Redirection to the bone marrow improves T cell persistence and antitumor functions

Abstract

A key predictor for the success of gene-modified T cell therapies for cancer is the persistence of transferred cells in the patient. The propensity of less differentiated memory T cells to expand and survive efficiently has therefore made them attractive candidates for clinical application. We hypothesized that redirecting T cells to specialized niches in the BM that support memory differentiation would confer increased therapeutic efficacy. We show that overexpression of chemokine receptor CXCR4 in CD8+ T cells (TCXCR4) enhanced their migration toward vascular-associated CXCL12+ cells in the BM and increased their local engraftment. Increased access of TCXCR4 to the BM microenvironment induced IL-15-dependent homeostatic expansion and promoted the differentiation of memory precursor-like cells with low expression of programmed death-1, resistance to apoptosis, and a heightened capacity to generate polyfunctional cytokine-producing effector cells. Following transfer to lymphoma-bearing mice, TCXCR4 showed a greater capacity for effector expansion and better tumor protection, the latter being independent of changes in trafficking to the tumor bed or local out-competition of regulatory T cells. Thus, redirected homing of T cells to the BM confers increased memory differentiation and antitumor immunity, suggesting an innovative solution to increase the persistence and functions of therapeutic T cells.

Keywords: Cancer immunotherapy; Chemokines; Immunology; T cells; Therapeutics.

Figure 1. Adoptively transferred T^CXCR4 demonstrate superior recruitment to the BM.

(A) Representative flow cytometry plots for CXCR4 expression in untreated CD8⁺ T cells (unstimulated), T^Control, or T^CXCR4. Gating based on “fluorescence minus 1” controls. CXCR4 median fluorescence index (MFI): 380 unstimulated; 587 GFP⁺ T^Control; 2,409 GFP⁺ T^CXCR4. (B and C) Equal mixtures of T^CXCR4 (CD45.1⁺) and T^Control (Thy1.1⁺) were injected into sublethally irradiated B6 mice. Representative plots of T^CXCR4 (red) and T^Control (blue) frequencies in BM, spleen (Sp), and LN at day 7 are shown in B. Summary graphs in C indicate mean ± SD T^CXCR4/T^Control ratio at timed intervals in BM, Sp, and LN (n = 6 per group at 3 and 24 hours, n = 4 per group at day 7). Statistical comparison was performed by Wilcoxon’s signed-rank test against a hypothetical ratio of 1.0 (dotted line). *P ≤ 0.05. (D) Box-and-whisker graphs for BM/LN ratio on day 14 following transfer of T^CXCR4 or T^Control to separate sublethally irradiated B6 mice, calculated by division of percent GFP⁺ of BM CD8⁺ T cells by percent GFP⁺ of LN CD8⁺ T cells (n = 6 T^CXCR4, n = 5 T^Control). (E) Box-and-whisker graphs of T^CXCR4/T^Control ratio in BM, Sp, and LN at day 7 following transfer into sublethally irradiated B6 mice (n = 4) and untreated Rag1ko mice (n = 10). (F) Box-and-whisker graphs of T^CXCR4/T^Control ratio in BM, Sp, and LN at day 7 following transfer into untreated B6 mice (n = 11) and untreated Rag1ko mice (n = 10). (G) Box-and-whisker graphs of ratio of T^CXCR4/T^Control in BM, Sp, and LN at day 7 following transfer into Rag1ko (n = 10), Rag1ko.Il15rako (n = 10), and Rag1ko.Il7ko (n = 4). Statistical comparisons in D and E were made using the Mann-Whitney test (2-tailed). *P ≤ 0.05, **P ≤ 0.01. All data are pooled from 2–3 independent experiments.

Figure 2. T^CXCR4 show enhanced motility and directed migration to vascular-associated CXCL12⁺ cells in BM.

(A) Left: Diagram showing strategy for calvarial imaging. Right: Intravital confocal calvarial imaging of transduced T cells (green) was performed 8 weeks after injection of T^Control and T^CXCR4 into separate Rag1ko mice. Representative maximum projection tile scans and corresponding high-magnification insets are shown following i.v. injection of anti-CXCL12–PE (red) and Cy5-dextran to identify vasculature (blue). Scale bars: 500 μm in low-magnification images, 50 μm in inset images. (B) Summary graph showing arrest coefficient data for time-lapse imaging of T^Control and T^CXCR4. Data are pooled from 5 mice (n = 2 T^Control and n = 3 T^CXCR4). Median tracking period was 8.5 min/cell, range 8.5–30 min/cell (total number of cells tracked n = 72 T^CXCR4, n = 16 T^Control). (C) Summary graph showing mean velocity of tracked cells. Statistical comparison was made using a t test (2-tailed). **P ≤ 0.01. (D) X-y graphs showing GFP intensity (x axis) versus distance (y axis) of individual T^Control (n = 54, left) and T^CXCR4 (n = 108, right) from CXCL12⁺ cells measured on static images derived from the same experiments in A–C. Inset to each graph shows Pearson’s correlation coefficient r and significance value.

