Survival of tissue-resident memory T cells requires exogenous lipid uptake and metabolism

Abstract

Tissue-resident memory T (T_RM) cells persist indefinitely in epithelial barrier tissues and protect the host against pathogens. However, the biological pathways that enable the long-term survival of T_RM cells are obscure. Here we show that mouse CD8⁺ T_RM cells generated by viral infection of the skin differentially express high levels of several molecules that mediate lipid uptake and intracellular transport, including fatty-acid-binding proteins 4 and 5 (FABP4 and FABP5). We further show that T-cell-specific deficiency of Fabp4 and Fabp5 (Fabp4/Fabp5) impairs exogenous free fatty acid (FFA) uptake by CD8⁺ T_RM cells and greatly reduces their long-term survival in vivo, while having no effect on the survival of central memory T (T_CM) cells in lymph nodes. In vitro, CD8⁺ T_RM cells, but not CD8⁺ T_CM cells, demonstrated increased mitochondrial oxidative metabolism in the presence of exogenous FFAs; this increase was not seen in Fabp4/Fabp5 double-knockout CD8⁺ T_RM cells. The persistence of CD8⁺ T_RM cells in the skin was strongly diminished by inhibition of mitochondrial FFA β-oxidation in vivo. Moreover, skin CD8⁺ T_RM cells that lacked Fabp4/Fabp5 were less effective at protecting mice from cutaneous viral infection, and lung Fabp4/Fabp5 double-knockout CD8⁺ T_RM cells generated by skin vaccinia virus (VACV) infection were less effective at protecting mice from a lethal pulmonary challenge with VACV. Consistent with the mouse data, increased FABP4 and FABP5 expression and enhanced extracellular FFA uptake were also demonstrated in human CD8⁺ T_RM cells in normal and psoriatic skin. These results suggest that FABP4 and FABP5 have a critical role in the maintenance, longevity and function of CD8⁺ T_RM cells, and suggest that CD8⁺ T_RM cells use exogenous FFAs and their oxidative metabolism to persist in tissue and to mediate protective immunity.

Extended Data Figure 1. CD8⁺ T_RM cells show increased gene expression of Fabp4 and Fabp5

a, Number of OT-I Thy1.1⁺ cells in infected skin at indicated time after infection. Graph shows mean ± s.d. of 10 mice. b, Venn diagram analysis of genes differentially expressed in pairwise comparisons between OT-I T_CM, T_EM and T_RM cells (day 30) relative to that of T_N cells (fold change cutoff, ≥ 2). c, Absolute qPCR analysis of Fabp gene expression in T_RM cells (day 30). ND, not detectable. d, qPCR assessments of Fabp4 and Fabp5 levels in indicated T cell subsets. OT-I Thy1.1⁺ cells were intravenously transferred into Thy1.2⁺ recipient mice 1 day before mice were infected with 2 × 10⁶ p.f.u. VACV_OVA by intratracheal infection. After 45 days, mice were euthanized. T_CM and T_EM cells were sorted from spleen and T_RM cells were sorted from lung. e, qPCR analysis of Pparg-knockdown efficiency by lentiviral vector encoding two specific and one scrambled siRNA in OT-I CD8⁺ T_RM cells. f, qPCR analysis of Fabp4 and Fabp5 gene expression in OT-I CD8⁺ T_RM cells from mice treated with or without GW9662. Graphs in c, d, e, f show mean ± s.d. from triplicates. β-actin was used as internal control and mRNA was normalized to T_N samples (d), or T_RM samples with scrambled siRNA transduction (e) or without GW9662 treatment (f). T cells from 15–20 mice were pooled for each sample. **P < 0.01; NS, not significant.

Extended Data Figure 2. Visualization of lipid distribution in mouse skin tissue by Bodipy staining

Mouse skin cryosections were stained for lipids (4,4-Difluoro-1,3,5,7,8-Pentamethyl-4-Bora-3a,4a-Diaza-s-Indacene (BODIPY 493/503), green) and nuclei (4′,6-diamidino-2-phenylindole (DAPI), blue). Images were acquired with a Leica TCS SP8 confocal microscope and analysed with ImageJ. n = 15 sections from 5 mice and 1 representative site is shown. Dashed line indicates epidermal–dermal boundary. Scale bar, 20 μm.

Extended Data Figure 3. OT-I Fabp4^−/−/Fabp5^−/− T_RM cells show a similar surface protein expression phenotype as wild-type cells

Thy1.2⁺ CD45.2⁺ recipient mice were given a 1:1 ratio of naive OT-I Thy1.1⁺ CD45.1⁺ wild-type and OT-I Thy1.1⁺ CD45.2⁺ Fabp4^−/−/Fabp5^−/− cells one day before mice were infected with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification. Subsequently, 45 days after infection, infected skin sites were collected and the phenotype of T_RM cells was examined by flow cytometry. Data are representative of three independent experiments (n = 5 mice per group). ESL, E-selectin ligand.

