ T cells were activated by residual antigen, as measured by expression of CD69 (Fig. 2 C). Importantly, this activation required cognate antigen stimulation, as neither i.n.-primed nor i.p.-primed Sendai-specific cells expressed CD69 after transfer into mice that had been primed with influenza x31 (Fig. 2 D).

To determine whether the inability of i.p.-primed cells to recognize residual antigen was a consequence of differential T cell localization within the MLN, we used a TCR transgenic system to visualize antigen-specific i.n.- and i.p.-primed cells in the same host (Fig. 3 A). As shown in Fig. 3 B, both i.n.-primed (CD90.1⁺, top) and i.p.-primed (CD45.1⁺, bottom) OT-I cells localized to the T cell– and dendritic cell–rich areas in the MLN. Furthermore, consistent with our previous observations, i.p.-primed OT-I cells in the MLN failed to express CD69, CD49a, and CD103 (Fig. S2). Therefore, despite being localized to similar areas of the LN, i.n.- and i.p.-primed cells have a differential ability to be reactivated in the MLN. Collectively with Fig. 2, these data demonstrate that the route of initial priming dictates the ability of memory CD8⁺ T cells to recognize residual antigen.

The finding that i.p.-primed memory CD8⁺ T cells are not activated by residual antigen in the MLN suggested that even if cognate antigen were present, i.p.-primed cells would exhibit less steady-state recruitment to the lung airways than i.n.-primed cells. To test this hypothesis, we performed dual adoptive transfer studies to directly compare i.n.- and i.p.-primed memory CD8⁺ T cells under identical conditions in the same host. At 1 wk after transfer, the relative recruitment of i.n.-primed cells to the airways was much higher than that of i.p.-primed cells (∼60 fold; Fig. 4 A and Fig. S3). In contrast, the relative recruitment to the MLN, lung parenchyma, and pleural cavity was similar between i.n.- and i.p.-primed cells, indicating that the inability of i.p.-primed cells to be reactivated by residual antigen impacts their ability to migrate specifically to the lung airways. Furthermore, the enhanced recruitment of i.n.-primed cells to the airways was dependent on antigen restimulation, as the relative recruitment between i.n.- and i.p.-primed cells was equivalent when transferred into mice primed with an irrelevant virus (Fig. 4 B). To investigate whether the enhanced recruitment to the airways was limited to steady-state conditions, we investigated the recruitment of i.n.- and i.p.-primed cells to the airways in response to localized inflammation. After administering CpG to the airways, both i.p.- and i.n.-primed memory CD8⁺ T cells were rapidly recruited to the airways with equal efficiency (Fig. 4 C). In addition, secondary effector CD8⁺ T cells generated from an equal number of i.n.- and i.p.-primed memory T cell precursors after transfer into a naive host were recruited to the airways with equal efficiency at the peak of the secondary response (Fig. 4 D). Therefore, the inability of i.p.-primed memory CD8⁺ T cells to be reactivated by residual antigen resulted in a dramatic impairment of their recruitment to the lung airways, but this impairment was manifest only during steady-state conditions.

Although the previous data have shown that the inability of i.p.-primed memory CD8⁺ T cells to be efficiently recruited to the lung airways is a result of their inability to be reactivated by residual antigen in the MLN, and that a subsequent i.n. challenge could confer airway homing to secondary effector T cells generated from i.p.-primed precursors (Fig. 4 D), it was unclear whether this homing was maintained into the memory stage. To determine whether the defective homing of i.p.-primed memory T cells under steady-state conditions could be corrected by a subsequent i.n. infection, we used a dual adoptive transfer approach to compare the migratory efficacy of secondary memory T cells that had been generated from primary i.n.- and i.p.-primed memory CD8⁺ T cells after secondary challenge. As shown in Fig. 5 A, the relative recruitment of secondary memory CD8⁺ T cells to the airways 30 d after challenge was equal regardless of initial priming. In support of these data, cells initially primed i.p. were now able to be reactivated by residual antigen to a similar degree as i.n.-primed cells. As shown in Fig. 5 B, secondary memory CD8⁺ T cells generated from i.p.- and i.n.-primed precursors (open bars and black bars, respectively) both expressed high levels of CD69 in the MLN, and this expression was comparable to primary host memory CD8⁺ T cells (gray bars). In addition, the ability of secondary memory CD8⁺ T cells initially primed i.p. to now recognize residual antigen correlated with increased expression of CD49a and CD103 (Fig. 5 C). Notably, the route of challenge is also important for the enhanced airway recruitment of secondary memory CD8⁺ T cells, as an i.p. challenge resulted in minimal recruitment of secondary memory cells generated from both i.n.- and i.p.-primed precursors (Fig. S4). Thus, the inability of i.p.-primed memory CD8⁺ to recognize residual antigen can be corrected by a subsequent i.n. virus challenge, and this correction fully restores the steady-state recruitment of memory CD8⁺ T cells to the lung airways.