Figure 3. Ag-activated T^CXCR4 retain a CD62L^hi phenotype.

Equal numbers of OT-I T^CXCR4 and T^Control were coinjected into Rag1ko mice, before prime-boost vaccination with relevant SIINFEKL peptide plus IFA (Antigen, n = 17) or irrelevant peptide plus IFA (No antigen, n = 8) on days 1 and 29. Tissues were harvested on day 8 (n = 4 per group), 29 (n = 3 Ag, n = 1 no Ag), and 36 (n = 9 Ag, n = 3 no Ag). Data from 4 independent experiments (with the exception of day 29 no-Ag group derived from 1 experiment). (A) Summary (mean ± SD) T^CXCR4/T^Control in BM, Sp, and LN over time for no-Ag (circles, dashed lines) and Ag groups (squares, solid line). Arrows indicate time of prime-boost vaccination. (B) Summary (mean ± SD) CD62L expression over time in T^CXCR4 (red) and T^Control (blue) in no-Ag (left) and Ag groups (right). (C) Representative plots for T^CXCR4 and T^Control accumulation and surface expression of CD44 and CD62L on day 36 in BM, Sp, and LN. Numbers denote frequencies of T^CXCR4 (red), T^Control (blue), and proportions of CD62L^hi and CD62L^lo cells (black). (D) Box-and-whisker graphs showing summary of T^CXCR4/T^Control ratios in BM, Sp, and LN on day 36 following transfer into Rag1ko (Il15ra WT, n = 9) and Rag1ko.Il15rako (Il15rako, n = 6) mice undergoing the same prime-boost vaccination schedule outlined in A. Data derived from 3 independent experiments. (E) Box-and-whisker graphs showing summary data for CD62L expression by T^Control or T^CXCR4 on day 36 in the same experiments outlined in D. In A and D, statistical comparisons were made by Wilcoxon’s signed-rank test against a hypothetical ratio of 1.0 (dotted line). *P ≤ 0.05, **P ≤ 0.01. In B, C, and E, statistical comparisons were made using the Mann-Whitney test (2-tailed). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure 4. Ag-activated T^CXCR4 adopt a less differentiated memory phenotype.

Equal numbers of OT-I T^CXCR4 and T^Control were coinjected into Rag1ko mice, which then underwent prime-boost vaccination with relevant SIINFEKL peptide plus IFA on days 1 and 29. Tissues were harvested on day 36 (n = 9); data are pooled from 4 independent experiments. (A and B) Representative flow cytometric histograms (A) and summary data (B) for expression of surface IL-15Rβ (CD122), intracellular Bcl2, EdU incorporation, and caspase-3 activity in T^Control (blue) versus T^CXCR4 (red) on day 36 in cells isolated from the BM. Statistical significance was tested using Wilcoxon’s ranked-sum test (2-tailed). *P ≤ 0.05, **P ≤ 0.01. (C and D) Representative flow cytometric histograms (C) and summary data (mean ± SD) (D) for CD62L and Bcl2 staining in BM T^CXCR4 gated according to GFP reporter expression (gates 1–4).Numbers shown as insets of the flow cytometric histograms relate to CD62L median fluorescence index (MFI) and proportion of Bcl2⁺ cells in the gated subset. Statistical significance was tested using the Mann-Whitney test (2-tailed). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. (E and F) Representative flow cytometric contour plots (E) and summary data (F) for frequency of splenic T^Control (blue) and T^CXCR4 (red) with short-lived effector cell (SLEC) (KLRG-1^hiCD127^lo), MPEC (KLRG-1^loCD127^hi), and exhausted (PD-1^hiEomes^hi) phenotypes on day 36 (n = 5). Statistical significance was tested using Wilcoxon’s ranked-sum test (2-tailed). *P ≤ 0.05.

Figure 5. Ag-activated T^CXCR4 have increased potential for polyfunctional cytokine generation.