Extended Data Figure 4. Relative contribution of FABP4 and FABP5 in fatty acid acquisition and long-term maintenance of skin CD8⁺ T_RM cells

a, Average MFI of Bodipy FL C₁₆ uptake by OT-I wild-type, Fabp4^+/−Fabp5^+/−, Fabp4^−/− and Fabp5^−/− T_RM cells. b–d, Number of OT-I wild-type, Fabp4^+/−Fabp5^+/−, Fabp4^−/− and Fabp5^−/− T_RM cells at different time points after VACV_OVA infection. Thy1.2⁺ CD45.2⁺ recipient mice were given a 1:1 ratio of naive OT-I Thy1.1⁺ CD45.1⁺ wild-type and OT-I Thy1.1⁺ CD45.2⁺ Fabp4/Fabp5 knockout cells (Fabp4^+/−Fabp5^+/− (b), Fabp4^−/− (c) or Fabp5^−/− (d) cells) one day before infection with 2 × 10⁶  p.f.u. VACV_OVA by skin scarification. Relative percentages of the two T cell populations were analysed longitudinally by flow cytometry. Graphs show mean ± s.d. of 5 mice per group. NS, not significant.

Extended Data Figure 5. OT-I Fabp4^−/−/Fabp5^−/− T_eff cells have similar proliferative capacity and tissue-homing receptor expression as wild-type counterparts

a, Quantification of Ki67⁺ cells in OT-I T_N, T_eff, T_CM, T_EM and T_RM cells. b, Average MFI of Bodipy FL C₁₆ uptake by OT-I T_N, T_eff, T_CM, T_EM or T_RM cells. Graphs show mean ± s.d. of 5 mice per group (a, b). *P < 0.05; **P < 0.01. c, Flow cytometric analysis of T cell proliferation and homing receptor expression on OT-I wild-type and Fabp4^−/−/Fabp5^−/− T cells. Thy1.2⁺ CD45.2⁺ recipient mice were given a 1:1 ratio of CFSE-labelled naive OT-I Thy1.1⁺ CD45.1⁺ wild-type and OT-I Thy1.1⁺ CD45.2⁺ Fabp4^−/−/Fabp5^−/− cells one day before mice were infected with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification. At 60 h after infection, proliferation and tissue-homing receptor expression of OT-I cells isolated from draining lymph nodes were analysed by flow cytometry. Data are representative of three independent experiments (n = 5 mice per group). ESL, E-selectin ligand; PSL, P-selectin ligand.

Extended Data Figure 6. Effect of Pparg lentiviral siRNA knockdown or PPARγ inhibition on CD8⁺ T_RM cell maintenance in peripheral tissue

a, Number of OT-I CD8⁺ T_RM transduced with scrambled siRNA or siRNA targeting Pparg. OT-I cells transduced with Pparg siRNA (together with the same number of congenically scrambled siRNA transduced OT-I cells) were cotransferred into recipient mice that were previously infected with VACV_OVA by skin scarification. After 40 days, mice were euthanized and the number of siRNA-transduced OT-I cells in the infected skin tissue were collected and counted by flow cytometry on the basis of the GFP marker. b, Number of OT-I CD8⁺ T_RM cells in infected skin from mice treated with or without GW9662. Thy1.2⁺ CD45.2⁺ recipient mice were infected with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification. 40 days later, mice were treated with GW9662 daily by intradermal injection for 5 days, after which mice were euthanized and T_RM cells were counted and analysed by flow cytometry. Graphs show mean ± s.d. of 5 mice per group. **P < 0.01.

Extended Data Figure 7. Gene expression profile of OT-I Fabp4^−/−/Fabp5^−/− T_RM cells

a, Principal-component analysis of gene-expression data for skin infiltrating T cells isolated at day 10 and day 30 after infection. b, Differentially expressed genes selected from a pairwise comparison between OT-I Fabp4^−/−/Fabp5^−/− T_RM and OT-I wild-type T_RM cells. Thy1.2⁺ CD45.2⁺ recipient mice were given a 1:1 ratio of naive OT-I Thy1.1⁺ CD45.1⁺ wild-type and OT-I Thy1.1⁺ CD45.2⁺ Fabp4^−/−/Fabp5^−/− cells one day before mice were infected with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification. At 30 days after infection, OT-I CD8⁺ T_RM cells were sorted from infected skin sites for gene microarray. c, qPCR analysis of genes involved in anti-inflammatory responses (Il10 and Socs2), cell apoptosis (Casp3 and Bcl2113) as well as immune responses (Ccr8, Ccl21, Il1r1, Il6, Cxcl13, Cxcl12, Ccl12, Gzmc, Gzma) in OT-I Fabp4^−/−/Fabp5^−/− T_RM cells compared to wild-type T_RM cells. Graphs show mean ± s. d. from triplicates. β-actin was used as internal control and mRNA was normalized to OT-I wild-type T_RM cells. mRNA was pooled from 15 mice from 3 independent biological groups (5 mice per group). **P < 0.01.

