Antigen-independent differentiation and maintenance of effector-like resident memory T cells in tissues

Abstract

Differentiation and maintenance of recirculating effector memory CD8 T cells (T(EM)) depends on prolonged cognate Ag stimulation. Whether similar pathways of differentiation exist for recently identified tissue-resident effector memory T cells (T(RM)), which contribute to rapid local protection upon pathogen re-exposure, is unknown. Memory CD8αβ(+) T cells within small intestine epithelium are well-characterized examples of T(RM), and they maintain a long-lived effector-like phenotype that is highly suggestive of persistent Ag stimulation. This study sought to define the sources and requirements for prolonged Ag stimulation in programming this differentiation state, including local stimulation via cognate or cross-reactive Ags derived from pathogens, microbial flora, or dietary proteins. Contrary to expectations, we found that prolonged cognate Ag stimulation was dispensable for intestinal T(RM) ontogeny. In fact, chronic antigenic stimulation skewed differentiation away from the canonical intestinal T cell phenotype. Resident memory signatures, CD69 and CD103, were expressed in many nonlymphoid tissues including intestine, stomach, kidney, reproductive tract, pancreas, brain, heart, and salivary gland and could be driven by cytokines. Moreover, TGF-β-driven CD103 expression was required for T(RM) maintenance within intestinal epithelium in vivo. Thus, induction and maintenance of long-lived effector-like intestinal T(RM) differed from classic models of T(EM) ontogeny and were programmed through a novel location-dependent pathway that was required for the persistence of local immunological memory.

Figure 1. Virus-specific CD8αβ T cells adopt a recently stimulated phenotype after migration to SI epithelium

(A) Phenotype of gp33-specific Thy1.1+ P14 CD8 T cells in spleen and SI IEL on the indicated days after LCMV Armstrong infection. All plots are gated on Thy1.1+ lymphocytes. (B) Kinetics of CD62L expression among P14 in spleen and SI IEL after LCMV infection. ***p<0.001, Student's unpaired t test. Error bars indicated SEM. n=5 per time point. (C) Phenotype of gp33-specific P14 CD8 T cells (top row is spleen, bottom row is SI IEL) 37 days after LCMV Armstrong infection. Phenotypes are representative of > 10 mice. (D&E) Mice were infected with the Armstrong or Cl-13 strains of LCMV, and 35 days later, the phenotype of H-2Db/gp33 MHC I tetramer+ splenocytes was compared. (D) Gated on lymphocytes, and (E) gated on CD8+ H-2Db/gp33 MHC I tetramer+ lymphocytes.

Figure 2. Induction and maintenance of activated phenotype is Ag-independent

(A) Naïve CD45.1+ OT-I × RAG−/− CD8 T cells were transferred to RAG−/− C57Bl/6 mice. 60 days later, OT-I were isolated from spleen and SI IEL and analyzed for the indicated markers (plots gated on CD45.1+ lymphocytes). One representative of 5 mice, experiment was performed 3 times. (B) OT-I were transferred to conventionally housed specific pathogen free (SPF) or germ-free RAG−/− mice. 27 days later, spleen and SI IEL were harvested and analyzed for the indicated markers (plots gated on H-2K^b/SIINFEKL tetramer+ lymphocytes). One representative of 5 mice. (C) PD-1 expression among transferred P14 isolated from spleen and SI IEL 2 months after LCMV Armstrong infection. One representative of 3 mice per experiment, experiment was performed twice. (D) Polyclonal CD8αβ+ lymphocytes were isolated from spleen and SI IEL of nur77-GFP transgenic mice or C57Bl/6 mice were examined for GFP and CD69 expression +/− systemic treatment with anti-CD3 Ab (Plots gated on CD8α+/CD8β+ lymphocytes). One representative of 3 mice per experiment, experiment was performed twice. (A-D) Spleen = solid gray, SI IEL = black line.