There is compelling evidence that memory T cells resident in peripheral tissues play a key role in mediating T cell–dependent protective immunity against microbial pathogens (Hansen et al., 2009; Woodland and Kohlmeier, 2009). Previous studies demonstrated that memory CD8⁺ T cell numbers in the lung airways undergo a gradual decline in the first several months after a respiratory virus infection before finally stabilizing at a relatively low number that is maintained for the life of the host. Several related avenues of investigation have shown that both antigen-dependent and -independent processes can contribute to memory T cell recruitment to the airways, leading to the hypothesis that the gradual decline of airway-resident memory T cells is a consequence of residual antigen clearance (Kohlmeier and Woodland, 2009). Although the current study supports this hypothesis, we have extended these findings by showing that residual antigen by itself is insufficient. Rather, we provide conclusive evidence that the enhanced recruitment of memory CD8⁺ T cells to the lung airways in the initial months after infection is mediated by two separate events: initial priming in the lung-draining LNs and recognition of residual antigen.

An important question raised by these studies is why i.p.-primed memory CD8⁺ T cells fail to be activated by cognate residual antigen. First, this failure is not the result of an altered distribution of i.p.-primed memory CD8⁺ T cells in the MLN (Fig. 3). Second, although it was possible that i.p. infection resulted in the generation of low-affinity memory CD8⁺ T cells, this is also an unlikely explanation because we observed the same differences between i.p.- and i.n.-primed TCR transgenic memory CD8⁺ T cells (Fig. S2 and not depicted). Third, there is no evidence that i.n.-primed memory CD8⁺ T cells prevent the access of i.p. memory CD8⁺ T cells to APCs because transient depletion of i.n.-primed memory CD8⁺ T cells failed to enhance access of i.p.-primed memory CD8⁺ T cells to residual antigen (Fig. S5). Nevertheless, despite all these similarities, only i.n.-primed memory CD8⁺ T cells were selectively activated by residual antigen in the MLN.

Overall, the current findings show that initial priming of CD8⁺ T cells in the lung-draining LNs during a respiratory infection enables these cells to access the secondary signals necessary for optimal recruitment to the lung airways under steady-state conditions. The requirement of both factors for the recruitment of memory CD8⁺ T cells to the lung airways is unique to this site, suggesting that a specific imprinting program occurs after a respiratory virus infection. Clearly, these findings have to be considered for the design of vaccines intended to generate memory CD8⁺ T cells capable of providing effective surveillance for respiratory pathogens.

Sendai virus (Enders strain) and influenza virus A/HK-x31 (x31) were grown, stored, and titered as previously described (Hikono et al., 2007). The recombinant influenza virus A/WSN/33 (WSN)-OVA_I and x31-OVA_I containing the OVA_257–264 (SIINFEKL) peptide in the neuraminidase stalk were provided by R. Dutton (Trudeau Institute, Saranac Lake, NY). Female C57BL/6J, B6.PL-Thy1a/Cy (CD90.1⁺), and B6.SJL-Ptprca Pep3/BoyJ (CD45.1⁺) mice were purchased from The Jackson Laboratory. OT-I transgenic mice (CD45.1⁺ and CD90.1⁺ background) were provided by R. Dutton. Animals were housed under specific pathogen-free conditions. Mice (6–8 wk) were infected i.p. (10⁴ EID₅₀ for Sendai, 1,333 PFU for WSN-OVA_I, and 5,667 PFU for x31-OVA_I) or i.n. (250 EID₅₀ for Sendai and 300 EID₅₀ for x31) with each virus in a variety of combinations as indicated. For i.n. infection, mice were anesthetized by i.p. injection of 2,2,2-tribromoethanol. Additionally, some groups of mice were challenged i.n. with 1 µg CpG ODN 1826 (InvivoGen). All animal studies were approved by the Trudeau Institute Animal Care and Use Committee.