Extended Data Figure 8. Effect of Cpt1a lentiviral siRNA knockdown or CPT1A inhibition on OT-I CD8⁺ T_RM cell maintenance in peripheral tissue

a, qPCR analysis of Cpt1a lentiviral siRNA knockdown efficiency in OT-I CD8⁺ T_RM cells. b, c, Quantification of OT-I wild-type and OT-I Fabp4^−/−/Fabp5^−/− T_RM cells in infected skin from mice treated with or without etomoxir (b) or trimetazidine (c). Thy1.2⁺ CD45.2⁺ recipient mice were given a 1:1 ratio of naive OT-I Thy1.1⁺ CD45.1⁺ wild-type and OT-I Thy1.1⁺ CD45.2⁺ Fabp4^−/−/Fabp5^−/− cells one day before infection with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification. Mice were administered etomoxir or trimetazidine daily by intradermal injection for 5 days from day 40 after infection, after which mice were euthanized and T_RM cells were analysed by flow cytometry and counted. Symbols represent individual mice. d, OCR and ECAR as measured by the Seahorse assay of skin-infiltrating T cells sorted at indicated time points after infection under basal conditions. Graphs show mean ± s.d. of triplicates. **P < 0.01; NS, not significant.

Extended Data Figure 9. Skin CD8⁺ T_RM cells residing in distal infection sites are deficient in long-time survival and virus clearance

a, Ratio of peripheral blood CD3 counts in FTY720 versus untreated mice, illustrating the kinetics of lymphoid sequestration by FTY720. n = 5 mice per group. b, Quantification of OT-I wild-type and OT-I Fabp4^−/−/Fabp5^−/− cells infiltrating distal infection sites at different time points after VACV_OVA infection. Thy1.2⁺ CD45.2⁺ recipient mice were given a 1:1 ratio of naive OT-I Thy1.1⁺ CD45.1⁺ wild-type and OT-I Thy1.1⁺ CD45.2⁺ Fabp4^−/−/Fabp5^−/− cells one day before infection with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification of the right ears. Relative percentages of the two T cell populations infiltrating left ears were analysed longitudinally by flow cytometry. Graphs show mean ± s.d. of 5 mice per group. c, qPCR analysis of distal infection skin viral load at six days after re-challenge. OT-I wild-type or OT-I Fabp4^−/−/Fabp5^−/− cells were adoptively transferred into μMT mice before infection with 2 × 10⁶ p.f.u. VACV_OVA by skin scarification of the right ears. 25 days later, mice were re-challenged with VACV_OVA at the left ear by skin scarification. U, unimmunized; I, immunized mice. Symbols represent individual mice. **P < 0.01; NS, not significant.

Extended Data Figure 10. Visualization of lipid distribution in human skin from patients with psoriasis by Bodipy staining

Cryosections of lesional scalp skin samples from patients with psoriasis were stained for lipid (4,4-Difluoro-1,3,5,7,8-Pentamethyl-4-Bora-3a,4a-Diaza-s-Indacene (BODIPY 493/503), green) and nuclei (4′,6-diamidino-2-phenylindole (DAPI), blue). Images were acquired with a Leica TCS SP8 confocal microscope and analysed with ImageJ. Dashed lines indicate epidermal–dermal boundary. Scale bar, 50 μm.

