T-Cell Activation by Cross-Presentation in Lymphoid Tissues Influences CTL Activity.

To examine the contributions of the direct presentation and cross-presentation pathways to CTL development, we transferred OT-I T cells into B6, bm1, or Alb-K^b.bm1 mice, and treated them 1 d later with rAAV.mOVA, with coinjection of rAAV.K^b for bm1 mice. CTL function was assessed 1 wk after OT-I transfer by in vivo cytotoxicity assay using SIINFEKL-pulsed B6 splenocytes as targets. CTL activity was detected only in B6 mice, in which cross-presentation was present (Fig. 1 D–F). The lack of CTL activity in bm1 recipients treated with rAAV.K^b (Fig. 1E) and Alb-K^b.bm1 (Fig. 1F) was not caused by a failure of OT-I T cells to proliferate to the same extent as in B6 mice, as the number of OT-I T cells recovered from these strains after rAAV.mOVA treatment were similar to those detected in B6 mice (Fig. S3).

To investigate the response of CD8 T cells with different specificity, we tested the outcome of CD8 Des T-cell activation following adoptive transfer into rAAV.K^b-treated syngeneic B10.BR (H-2^k) recipients. The Des TCR is specific for H-2K^b complexed with endogenous peptide. Hence, in rAAV.K^b-treated B10.BR recipients, Des T cells are predicted to recognize H-2K^b expressed by hepatocytes, but not epitopes derived from the H-2K^b protein presented in the context of recipient H-2^k molecules. This model therefore essentially excludes the contribution of recipient cross-presentation to Des T-cell activation (13, 16, 17). Consistent with our previous findings in Alb-K^b transgenic mice that express H-2K^b only on hepatocytes (13, 16, 17), Des T-cell activation at 3 h after transfer into rAAV.K^b-treated mice was restricted to the liver (Fig. 1G). Despite efficient intrahepatic activation leading to proliferation of transferred Des T cells (Fig. S4), CTL function never developed in vivo (Fig. 1H). Consistent with our previous findings (13), these results also suggested a potential requirement for antigen presentation in secondary lymphoid organs for effective CTL formation.

CTLs Generated in the Presence of Extrahepatic Cross-Presentation Were Silenced by Persistent Intrahepatic Antigen Expression.

Although OVA-specific cytotoxicity was consistently detected in rAAV.mOVA-treated B6 mice at 1 wk after OT-I transfer, this response was not as effective as that seen after OT-I stimulation with OVA-coated splenocytes and LPS (Fig. 1D, black bar indicates positive control). Extension of in vivo cytotoxicity measurements to week 3 after OT-I transfer revealed no detectable residual OT-I CTL activity in rAAV.mOVA-treated B6 mice (Fig. 2A). This was not caused by T-cell deletion, as OT-I cells were still present and were sequestered in the livers of recipient mice (Fig. 2B), which continued to express high levels of OVA (Fig. 2C). Restimulation of these mice by i.v. transfer of OVA-coated splenocytes and LPS led to an increase in OT-I T cell numbers but did not restore cytotoxic function (Fig. 2B), suggesting that intrahepatic T cells retained their ability to recognize antigen and proliferate but were functionally silenced by extrinsic or intrinsic mechanisms. A second cohort of OT-I T cells transferred into these mice proliferated to the same extent as those transferred into rAAV.mOVA-treated controls that had not received the first cohort of OT-I T cells (Fig. 2D), confirming antigen availability at this time point. Together, these results indicate that, although CTLs were generated via cross-presentation in secondary lymphoid tissues, they were subsequently “silenced” at the functional level between 1 and 3 wk postactivation.

Open in a separate window

Fig. 2.