For the dual adoptive transfer approach, congenic donor mice were either infected i.p. or i.n. with Sendai virus. 30 d later, CD8-enriched splenocytes were combined such that the number of Sendai NP_324-332/K^b-specific CD8⁺ T cells in each donor population was equal (7 × 10⁵ Sendai-specific cells from each donor). The combined cells were transferred intravenously into recipient mice that had received x31 virus or Sendai virus by i.n. infection 30 d before transfer. Alternatively, if the recipient mice were to receive a challenge infection, a mixture of 5 × 10³ Sendai-specific cells from each donor was transferred intravenously into naive recipient mice. 1 d later, recipient mice were challenged i.n. with Sendai virus.

Tissues were harvested and stained, as previously described (Kohlmeier et al., 2008). In brief, single cell suspensions were stained with combinations of APC- or PE-conjugated Sendai NP_324-332/K^b tetramer, APC-conjugated influenza NP_366-374/D^b tetramer, and APC-conjugated OVA_257–264/K^b tetramer. All tetramers were generated by the Trudeau Institute Molecular Biology Core. Tetramer-labeled cells were then stained with antibodies to CD8, CD27, CD43 (1B11), CD49a, CD45.1, CD44, CD62L, CD69, CD90.2, CD103, CD122, CD127, Ly6c, CXCR3, and KLRG1. Purified anti-CD49a mAb was labeled with R-Phycoerythrin (RPE) using the PhycoLink RPE Conjugation kit (ProZyme, Inc.). All anti–mouse mAbs used were purchased from BD, eBioscience, BioLegend, R&D Systems, or SouthernBiotech. Samples were run on FACSCalibur (BD) or CyAn (Dako) flow cytometers, and data were analyzed using FlowJo (Tree Star, Inc.).

The standardized number of tetramer⁺ cells in each tissue as compared with spleen were calculated as follows: [(the number of tetramer⁺ cells in each tissue/the number of tetramer⁺ cells in spleen) × 100]. Then, the relative recruitment of CD8⁺ T cells from the spleen to each tissue was calculated as follows: (the standardized number of tetramer⁺ cells derived from the experimental group/the standardized number of tetramer⁺ cells derived from the control group). Each symbol represents an individual mouse.

Naive OT-1 cells (CD90.1⁺) were sorted using a FACSVantage (BD) cell sorter with DIVA enhancement software and transferred i.v. into recipient (C57BL/6) mice. 1 d later, mice were infected i.n. with x31-OVA_I. 30 d after infection, naive OT-1 cells (CD45.1⁺) sorted as previously described were also transferred i.v. into recipient mice, and mice were then challenged i.p. with WSN-OVA_I. MLNs were harvested from recipient mice, embedded in OCT (Sakura Finetek), and frozen over liquid nitrogen. Frozen sections (6 µm) were fixed in acetone/ethanol, air dried, and stained with combinations of anti-CD45.1-FITC, anti-CD90.1-APC, anti-CD11c-PE, anti-CD3-PE, and anti-CD19-PE. All frozen sections were viewed with a microscope (Axiovert 200M; Carl Zeiss, Inc.) and images were recorded with a digital camera (AxioCam; Carl Zeiss, Inc.).

Homogenized lungs were serially diluted and added to MDCK monolayers. After incubation at 35°C for 48 h, cells were fixed with acetone and stained with a combination of goat anti-parainfluenza mAb and biotin-labeled anti–goat IgG mAb (Abcam). Specific reaction was detected using alkaline phosphatase-conjugated streptavidin (Dako) and nitroblue tetrazolium-5-bromo-4-chloro-3-indolyl phosphate (Sigma-Aldrich) as a chromogen. Foci were enumerated microscopically.

Fig. S1 shows that residual antigen is present in the MLN after i.p. infection. Fig. S2 shows that i.p.-primed memory CD8⁺ T cells fail to express CD49a and CD103 in the MLN of i.n.-primed hosts. Fig. S3 shows representative staining and cell number data from the dual transfer experiments shown in Fig. 4. Fig. S4 shows that the route of secondary challenge can reprogram memory CD8⁺ T cells regardless of the route of initial priming. Fig. S5 shows that depleting i.n.-primed memory CD8⁺ T cells does not enable i.p.-primed cells to recognize residual antigen.

The authors thank Marcy Blackman and William Reiley for critically reviewing the manuscript.

This work was supported by National Institutes of Health grants AI067967, AI076499, and AG021600 (D.L. Woodland), AI083610 (J.E. Kohlmeier), AI074922 (D.M. Jelley-Gibbs), a grant from the Uehara Memorial Foundation (S. Takamura), and funds from the Trudeau Institute.

The authors have no conflicting financial interests.