CXCR4 has a dual role in improving the efficacy of BCMA-redirected CAR-NK cells in multiple myeloma

Abstract

Multiple myeloma (MM) is a plasma cell disease with a preferential bone marrow (BM) tropism. Enforced expression of tissue-specific chemokine receptors has been shown to successfully guide adoptively-transferred CAR NK cells towards the malignant milieu in solid cancers, but also to BM-resident AML and MM. For redirection towards BM-associated chemokine CXCL12, we armored BCMA CAR-NK-92 as well as primary NK cells with ectopic expression of either wildtype CXCR4 or a gain-of-function mutant CXCR4^R334X. Our data showed that BCMA CAR-NK-92 and -primary NK cells equipped with CXCR4 gained an improved ability to migrate towards CXCL12 in vitro. Beyond its classical role coordinating chemotaxis, CXCR4 has been shown to participate in T cell co-stimulation, which prompted us to examine the functionality of CXCR4-cotransduced BCMA-CAR NK cells. Ectopic CXCR4 expression enhanced the cytotoxic capacity of BCMA CAR-NK cells, as evidenced by the ability to eliminate BCMA-expressing target cell lines and primary MM cells in vitro and through accelerated cytolytic granule release. We show that CXCR4 co-modification prolonged BCMA CAR surface deposition, augmented ZAP-70 recruitment following CAR-engagement, and accelerated distal signal transduction kinetics. BCMA CAR sensitivity towards antigen was enhanced by virtue of an enhanced ZAP-70 recruitment to the immunological synapse, revealing an increased propensity of CARs to become triggered upon CXCR4 overexpression. Unexpectedly, co-stimulation via CXCR4 occurred in the absence of CXCL12 ligand-stimulation. Collectively, our findings imply that co-modification of CAR-NK cells with tissue-relevant chemokine receptors affect adoptive NK cell therapy beyond improved trafficking and retention within tumor sites.

Keywords: BCMA; NK cells; adoptive T cell therapy; chemokine receptor CXCR4; chimeric antigen receptor; multiple myeloma.

Figure 1

Enforced CXCR4 expression improves migration of BCMA-CAR NK cells towards chemotactic CXCL12 gradients. (A) Schematic of retroviral constructs encoding a BCMA-CAR (CAR), equipped with either a WT CXCR4 (CAR-CXCR4) or mutant CXCR4 (CAR-CXCR4^R334X) chemokine receptor sequence. LTR, long terminal repeat; tm, transmembrane region; ic, intracellular signaling domain; SP, signal peptide. (B) Schematic of heptahelical CXCR4^R334X mutant. (C) FACS plots depicting the generation of BCMA-CAR⁺ NK-92 cells. NK-92 cells underwent retroviral transduction with the aforementioned BCMA-CAR-encoding constructs. BCMA-CAR⁺ expressing NK-92 cells, as detected by cognate FITC-conjugated BCMA-peptide, were enriched twice by FACS to generate stable BCMA-CAR⁺ NK-92 cell lines. (D) Of these BCMA-CAR⁺ NK-92 cells, the expression of CXCR4 was assessed by flow cytometry, which was (E) quantified as ΔgMFI (values represent CXCR4 stained cells minus isotype control). Statistics calculated by an unpaired t-test, error bars ± SEM. (F) FACS histograms showing CXCR4 expression on pNK cells at day +9 of ex vivo expansion (2-days post-transduction; untransduced (UT); isotype, dark filled histogram; one representative example is shown. The gating strategy used to define CD56⁺CD16⁺ pNK cells is shown on the left plot. NK cells were stimulated with IL-2 (500 U/ml) and IL-15 (20 ng/ml), in the presence of anti-CD2/CD335 activator beads. (G) Quantification as in (E), ΔgMFI values are CXCR4 stained cells minus isotype control. An unpaired t-test was applied, error bars ± SEM. (H) Western blot assessing P-ERK1,2^T202/Y204 phosphorylation of untreated and CXCL12-stimulated (25 ng/ml, 5 min) CAR, CAR-CXCR4 and CAR-CXCR4^R334X NK-92 cells. Numbers between the gel lines give the ratios between P-ERK 1,2 and total ERK1,2. (I) Transwell migration assay assessing migratory capacity of CAR (grey bars) and CAR-CXCR4 (pink bars) NK-92 cells exposed to 6.25 ng/mL or 25 ng/mL CXCL12 for 4 hrs. Relative control (rel. ctrl.) represents unstimulated cells. Statistics calculated by an unpaired t-test, error bars ± SEM. (J) Quantification of migrated CAR (grey bar), CAR-CXCR4 (pink bar) and CAR-CXCR4^R334X (green bar) pNK cells exposed to 25 ng/mL CXCL12. Threshold set arbitrarily at 1.5-fold. Statistics calculated by an unpaired t-test, error bars ± SEM. P values are given. NK cells were used at day 12–15 after start of the culture. Data points in E, G, I, J represent independent biological replicates.