Cytotoxic OT-I T cells were silenced at 3 wk by high but not low levels of intrahepatic antigen. (A) In vivo cytotoxicity in the spleen at 1 wk and 3 wk after transferring 10⁶ OT-I LN cells into B6 mice treated with rAAV.mOVA (5 × 10¹⁰ vgc) the following day. (B) Ability of 10⁷ OVA-coated splenocytes and 1 µg LPS (OVA spl LPS) delivered i.v. to restore proliferation (Left) or cytotoxicity (Right) of OT-I T cells transferred into B6 mice and treated with rAAV.mOVA (5 × 10¹⁰ vgc) 6 wk earlier. Total T-cell numbers (in spleen, liver, LNs, and blood) and cytotoxicity in the spleen were measured 7 d after restimulation. Bars show mean ± SD measured in three mice per group. (C) OVA expression in liver 3 wk after transferring 10⁶ OT-I LN cells into B6 mice treated with rAAV.mOVA (5 × 10¹⁰ vgc) the following day. Controls did not receive OT-I cells or were treated with rAAV.K^b (5 × 10¹⁰ vgc). Images are representative of livers from three mice per group. (D) OT-I T cells were not able to clear OVA-expressing hepatocytes in mice treated with a high dose of rAAV.mOVA. B6 mice were adoptively transferred with a first cohort of 10⁶ CD45.2⁺ OT-I LN cells or PBS solution and then, 1 d later, treated with rAAV.mOVA (5 × 10¹⁰ vgc). After 3 wk, a second cohort of 2 × 10⁶ CFSE-labeled CD45.1⁺ OT-I cells was adoptively transferred into the same recipient mice. Proliferation of the second cell cohort was assessed by gating on liver CD8⁺ CD45.1⁺ 2 d after transfer and assessing CFSE dilution. Flow plots are representative of five or seven mice per group from two independent experiments. (E) Expression of OVA in the liver 7 d after high- and low-dose rAAV.mOVA treatment of B6 mice. OVA expression was detected by immunofluorescence as per C. (F) OT-I cytotoxicity in the spleen at week 1 and week 3 after transfer of 10⁶ OT-I LN cells into B6 mice treated with high- or low-dose rAAV.mOVA as indicated. (G) Efficient clearance of OVA-expressing hepatocytes in B6 mice by week 3 after transfer of 10⁶ OT-I LN cells and low-dose inoculation with rAAV.mOVA (5 × 10⁸ vgc) compared with mice that did not receive OT-I T cells. (H) OT-I T cells cleared detectable expression of OVA by hepatocytes in mice treated with a low dose of rAAV.mOVA. B6 mice were adoptively transferred with a first cohort of 10⁶ CD45.2⁺ OT-I LN cells or PBS solution and then, 1 d later, treated with rAAV.mOVA (5 × 10⁸ vgc). After 3 wk, a second cohort of 2 × 10⁶ CFSE-labeled CD45.1⁺ OT-I cells was adoptively transferred into the same recipient mice. Proliferation of the second cell cohort was assessed in the liver by gating on CD8⁺ CD45.1⁺ lymphocytes 2 d after transfer and assessing CFSE dilution. Proliferation of the second OT-I cohort was only observed in controls that did not receive the first cohort of OT-I T cells. Flow plots are representative of at least seven mice per group in two independent experiments.

Lower Frequencies of Antigen-Expressing Hepatocytes Led to Development of CTLs That Persistently Maintained Function.

The induction of antigen expression on almost all hepatocytes after a standard dose of rAAV treatment is akin to that observed in transgenic mice in which antigen is continually expressed on hepatocytes, and to liver transplantation, in which all liver cells express donor antigen. However, it is unlikely that these levels of antigen expression are achieved early in infections by hepatotropic viruses or by current gene therapy protocols. To investigate the effects of lower levels of antigen expression, we reduced the dose of nonreplicative rAAV vector administered. B6 mice were treated with 1/100th of the standard dose of rAAV.mOVA (5 × 10⁸ vgc), which resulted in OVA expression by a small proportion of hepatocytes (Fig. 2E). Following administration of low-dose rAAV.mOVA, OT-I also developed CTL function at week 1 in B6 mice (Fig. 2F). However, in contrast to the “silencing” of OT-I function observed after high-dose rAAV treatment, CTL activity in the low-dose–treated group persisted at week 3 (Fig. 2F). In these mice, no detectable OVA expression was found in the liver 3 wk following treatment (Fig. 2G). The complete clearance of residual OVA-expressing cells was confirmed by the failure of a second cohort of adoptively transferred OT-I cells to be activated and proliferate in these recipients (Fig. 2H). Thus, in B6 mice, low numbers of OVA-expressing hepatocytes were sufficient to trigger OT-I CTL that mediated clearance of antigen expression, and this was associated with persisting CTL activity at week 3.

A Threshold in the Number of Antigen-Expressing Hepatocytes Determines CD8 T-Cell Outcome.