Figure 3. Persistent antigen alters canonical IEL phenotype

(A) C57Bl/6J mice were infected with the Armstrong (cleared) or Cl-13 (chronic) strains of LCMV. 21 days later, lymphocytes were isolated from spleen and SI IEL (plots gated on CD8α+ lymphocytes). Percent of CD103+ cells among H-2D^b/gp33 tetramer+ (upper right) or tetramer- (lower right) is indicated. (B) Mean percent of H-2D^b/gp33 tetramer+ IEL that were CD103+, n=5. ***p<0.001, Student's unpaired t test. Error bars indicated SEM. (C) Naïve OT-I were transferred to ovalbumin bearing 232–4 mice and 150 days later isolated SI IEL were examined for expression of CD103 (black histogram gated on OT-I, gray histogram gated on CD8α+ lymphocytes). All plots are representative of at least three mice per experiment and each experiment was performed at least twice.

Figure 4. Cytokine milieu can induce canonical IEL phenotype

(A) C57Bl/6J mice were infected with LCMV Armstrong one day after transfer of P14. 4.5 or 30 days after LCMV infection, splenocytes were isolated, cultured with 0.1ng/ml TGFb for 60h, and examined for CD103 expression by flow cytometry. (B) D4.5 splenocytes (as in A) were treated with 0.1ng/ml TGFb for 60h +/− gp33 cognate peptide. (C&D) D4.5 splenocytes were cultured for 40h with the indicated cytokines, then examined for the expression of CD103, Ly6C, and CD69 by flow cytometry. All plots are representative of at least three mice per experiment and each experiment was performed at least three times.

Figure 5. TGFβ induces CD103+/CD69+ SI IEL population in vivo

(A–B) Thy1.1+ OT-I and Thy1.1/Thy1.2 dnTGFbRII OT-I were mixed in a 1:1 ratio and co-transferred to C57Bl/6J mice. The following day, recipients were infected with VSVova, and lymphocytes were isolated from spleen and SI IEL 7 days later, and analyzed for CD103 and CD69 expression by flow cytometry. Plots gated on WT OT-I (black) or dnTGFbRII OT-I, as indicated. (C) Number of recovered WT (black) and dnTGFβRII OT-I (gray) SI IEL among the indicated subsets. (D) Identical experimental design as in A–C, however WT and mutant P14 were substituted for OT-I and mice were infected with LCMV Armstrong. **p<0.01, Student's unpaired t test. Error bars indicated SEM. Flow plots are representative of five mice per experiment. Each experiment was performed at least three times.

Figure 6. TGFβ and CD103 expression is required for maintenance of IEL memory

(A) C57Bl/6J mice were infected with LCMV Armstrong one day after transfer of a 1:1 ratio of WT P14 and CD103−/− P14. 4.5, 7, and 30 days later, the ratio of each population was assessed in the indicated tissues. (B–F) C57Bl/6J mice that received either WT Thy1.1+ P14 or CD103−/− Thy1.1+ P14 and infected with LCMV Armstrong were analyzed by immunohistochemistry 30 days later. (B) The indicated region of frozen small intestine `pinwheels' were assessed for SI IEL (C) and SI LPL (D, representative images, DAPI indicates nuclei, Thy1.1 indicates P14). (E) Number of WT or CD103−/− P14 counted in epithelium or lamina propria per 10⁶ nucleated cells (approximately 1.5×10⁵ nuclei were counted per mouse within the indicated region). n=3. (F) The ratio of SI IEL vs. LPL P14 that were counted in B–E. ***p<0.001, *p<0.05, Student's unpaired t test. Error bars indicated SEM.

Figure 7. Signature IEL markers are expressed in other non-lymphoid tissues

(A–B) C57Bl/6J mice were infected with LCMV Armstrong one day after receiving naïve Thy1.1+ P14. Three months later, lymphocytes were isolated from the indicated tissues and analyzed by flow cytometry for expression of CD103, (A) granzyme B, or (B) CD69 and CD62L. Plots gated on CD8+ Thy1.1+ lymphocytes. (C) Naïve CD45.1+ OT-I × RAG−/− CD8 T cells were transferred to RAG−/− C57Bl/6 mice. 60 days later, OT-I were isolated from the indicated tissues and analyzed for CD103, CD69, and CD62L expression. Plots gated on CD8+ CD45.1+ lymphocytes.