Figure 2

CXCR4 co-expression enhances cytolytic activity of BCMA-CAR NK cells in the absence of CXCL12 stimulation. (A) A flow cytometry-based cytotoxicity assay was performed by co-culturing CAR-transduced NK-92 cells and Raji^BCMA target cells. Representative FACS plot depicting a 17-hour co-culture between Raji^BCMA cells with either SP6-CAR as negative control, CAR, CAR-CXCR4 or CAR-CXCR4^R334X NK-92 cells at a 0.25:1 E:T ratio. Top panel displays Raji^BCMA- or transduced NK-92-only controls. Bottom panel shows the co-cultures between Raji^BCMA cells (CD19⁺) and transduced-NK-92 cells (CD56⁺). The percentage of specific-Raji^BCMA cell killing is calculated using the formula: Killing rate %= (1- (tumor cells in coculture/tumor cells only)) x 100. (dependent upon E:T ratio). Numbers in the gates are the percentages of effector (CD56⁺) and target cells (CD19⁺), respectively. (B) Quantification of specific-Raji^BCMA cell killing (%) for the aforementioned co-cultures at the indicated E:T ratios (n=13–18 independent biological replicates). Statistics calculated by a paired t-test. (C) Representative FACS plot of a degranulation assay detecting LAMP-1 expression on CD56⁺ SP6-CAR as negative control, CAR, CAR-CXCR4 and CAR-CXCR4^R334X NK-92 cells following co-culture with Raji^BCMA cells for 30-min. Frequencies of LAMP-1⁺ NK-92 cells were calculated by subtracting percentage LAMP-1⁺ expression on transduced-NK-92 cells alone (grey plot) by percentage LAMP-1⁺ expression on NK-92 cells co-cultured with Raji^BCMA cells (red plot). Numbers within gates are the frequencies in percent of LAMP-1⁺ cells. (D) Quantification bar plots of (C) after 30- and 60-min stimulation. Statistics calculated by an unpaired t-test. (E) Schematic of G protein-deficient CXCR4^R134N mutant and retroviral construct. The position of the conserved DRY motif is indicated. (F) Representative FACS plot of a degranulation assay following a 30-minute co-culture between transduced NK-92 cells (including CAR-CXCR4^R134N mutant) and Raji^BCMA cells in the presence or absence of 25 ng/ml CXCL12. Numbers below gates are the frequencies in percent of LAMP-1⁺ cells upon stimulation. (G) Quantification of aforementioned degranulation assays (n=5) with (CXCL12; hatched bars) or without (w/o CXCL12; solid bars) CXCL12. Statistics calculated using an unpaired t-test comparing different constructs and a paired t-test assessing each construct with or without CXCL12. Error bars ± SEM. P values are given; data points in (D), (G) represent independent biological replicates.