These data indicated that, in B6 mice, the number of hepatocytes transduced was a key determinant of CD8 T-cell fate: high intrahepatic antigen load was associated with silencing of CTL. To determine the level of initial hepatocyte transduction required for CTL silencing and to explore the effects of antigen expression levels more akin to those occurring during hepatotropic viral infections (4), we assessed OT-I CTL activity in B6 mice treated with intermediate doses of rAAV 3 wk earlier. To facilitate quantification of antigen-expressing hepatocytes, we generated a new rAAV vector that expressed GFP bicistronically with mOVA (rAAV.GFP-mOVA). Administration of a standard dose of this new vector (5 × 10¹¹ vgc) also led to 100% transduction of hepatocytes, indicating comparable efficacy of this vector with rAAV.mOVA (Fig. 3A). Similar to rAAV.mOVA-treated mice, CTL activity was not observed in mice treated with the standard (high) dose of rAAV.GFP-mOVA, but was detected in mice treated with lower doses of rAAV. CTL function was maintained only when less than 25% of hepatocytes were transduced, indicating that a threshold existed for initial transduction levels required for CTL silencing (Fig. 3B).

Open in a separate window

Fig. 3.

A threshold of intrahepatic antigen expression determines the functional outcome of CD8 T-cell responses. (A) GFP expression in B6 mice treated with PBS solution or 10¹¹ vgc of rAAV.GFP-mOVA for 7 d. Deconvolution micrographs of fixed liver sections stained with DAPI (blue) and phalloidin-Alexa Fluor 594 (red) with GFP expression in green show 100% transduction efficiency. (B) In vivo cytotoxicity 3 wk after transfer of 10⁶ OT-I LN cells into B6 mice treated with a range of rAAV.GFP-mOVA doses, plotted against proportion of GFP⁺ hepatocytes in mice not injected with T cells.

Low Frequencies of Antigen-Expressing Hepatocytes Could Promote CTL Development in the Absence of Cross-Presentation, Depending on T-Cell Specificity.

Based on our findings that cross-presentation was associated with CTL generation in B6 mice treated with high-dose rAAV.mOVA (Fig. 1), we hypothesized that exclusion of cross-presentation would also prevent CTL development after low-dose rAAV.mOVA treatment. To test this, OT-I T cells were transferred into Alb-K^b.bm1 mice treated with a low dose (5 × 10⁸ vgc) of rAAV.mOVA, or into bm1 mice coadministered high-dose rAAV.K^b and low-dose rAAV.mOVA. Surprisingly, we found high levels of CTL activity at 1 and 3 wk after OT-I transfer in both recipient strains despite the exclusion of cross-presentation (Fig. 4). CTL development was not mediated by residual cross-presenting APCs cotransferred with OT-I T cells, as robust CTL responses were also found when donor cells from OT-I mice backcrossed to bm1 were used (Fig. S5). These results showed that, after low-dose rAAV treatment, direct activation of OT-I T cells exclusively by OVA-expressing hepatocytes was able to prime CTLs by week 1 and that these CTLs were not silenced at week 3.

Open in a separate window

Fig. 4.

Exclusion of cross-presentation at low antigen doses did not prevent OT-I T cells developing into CTL after stimulation with high-affinity WT OVA. In vivo cytotoxicity at week 1 and week 3 following adoptive transfer of 10⁶ OT-I T cells into B6, Alb-K^b.bm1, or bm1 mice treated with high-dose (5 × 10¹⁰ vgc) or low-dose (5 × 10⁸ vgc) rAAV.mOVA and/or rAAV.K^b the following day, as indicated. OVA spl LPS, B6 mice injected with OVA-coated B6 splenocytes and 1 µg LPS 7 d earlier.

It was unexpected that hepatocytes could induce OT-I development into cytotoxic effectors by direct-presentation, as we have shown consistently in the past that naive CD8 Des T cells exclusively activated by hepatocytes developed poor CTL function and/or died prematurely, resulting in undetectable CTL activity in vivo (18). However, the outcome of intrahepatic Des T-cell activation had never been assessed in the setting of low H-2K^b antigen expression. We therefore measured Des CTL function in B10.BR mice treated with low-dose rAAV.K^b (5 × 10⁸ vgc). Des T cells underwent activation and proliferation in these animals (14), but were unable to kill H-2K^b–expressing target cells in vivo (Fig. 5A). In these low-dose rAAV.K^b-treated mice that also received Des T cells, H-2K^b–expressing hepatocytes were still present at week 3, as demonstrated by immunofluorescent staining (Fig. 5B) and by their ability to stimulate proliferation of a second cohort of adoptively transferred naive Des T cells (Fig. 5C). These results suggest that, although Des T cells proliferated in response to low frequencies of H-2K^b-expressing hepatocytes, they failed to develop sufficient CTL function to clear antigen-expressed hepatocytes. Therefore, in contrast to OT-I T cells, Des T cells exclusively activated by hepatocytes did not develop into CTL at low antigen doses.