Equal numbers of OT-I T^CXCR4 and OT-I T^Control were coinjected into Rag1ko mice, which then underwent prime-boost vaccination with relevant SIINFEKL peptide plus IFA on days 1 and 29. T cells were isolated from the spleen on day 36 (n = 7). (A) Representative flow cytometric contour plots showing IFN-γ and TNF-α intracellular costaining in OT-I T^CXCR4 and OT-I T^Control after ex vivo stimulation with relevant peptide with gates set according to stimulation with irrelevant peptide. (B) Summary data for IFN-γ, TNF-α, and IL-2 generation from OT-I T^Control (blue) and OT-I T^CXCR4 (red) in the same assays. Statistical significance tested using the Wilcoxon ranked sum test (two-tailed), *P ≤ 0.05, **P ≤ 0.01. (C) Pie charts depicting polyfunctional cytokine generation in T^Control and T^CXCR4 according to Boolean combination gates identifying IFN-γ⁺, TNF‑α⁺, and IL-2⁺ cells. (D) Transduced OT-I T^CXCR4 and OT-I T^Control were stimulated in vitro with relevant or irrelevant peptide and in the absence or presence of 500 ng/ml of recombinant murine CXCL12. Representative flow cytometric contour plots showing CD44 and CD62L surface expression. Data shown are representative of 2 independent experiments. (E) Summary data (mean ± SD) for intracellular IFN-γ generation after in vitro stimulation with anti-CD3 in the presence or absence of CXCL12 (n = 3 from 3 independent experiments).

Figure 6. Resting memory T^CXCR4 possess a memory precursor–like signature.

(A) Heatmap showing the relative expression levels of the genes differentially expressed between resting memory OT-I T^CXCR4 and controls (fold change ≥1.5, P ≤ 0.01); labels identify specific genes as discussed in the text. (B) The network map displays the Reactome gene sets enriched in OT-I T^CXCR4 versus OT-I T^Control. Node area indicates the size of the gene set; color code reflects enrichment in OT-I T^CXCR4 (red) or OT-I T^Control (blue); color intensity is proportional to statistical significance. Clusters of functionally related gene sets were manually circled and assigned a label. NES, normalized enrichment score. (C) Gene set enrichment analysis–based (GSEA-based) assessment of the stage of antigenic response of OT-I T^CXCR4 and OT-I T^Control according to the 10 phase-specific gene sets identified by Best et al. (42). Results are represented in a BubbleGUM plot, in which stronger and more significant enrichments are represented by larger and darker bubbles, colored red for OT-I T^CXCR4 or blue for OT-I T^Control. (D) GSEA plots showing that memory OT-I T^CXCR4 upregulate genes associated with early memory cells and with increased responsiveness to IL-15.

Figure 7. T^CXCR4 demonstrate enhanced tumor protection.

All mice underwent B6→BALB/c BMT. (A) A20 tumors were implanted on day 0; 2 days later, recipients received CD8⁺ allo-T^Control (blue circles, n = 5), CD8⁺ allo-T^CXCR4 (red circles, n = 5), or no T cells (black triangles, n = 3). Graph shows tumor size at timed intervals. (B) A20.hCD34⁺ cells were given by intraosseous injection to the left tibia on day 0; 2 days later, recipients received CD8⁺ allo-T^Control (blue circles, n = 4 day 11, n = 9 day 18), 1 × 10⁶ CD8⁺ allo-T^CXCR4 (red circles, n = 4 day 11, n = 9 day 18), or no T cells (black triangles, n = 2 day 18). Graphs show mean ± SD A20.hCD34⁺ accumulation in ipsilateral (left) and contralateral (right) tibia. Data pooled from 2 experiments. (C) CD8⁺ allo-T^Control or allo-T^CXCR4 were given on day 2 (n = 5 per group); graph shows absolute numbers of CD62L^hi and CD62L^lo donor CD8⁺ T cells on day 10 post-BMT. *P ≤ 0.05, **P ≤ 0.01 Mann-Whitney test, 2-tailed. (D) Allo-T^Control and allo-T^CXCR4 (n = 7 per group), or no T cells (n = 3), were given on day 2; graph shows in vivo specific cytotoxicity against BALB/c B cells on day 10 post-BMT. ND, no data. (E) A20 tumors were implanted s.c. on day 0; 2 days later, BMT recipients received no T cells (black triangles, n = 3), CD3⁺ allo-T^Control (blue circles, n = 5), or CD3⁺ allo-T^CXCR4 (red circles, n = 5). Graph shows tumor size at timed intervals. (F) Experimental design as in E. Weight ratio and histological GVHD score on day 10 post-BMT (allo-T^Control, n = 13; allo-T^CXCR4, n = 13; no T cells, n = 3). Data pooled from 2 experiments. (G) On day 2, luc⁺ CD3⁺ allo-T^Control or allo-T^CXCR4 were transferred to BMT recipients bearing subcutaneous A20 tumors. T cell infiltration was monitored at timed intervals (mean ± SD, n = 3 per group). (H) Mean ± SD luc⁺ Treg accumulation at timed intervals within A20 tumors following cotransfer on day 2 in a 1:1 ratio with non-luc⁺ T^Control (blue squares) or T^CXCR4 (red squares).