Figure 3

Multiple Myeloma cells with endogenous BCMA expression are more susceptible to BCMA CAR NK-92 cells equipped with the CXCR4 chemokine receptor. (A) A flow cytometry-based cytotoxicity assay was performed by co-culturing CAR-transduced NK-92 cells and MM.1S-luc.GFP target cells. Representative FACS plot depicting a 6-hour co-culture between MM.1S-luc.GFP cells with either SP6-CAR as negative control, CAR, or CAR-CXCR4 NK-92 cells at a 0.25:1 E:T ratio. Effector NK-92 cells were detected by anti-CD56 staining, and target cells were GFP labeled for discrimination. Numbers on the bottom gates are percentages of viable tumor target cells after co-culture. (B) The percentage of specific-MM.1S-luc.GFP cell killing is calculated as in Figure 2A ). (dependent upon E:T ratio). Values displayed are mean ± SEM (n=4 independent experiments). Statistics calculated by a paired t-test. (C) A representative FACS plot of co-cultures between primary MM specimen and NK-92 effector cells. Effector cells were labeled with eFluor670⁺ for better distinction from MM target cells, defined by CD38^+high eFluor670^-. Because of the low frequency of primary MM target cells in the specimen, killing rates were calculated relative to the negative control, SP6 CAR. Statistics calculated by a paired t-test; n=4 independent donor samples tested. (D) Formula to quantify the killing rate is Killing rate %= (1- (tumor cells in coculture with test CAR/tumor cells in coculture with SP6 CAR control)) x100.

Figure 4

CXCR4 overexpression prolongs BCMA-CAR surface expression and is associated with enhanced levels of CAR-mediated trogocytosis. (A) Representative FACS histograms showing the kinetics of BCMA-CAR surface expression (as detected by IgG staining) on CAR (grey histograms) and CAR-CXCR4 (pink histograms) NK-92 cells following exposure to BCMA-beads for the indicated timepoints. (B) Quantification of BCMA-CAR surface expression. gMFI values were set arbitrarily at 100% at 0 min. The change in gMFI over time was quantified as a percentage of initial surface expression (n=3 independent biological replicates). Statistics calculated using an unpaired t-test (CAR versus CAR-CXCR4). (C) Representative FACS histograms of CXCR4 expression on CAR-CXCR4 NK-92 cells after exposure to BCMA-beads (pink histograms) or CXCL12 (grey histograms; pick-dotted line) for the indicated timepoints. (D) Quantification of CXCR4 surface expression, calculated as for (B) (BCMA-beads n=4; CXCL12 n=5 independent biological experiments). Statistics calculated using a paired t-test (BCMA-beads; timepoint versus ‘0 min’) or Wilcoxon test (CXCL12; timepoint versus ‘0 min’), where appropriate according to sample size and distribution. (E) Representative FACS plots depicting the proportion (%) of CD56⁺ CD19⁺ SP6-CAR, CAR, CAR-CXCR4 and CAR-CXCR4^R334X NK-92 cells (ordinarily CD19^-) following 30 min co-culture with CD19⁺ Raji^BCMA target cells, indicative of CAR activation-induced trogocytosis. Cell gating was on single cells (FSC-W/FSC-H). Top panel shows ‘NK-92 only’ control cells, while the bottom panel represents the co-culture. (F) Violin plots quantifying the ratio of SP6-CAR (white plot), CAR (grey), CAR-CXCR4 (pink) and CAR-CXCR4^R334X (green) NK-92 cells that had undergone (TROG+) or had not yet experienced (TROG-) trogocytosis of Raji^BCMA target cells (TROG+/TROG-). Individual datapoints for independent biological replicates are shown. Statistics calculated using a paired t-test. (G) Representative FACS plots detecting LAMP-1 expression on the aforementioned CAR-CXCR4^TROG- (CD56⁺ CD19^-; top panel) and CAR-CXCR4^TROG+ (CD56⁺ CD19⁺; bottom panel) NK-92 cells following a 30 min co-culture with Raji^BCMA cells. Frequencies of LAMP-1⁺ CAR-CXCR4^TROG- or CAR-CXCR4^TROG+ cells were calculated by subtracting percentage LAMP-1⁺ expression on CAR-CXCR4 NK-92 cells alone (black contour plot) by percentage LAMP-1⁺ expression on either CAR-CXCR4^TROG- or CAR-CXCR4^TROG+ NK-92 cells (pink contour plot). Numbers within gates are the frequencies in percent of LAMP-1⁺ cells. (H) Quantification bar plots of (G). Data points represent independent biological replicates. Statistics calculated by a paired t-test. P values are given; statistical significance P<0.05. Error bars ± SEM.