Open in a separate window

Fig. 5.

Lack of Des CTL development at low antigen dose was not a result of low pMHC occupancy. (A) In vivo cytotoxicity specific to H-2K^b+ target splenocytes at week 1 and week 3 after adoptive transfer of 1.5 × 10⁶ DesRAG^−/− LN cells into B10.BR mice treated with high- (5 × 10¹⁰ vgc) or low-dose (5 × 10⁸ vgc) rAAV.K^b the following day. (B) Persisting H-2K^b expression in B10.BR mice treated with high- and low-dose rAAV.K^b, whether or not they received 10⁶ DesRAG^−/− LN cells the day before rAAV treatment. (C) Des T cells were unable to clear H-2K^b+ hepatocytes after 3 wk in mice treated with rAAV-K^b. B10.BR mice were adoptively transferred with a first cohort of 10⁶ DesRAG^−/− LN cells or PBS solution and then, 1 d later, treated with rAAV.K^b (5 × 10¹⁰ or 5 × 10⁸ vgc). After 3 wk, a second cohort of 2 × 10⁶ CFSE-labeled DesRAG^−/− cells was adoptively transferred into the same recipient mice. Proliferation of the second cell cohort was assessed by gating on CD8⁺ CFSE⁺ cells 2 d after transfer and assessing CFSE dilution. (D) Total numbers of Des T cells and numbers of recently activated CD69^hi Des T cells were quantified in the liver at 3 h after 10⁶ DesRAG^−/− LN cells were injected into B10.BR or Alb-K^b mice treated with rAAV.GFP-KVIT 7 d earlier. Bar graphs show mean ± SD of three mice. (E) In vivo cytotoxicity measured at weeks 1 and 3 after transfer of 10⁶ DesRAG^−/− LN cells into B10.BR mice treated with indicated doses of rAAV.K^b and/or rAAV.GFP-KVIT the following day. Met-K^b mice receiving 1.5 × 10⁶ DesRAG^−/− LN cells 5 d before cytotoxicity assay were used as positive controls. Bar graphs show mean percent ± SD of specific killing measured in three mice per group for two independent experiments.

Increasing Peptide:MHC Complex Occupancy Did Not Alter the Outcome of Des T-Cell Activation by Hepatocytes.

One potential explanation for the different functional outcomes observed after Des T-cell and OT-I T-cell stimulation in the liver was the amount of antigen expressed in each hepatocyte, as this would affect the number of peptide:MHC complexes (pMHC) available for TCR recognition and hence T-cell activation. It has been reported that the self-peptides recognized by the Des TCR in the context of H-2K^b are present only at low levels, estimated at 200 copies per cell, corresponding to ~0.3% H-2K^b occupancy (19). In contrast, H-2K^b occupancy by SIINFEKL can range from 10% to 80% (corresponding to 3,000–85,000 H-2K^b:SIINFEKL complexes per cell) depending on the form and amount of OVA expressed (20). To increase the number of H-2K^b:self-peptide complexes available to Des T cells, we replaced SIINFEKL in our expression construct with KVITFIDL, one of three epitopes that associate with H-2K^b for Des TCR recognition. Treatment of Alb-K^b mice with this vector (rAAV.GFP-KVIT) increased Des T-cell activation and retention in the liver compared with untreated controls (Fig. 5D), suggesting that treatment with rAAV.GFP-KVIT was effective in increasing the numbers of H-2K^b:KVITFIDL complexes. However, in contrast to the outcome of OT-I responses following administration of low-dose OVA, in which persisting CTL responses were observed, we did not detect Des CTL generation in B10.BR mice coadministered low-dose rAAV.K^b with high-dose rAAV.GFP-KVIT (Fig. 5E), suggesting that differences in MHC occupancy did not explain the different fates of OT-I and Des T cells.

TCR:pMHC Affinity Influences the Outcome of T-Cell Activation in Response to Liver-Expressed Antigen.