Figure 1. Skin CD8⁺ T_RM cells show increased expression of FABP4 and FABP5

a, Principal-component analysis (PCA) of gene-expression data for CD8⁺ T cell subtypes. Each time point represents an individual experiment wherein mRNA was pooled from 15–20 mice from 3–4 independent biological groups (5 mice per group). Numbered dots are for skin T cells derived after infection for the indicated number of days. b, Pearson correlation coefficients among CD8⁺ T cell subtypes. c, Heatmap of differentially expressed genes selected from a pair-wise comparison between OT-I T_RM (day 30) and T_CM cells. d, qPCR analysis of Fabp4 and Fabp5 expression in T_N, T_CM, T_EM and T_RM cells (day 30). e, qPCR analysis of Fabp4 and Fabp5 gene expression in skin CD103⁻ and CD103⁺ T_RM cells (day 30). f, Immunofluorescence staining of FABP4 (top) and FABP5 (bottom) in OT-I T_RM cells 30 days after infection. Scale bar, 20 μm. g, qPCR analysis of Pparg expression in T_N, T_CM, T_EM and T_RM (day 30). h, Effect of lentiviral Pparg siRNA knockdown (KD) on Fabp4 and Fabp5 expression in OT-I CD8⁺ T_RM cells. Graphs in d, e, g, h show mean ± s.d. from triplicates. β-actin was used as internal control and mRNA was normalized to T_N cells (d, e, g) or T_RM cells transduced with a lentiviral vector encoding scrambled siRNA (h). T cells from 15–20 mice were pooled for each group. **P < 0.01; NS, not significant.

Figure 2. Loss of Fabp4 and Fabp5 decreases fatty-acid uptake and metabolism by CD8⁺ T_RM cells and impairs their long-term maintenance

a, Representative histograms and average mean fluourescence intensity (MFI) of Bodipy FL C₁₆ uptake by T_N, T_CM, T_EM or T_RM cells (day 30). b, Representative histograms and average MFI of Bodipy FL C₁₆ uptake in OT-I wild-type and Fabp4^−/−/Fabp5^−/− T_RM cells with or without pre-incubation with unlabelled palmitate (Palm). c, Number of OT-I wild-type and Fabp4^−/−/Fabp5^−/− T_CM and T_RM cells at different time points after infection in spleen and skin. d, Immunofluorescence staining and quantification of OT-I wild-type and Fabp4^−/−/Fabp5^−/− T_RM cells at infected skin sites 45 days after infection. FOV, field of view. Scale bar, 50 μm. e, Representative histograms and quantification of annexin V⁺ cells in OT-I wild-type and Fabp4^−/−/Fabp5^−/− T_RM cells 45 days after infection. f, OCR of OT-I T_CM, T_RM and Fabp4^−/−/Fabp5^−/− T_RM cells (day 30) under basal conditions and in response to indicated mitochondria inhibitors. Results were normalized to control cells treated with bovine serum albumin (BSA). Eto, etomoxir. Graphs show mean ± s.d. of triplicates. g, Effect of lentiviral Cpt1a siRNA knockdown on OT-I CD8⁺ T_RM survival in vivo in infected tissue. h, Representative dot plots and percentage of wild-type and Fabp4^−/−/Fabp5^−/− T_CM and T_RM cells 45 days after infection. Cells were gated on VACV-specific pentamer⁺ CD8⁺ T cells. Graphs show mean ± s.d. of 5 (a, b) or 10 (c) mice, or 50 fields from 5 mice (d, 10 fields per mouse), or symbols represent individual mice (e, g, h). *P <0.05; **P < 0.01; NS, not significant.

Figure 3. Skin Fabp4^−/−/Fabp5^−/− CD8⁺ T_RM cells fail to protect mice against viral infectious challenge

a, Schematic of experimental design. i.p., intraperitoneal injection; s.s., b, qPCR analysis of skin viral load at six days after re-infection. U, un-immunized mice; I, immunized mice. c, d, Representative dot plots and quantification of IFNγ secretion by OT-I wild-type or Fabp4^−/−/Fabp5^−/− T_RM cells. Symbols represent individual mice (b, d). e, Schematic of experimental design. f, g, Body weight (BW) (f) and survival measurements (g) of WR-VACV challenged mice, with or without FTY720 treatment. Data are representative of three independent experiments (n = 10 mice per group). *P < 0.05; **P < 0.01; NS, not significant.

Figure 4. Human skin CD8⁺ T_RM demonstrate increased expression of FABP4 and FABP5

a, b, Representative histograms and average MFI of intracellular staining of FABP4 (a) and FABP5 (b) in human peripheral blood mononuclear T_N, T_CM, and T_EM cells compared to facial skin T_RM. c, Haematoxylin and eosin staining of a human psoriatic lesion. Scale bar, 100 μm. d, Immunofluorescence staining of CD8 and CD69 expression in psoriatic human skin. Dashed line indicates epidermal–dermal junction. Scale bar, 50 μm. e, Immunofluorescence staining of FABP4 (top) and FABP5 (bottom) expression in skin CD8⁺ T_RM cells from patients with psoriasis. Scale bar, 20 μm. f, Representative histograms and average MFI of Bodipy FL C₁₆ uptake by human T_N, T_CM, T_EM or T_RM cell subsets. a, b, f, Graphs show mean ± s.d. of 5 individuals; c–e, Representative images of n = 15 sections from 5 individuals. **P < 0.01.