Figure 5

Enforced CXCR4 expression is associated with increased P-ZAP-70^Y319 recruitment into the IS and boosted BCMA-induced proximal signaling. (A) Exemplary single plane micrographs depicting IS formation between Raji^BCMA cells and CAR, CXCR-CXCR4 or CAR-CXCR4^R334X NK-92 cells during 5 min co-incubations. The immunological synapse was identified by accumulation of actin cytoskeleton at the interface of NK-92 cells and Raji^BCMA target cells. Cells were probed for CXCR4 (blue), P-ZAP-70^Y319 (green), F-actin (red; representative of IS interface) and BCMA (grey; target cell marker). The region of interest (ROI) represents the IS, where P-ZAP-70^Y319 localizes to. Scale bar, 5 µm. Representative image, acquired on an LSM-980 airyscan microscope, equipped with a 63x Plan Apochromat NA 1.40 oil objective. Digital images were obtained with ZEN software (Zeiss), and further processed with Fiji. (B) Quantification of phosphorylated P-ZAP-70^Y319 MFI within the ROI following 5- or 20-minute co-incubations. CAR=grey bar, CAR-CXCR4=pink bar and CAR-CXCR4^R334X=green bar. Statistics calculated using an unpaired t-test. Each data point represents one effector-target NK-92-Raji^BCMA cell conjugate. At least n=2 independent experiments were performed. (C) Representative immunoblot assessing P-ZAP-70^Y319 phosphorylation of unstimulated and BCMA-bead-stimulated (5-minute) SP6-CAR control, CAR, CXCR-CXCR4 and CAR-CXCR4^R334X NK-92 cells. (D) Quantification (densitometry) of P-ZAP-70^Y319 phosphorylation for the aforementioned Western blots (n=8 independent biological replicates), calculated by dividing P-ZAP-70^Y319 signal intensity by total ZAP-70 signal intensity. SP6-CAR control=white bars, CAR=grey bars, CAR-CXCR4=pink bars, and CAR-CXCR4^R334X=green bars. Individual datapoints shown. Statistics calculated using a Mann-Whitney test. (E) Kinetics of ERK1,2 phosphorylation assessed by immunoblotting of BCMA-bead stimulated CAR-transduced NK-92 cells. In C) and E), numbers between the gel lines give the ratio between phosphorylated ZAP-70 and phosphorylated ERK 1,2 and their total protein forms ERK1,2 or AKT, respectively. (F) Quantification of P-ERK1,2^T202/Y204 occurrence. The ratio of total ERK1,2 and phosphorylated P-ERK1,2 is given (n=6 independent biological replicates). Statistics calculated using an unpaired t-test. Error bars ± SEM.

Figure 6

Enforced CXCR4 expression lowers the antigen threshold for BCMA-CAR activation and heightens BCMA-CAR sensitivity. (A) Flow cytometry analysis of adhesion molecule expression in BCMA-CAR and CAR-CXCR4 enhanced NK-92 cells. Numbers in the histograms indicate gMFI values, grey solid graphs depict isotype controls. (B) CAR-transduced NK-92 cells and Raji^BCMA target cells were co-cultured as in Figures 2C, F . Conjugate formation was performed either with or without CXCL12 supplementation (25 ng/ml). After 30 min, cells were stained for CD56 and CD19. Numbers in the gates indicate the percentage of cells forming conjugates. (C) Quantification of conjugate formation, as determined by cell doublets (FSC-H/FSC-W) and CD56⁺/CD19⁺ double positive cells. SP6-CAR ctrl=white plot, CAR=grey plot, CAR-CXCR4=pink plot, and CAR-CXCR4^R334X=green plot. Individual datapoints for biological replicates are shown. An unpaired and paired t-test was applied, where appropriate. (D) Representative immunoblot examining rapid P-ERK1,2^T202/Y204 phosphorylation of CAR and CXCR-CXCR4 NK-92 cells exposed to decreasing amounts of cognate BCMA antigen (BCMA-beads; 5, 1.67, 0.56, 0.19 and 0 µL) for 5 minutes. (E) Quantification (densitometry) of P-ERK1,2^T202/Y204 phosphorylation for the abovementioned western blots (n=7 independent biological replicates), calculated by P-ERK1,2^T202/Y204 signal intensity divided by ERK1,2 signal intensity. CAR=grey bars and CAR-CXCR4=pink bars. Statistics calculated using a Wilcoxon test.