Another potential explanation for the different functional outcomes observed after Des and OT-I T-cell stimulation by hepatocytes was the differing TCR affinities for their respective ligands. Low-affinity TCR:pMHC interactions have been reported to be less effective at activating CD8 T cells and supporting effector expansion (21, 22). It is possible that the Des TCR affinity to H-2K^b complexed to one or all of the three defined self-peptides was lower than the OT-I TCR affinity to H-2K^b:SIINFEKL. As Des and OT-I TCR affinities for their ligands have not been compared, we expressed a lower-affinity ligand for OT-I and tested if its expression in hepatocytes would result in failure to elicit CTL function. We replaced the fourth asparagine residue of the WT SIINFEKL (OVA) with a threonine to generate SIITFEKL (OVA-T4) to create the rAAV.GFP-mOVA-T4 vector. OVA and OVA-T4 are reported to demonstrate similar binding efficiencies to H-2K^b, but the H-2K^b:OVA-T4 complex binds with lower affinity to the OT-I TCR compared with H-2K^b:OVA (22). To assess the effect of varying pMHC:TCR affinity on CD8 T-cell priming by hepatocytes, we cocultured naive OT-I T cells with hepatocytes isolated from B6 mice that had been transduced in vivo to express OVA or OVA-T4. Low numbers of hepatocytes expressing WT OVA acted as efficient APCs to induce activation and proliferation of most OT-I T cells (Fig. S6). A significant proportion of OT-I T cells stimulated by WT OVA-expressing hepatocytes also expressed IFN-γ, suggesting that they were functional. In contrast, hepatocytes expressing the low-affinity OVA-T4 variant induced poor OT-I activation, proliferation, and IFN-γ production (Fig. S6), akin to that previously described for Des T-cell responses (13). The suboptimal OT-I response to OVA-T4-expressing hepatocytes was reversed by addition of exogenous IL-2, suggesting that presentation of a high-affinity antigen by hepatocytes could remove the requirement for additional cosignals to mediate effective priming, which was required during presentation of lower-affinity antigen.

To test the ability of hepatocytes expressing low- or high-affinity antigen to induce CTLs in vivo, OT-I T cells were transferred into B6 mice treated with rAAV.GFP-mOVA or rAAV.GFP-mOVA-T4 vectors. Both vectors had comparable transduction efficacies (Fig. 6A and Fig. S7), and their administration to mice led to activation and proliferation of most transferred OT-I T cells in the liver and lymphoid organs (Fig. 6B). However, when CTL activity was measured at week 3, only expression of the high-affinity WT OVA, but not the lower-affinity OVA-T4 variant, led to CTL generation after low-dose rAAV vector treatment (Figs. 4 and ​and6C).6C). Higher rAAV doses did not elicit CTL activity at 3 wk for either variant (Fig. 6C).

Open in a separate window

Fig. 6.

A low-affinity OVA variant expressed in hepatocytes following rAAV transduction did not lead to detectable CTL function at 3 wk despite promoting efficient OT-I recruitment and proliferation. (A) GFP expression in B6 mice treated 7 d prior with PBS solution or 1 × 10¹¹ vgc of rAAV.GFP-mOVA or rAAV.GFP-mOVA-T4. Deconvolution micrographs of fixed liver sections stained with DAPI (blue) and phalloidin-Alexa Fluor 594 (red) with GFP expression in green. (B) Expression of CD69 on CD45.1⁺CD8⁺ OT-I T cells at 3 h after transfer of 5 × 10⁶ CD45.1⁺ OT-I LN cells into B6 mice pretreated with 1 × 10¹¹ vgc rAAV.GFP-mOVA or rAAV.mOVA-T4 (black lines) compared with untreated B6 controls (shaded; histograms; Upper), and proliferation of CD45.1⁺ donor cells 36 h after transfer (dot plots; Lower). (C) In vivo cytotoxicity assay at week 3 after transfer of 10⁷ CD45.1⁺ OT-I LN cells into B6 mice treated with indicated doses of rAAV.GFP-mOVA or rAAV.GFP-mOVA-T4 1 d after T-cell transfer.

Together, these results suggest that, when a low percentage of hepatocytes expressed antigen, CD8 T cells activated by higher-affinity TCR:pMHC interactions could develop effector function whereas lower affinity interactions, although still able to stimulate proliferation, were insufficient to induce full functional differentiation in the absence of cosignals that are not provided by hepatocytes.

High Levels of Antigen Expression in the Liver Promote T-Cell Exhaustion.

To characterize the mechanisms involved in CTL silencing at high antigen dose, we phenotyped intrahepatic OT-I T cells isolated from mice treated with high or low doses of rAAV.mOVA at week 3 after adoptive transfer. OT-I T cells from the liver of mice treated with a high dose of rAAV.mOVA expressed very high levels of the PD-1 inhibitory receptor, a hallmark of exhausted T cells (Fig. 7 A and B). A significant proportion of OT-I T cells from liver also expressed Tim-3, but there were no significant differences in Tim-3 expression between OT-I T cells from the high- and low-dose treatment groups (Fig. 7 A and B). T-cell exhaustion was confirmed at the functional level. OT-I T cells from the livers of mice treated with high-dose rAAV.mOVA did not exhibit degranulation of cytotoxic granules or express IFN-γ upon ex vivo restimulation (Fig. 7 C and D). In contrast, most OT-I T cells isolated from the low-dose treatment group expressed low to intermediate levels of PD-1 (Fig. 7 A and B), degranulated efficiently, and coexpressed IFN-γ upon ex vivo restimulation (Fig. 7 C and D). There were significantly higher numbers of intrahepatic OT-I T cells that degranulated and expressed IFN-γ in mice treated with low-dose rAAV compared with mice treated with high-dose rAAV (Fig. 7D). These findings were consistent with the results of in vivo CTL assays. In OT-I T cells isolated from the livers of mice treated with a low rAAV.mOVA dose, but not mice treated with a high dose of rAAV.mOVA, quantitative real-time–PCR confirmed expression of mRNA for IFN-γ and also demonstrated expression of TNF-α mRNA upon peptide restimulation (Fig. S8). However, no IL-2, IL-4, IL-17, or IL-10 mRNA expression was detected in OT-I T cells from either group of mice, suggesting that OT-I T cells did not differentiate into Tc2-, Tc17-, or IL-10–producing regulatory CD8 T cells.

Open in a separate window

Fig. 7.

High-dose rAAV treatment promoted and maintained a functionally exhausted phenotype. (A and B) B6 mice adoptively transferred with 3 × 10⁶ CD45.1⁺ OT-I LN cells were treated 1 d later with a low dose (5 × 10⁸ vgc) or high dose (5 × 10¹⁰ vgc) of rAAV.mOVA. Liver leukocytes were harvested 3 wk later to assess cell surface expression of PD-1 and Tim-3. Flow plots were gated on CD8⁺ CD45.1⁺ cells and are representative of seven mice per group in two independent experiments. (B) Histograms represent mean ± SEM of four mice per group. Data representative of two separate experiments. (C and D) Cytotoxic degranulation (CD107 binding) and expression of IFN-γ upon ex vivo peptide restimulation following the experimental protocol outlined for A and B. Flow plots are representative of seven mice per group in two independent experiments. (D) Histograms represent mean ± SEM of four mice per group. This experiment was repeated once (n = 3 mice per group). (E–G) B6 mice adoptively transferred with 3 × 10⁶ CD45.1⁺ OT-I LN cells were treated 1 d later with a low dose (5 × 10⁸ vgc) or high dose (5 × 10¹⁰ vgc) of rAAV.mOVA. Liver leukocytes were harvested 1 wk later and transferred into secondary B6 recipients that had been injected 1 wk earlier with a low dose (5 × 10⁸ vgc) or high dose (5 × 10¹⁰ vgc) of rAAV.mOVA. Three weeks after secondary T-cell transfer, livers were harvested to assess (F and G) expression of PD-1 and Tim-3 and (E) cytolytic function (CD107 binding) and IFN-γ production of intrahepatic OT-I T cells. Dot plots were gated on CD8⁺ CD45.1⁺ cells and are representative of three mice per group. (E) Histograms represent mean ± SEM of three mice per group. Data are representative of two independent experiments.

We also observed similar outcomes in terms of IFN-γ production and cytotoxic degranulation when low numbers (n = 500) of CD8 OT-I T cells were transferred (Fig. S9), suggesting that the influence of rAAV dose on CD8 T-cell outcome was not caused by the high precursor frequency of OT-I T cells used in this study, but is likely to affect outcomes at more physiological precursor frequencies of antigen-specific T cells.

Animals.

C57BL/6 (H-2^b) mice were purchased from the Animal Resources Centre (ARC). B10.BR (H-2^k, CD45.1), Des, Des backcrossed into the recombinase activating gene-deficient background (DesRAG^−/−) Alb-K^b, and Met-K^b mice have been described (13, 17). OT-I (41) and H-2K^bm1 (15) mice were kindly provided by William Heath (University of Melbourne, Melbourne). Alb-K^b and OT-I mice were backcrossed to H-2K^bm1. OT-I mice were also backcrossed to B6.SJL-Ptprc^a Pep3^b/BoyJ mice (ARC). Mice were maintained under specific pathogen-free conditions in the Centenary Institute (Sydney). Experimental protocols were approved by the Sydney University Animal Care and Ethics Committee and Sydney Local Health District Animal Ethics Committee.

Quantifying Hepatocyte Antigen Expression by Flow Cytometry.

Hepatocytes were isolated as previously described (42). Hepatocytes were stained with biotinylated anti–H-2K^b (AF6-88.5; BD Pharmingen), anti–H-2K^k (36-7-5; BD), and streptavidin-phycoerythrin (PE) (Invitrogen); or with rabbit anti-OVA (clone 1670; provided by Andrew Lew, Walter and Eliza Hall Institute of Medical Research, Melbourne) and Alexa Fluor 488-conjugated anti-rabbit IgG (Invitrogen). Cells were stained with propidium iodide (3 μg/mL) before acquisition on a BD LSR II flow cytometer (BS Biosciences), and were analyzed in FlowJo 9.4.11 (Tree Star).

Confocal and Epifluorescence Analysis of Liver Sections.

Livers were perfused in situ with PBS solution followed by 10% (vol/vol) neutral buffered formalin and transferred to 10% (wt/vol) sucrose followed by 30% (wt/vol) sucrose solutions before embedding in optimal cutting temperature (OCT) compound (TissueTek). Sections (12 μm) were stained with rabbit anti-OVA (clone 1670), rat-anti-F4/80 (CI:A3-1; hybridoma) or rat anti-CD31 (MEC 13.3; BD Pharmingen) and Alexa Fluor 488 or Alexa Fluor 594-conjugated secondary antibodies (Invitrogen) before image acquisition on a DeltaVision microscopy imaging system with a 10× objective and Photometrics CoolSnap HQ2 camera or a Leica SP5 Confocal microscope (Leica Microsystems).

Purification, Adoptive Transfer, and Analysis of Donor Cells by Flow Cytometry.

Single-cell suspensions of peripheral LN cells from donor OT-I, Des, or DesRAG^−/− mice were isolated as previously described (17, 25) and labeled with 5 μM carboxyfluorescein succinimidyl ester (CFSE) where indicated. One million to eight million LN cells were injected i.v. 7 d after rAAV treatment (for assays to determine primary site of activation) or 1 d before rAAV treatment (for in vivo cytotoxicity assays). Liver, spleen, LNs, and blood were harvested and stained as described previously (7). Antibodies were from BD Pharmingen unless otherwise stated, and included CD45.1 Horizon V450 (A20); CD8-Pacific Blue or CD8-allophycocyanin (53-6.7); CD69-PE, CD69-allophycocyanin, or CD69-PerCP-Cy5.5(H1.2F3); and CD44-allophycocyanin-Cy7 (IM7; Biolegend). Cells were stained with 0.1 μg/mL DAPI (Invitrogen) before acquisition on a BD LSR Fortessa flow cytometer (BD Biosciences), and were analyzed in FlowJo 9.4.11 (Tree Star).

In Vivo Cytotoxicity Assay.

Cytotoxicity of transferred Des T cells toward H-2K^b+ target splenocytes from 178.3 mice relative to control B10.BR splenocytes was measured as described previously (17). OT-I cytotoxicity was determined by their ability to kill SIINFEKL (2 μg/mL)-pulsed B6 splenocytes relative to unpulsed splenocytes. Target and control splenocytes were labeled with 5 μM or 0.5 μM CFSE and mixed 1:1, and 2 × 10⁷ total cells were injected 16 h before harvesting; the percentage of specific killing was calculated as described previously (17).

The authors thank the Centenary Institute Animal Facility and Advanced Cytometry Facility for their technical support, Dr. Frank Carbone for the mOVA cDNA, and Dr. Andrew Lew for the anti-OVA antibody used in this study. This work was supported by the National Health and Medical Research Council (NHMRC) Australia Program Grant 571408 and NHMRC Senior Research Fellowship 511903 (to P.B.). J.R.G. holds the Wenkart Chair of the Endothelium.