CD70-Targeting CAR NK Cells Overcome BCMA Downregulation and Improve Survival in High-risk Multiple Myeloma Models

CD70 is highly expressed in high-risk multiple myeloma and correlates with poor survival. CD70-targeting CAR NK cells exert strong anti-myeloma cytotoxicity, comparable to CAR T cells, including against BCMA-knockout models.

Abstract

CD70 is highly expressed in many cancers, including multiple myeloma. We show in two cohorts of patients with multiple myeloma that CD70 is elevated in several high-risk disease categories and correlates with poor survival. These findings were validated using single-cell RNA sequencing, flow cytometry, and IHC. Moreover, we demonstrate the feasibility of targeting CD70 in myeloma using NK cells engineered with a chimeric antigen receptor (CAR) incorporating the CD70 cognate receptor CD27 and IL-15 (CAR27/IL-15). CAR27/IL-15 NK cells exerted potent in vitro and in vivo cytotoxicity against CD70⁺ multiple myeloma cells, comparable with CAR27/IL-15 T cells, and remained effective in BCMA knockout models. Collectively, these results establish CD70 as a promising therapeutic target for high-risk multiple myeloma, particularly for patients who relapse after BCMA-directed therapy, providing preclinical support for the ongoing phase I/II clinical trial of CD70-targeting CAR NK cells (NCT05092451).

Significance:

We demonstrate that CD70 expression is elevated in patients with high-risk multiple myeloma and in patients with t(4;14) translocation. CD70-targeting CAR NK cells exhibit potent cytotoxicity against CD70⁺ multiple myeloma cells and significantly improve survival in xenograft mouse models of multiple myeloma, even in the absence of BCMA expression.

See related commentary by Benson Jr and Caligiuri, p. 166

Introduction

Multiple myeloma (MM) is the second most frequent hematologic malignancy and remains incurable despite significant advances in treatment options. High-risk MM, characterized by specific cytogenetic abnormalities or relapsed/refractory MM (RRMM) disease, poses a particularly difficult therapeutic challenge even with high-dose chemotherapy and autologous stem cell transplantation (1). Clinical data from two FDA-approved chimeric antigen receptor (CAR) T-cell therapies targeting B-cell maturation antigen (BCMA) support the efficacy of CAR-based immunotherapy, with high overall objective response rates in multiple myeloma (2, 3). However, CAR T-cell therapies present a number of challenges, including the high cost and logistical complexity of manufacturing autologous CAR T-cell products, the risk of GVHD in the allogeneic setting (4), and treatment-related toxicities such as cytokine release syndrome, immune effector cell–associated neurotoxicity syndrome, and parkinsonism (5–7). Furthermore, despite initial responses in multiple myeloma, most patients treated with BCMA-targeted CAR T-cell therapy eventually relapse (2, 3), and there is currently no standard of care for those who progress after BCMA-directed therapy.

Engineered NK cells derived from cord blood (CB) offer a potential alternative to CAR T-cell therapy. NK cells are critical members of the innate immune system, operating as the first line of defense against viruses and cancer cells (8). Unlike T cells, NK cells possess endogenous cytotoxic activity against myeloma through their germline-encoded activating receptors, allowing them to recognize and eliminate tumor cells independent of antigen loss. They do not cause GVHD or other alloimmune or autoimmune toxicities and thus can provide a source of allogeneic “off-the-shelf” therapies that are scalable and safe (9). Our previous studies have demonstrated the feasibility and safety of CAR NK cells in B-cell malignancies, establishing their potential as an effective therapeutic platform (10, 11).

CD70 is a member of the TNF superfamily and the natural ligand for the cytokine receptor CD27 (12). CD70 is typically only expressed transiently on a small subset of activated T, B, and dendritic cells but is upregulated in various malignancies (13, 14). Although CD70 has been extensively studied in solid tumors and some hematologic malignancies, its role in multiple myeloma has been less explored (14, 15). Here, we demonstrate that CD70 is significantly overexpressed in patients with advanced multiple myeloma, particularly those with high-risk cytogenetics or with the t(4;14) translocation, and that its expression correlates with worse overall survival. Importantly, we demonstrate that CD70 expression is upregulated following BCMA loss, highlighting its potential as a therapeutic target in relapsed multiple myeloma.

Given the urgent need for effective alternative immune therapies to overcome BCMA resistance, we developed CD70-targeting CAR NK cells [CD27-based CAR/IL-15 (CAR27/IL-15)] by fusing the extracellular domain of CD27 to a signaling endodomain and coexpressing IL-15 to enhance persistence and expansion. We evaluated their anti-multiple myeloma efficacy in vitro and in vivo, demonstrating that these CAR NK cells effectively eliminated CD70⁺ multiple myeloma cells, including BCMA knockout (KO) models, and significantly improved survival in xenograft mouse models of multiple myeloma. These results establish CD70-targeting CAR NK cells as a promising therapeutic option for patients with high-risk multiple myeloma and RRMM, particularly for patients who progress after BCMA-targeted therapies.

Results

Expression of CD70 in Multiple Myeloma in the CoMMpass Dataset

To evaluate CD70 as a potential therapeutic target for multiple myeloma, we analyzed its expression in RNA sequencing (RNA-seq) data from newly diagnosed multiple myeloma (NDMM) using the Multiple Myeloma Research Foundation (MMRF) CoMMpass dataset. Chromosomal abnormalities, including translocations and mutations, are known to be critical primary events for multiple myeloma initiation. We found that different subgroups had different CD70 mRNA expression levels. For example, CD70 mRNA was significantly increased in patients with high-risk cytogenetics as defined by the International Myeloma Working Group (16), including t(4;14), t(14;16), amp1q, and del17p, compared with patients with the t(11;14) translocation, which exhibited lower levels of CD70 mRNA expression and was thus used as a comparator group [Fig. 1A; P ≤ 0.001 for each comparison with t(11;14) and P < 0.05 for t(14;16) vs. t(11;14)]. Patients with advanced disease, as indicated by International Staging System (ISS) stage 2 or 3, had significantly higher levels of CD70 mRNA compared with ISS stage 1 (Fig. 1B; P = 0.049 for the comparison between ISS stages 1 and 2 and P = 0.0028 for the comparison between ISS stages 1 and 3), although no significant difference was observed between ISS stages 2 and 3. Moreover, we found no significant difference in CD70 mRNA abundance when stratified by sex or race/ethnicity (Supplementary Fig. S1A and S1B).

Figure 1.

CD70 expression in patients with multiple myeloma. Comparison of CD70 mRNA levels based on (A) RNA levels from the MMRF CoMMpass dataset of different translocations, Amp1q and Del17p, in patients with NDMM and (B) the ISS (n = 561). C, Paired comparison of CD70 mRNA in NDMM and at first relapse (left). The difference between the expression levels is plotted (right, relapsed–NDMM) with a dotted line at the mean of differences (0.609, P = 0.0028, paired t test, n = 61). D, Patients were stratified into tertiles based on CD70 mRNA expression, and overall survival (OS) was plotted (n = 792). Color shading represents the 95% confidence interval. E, Bone marrow aspirates from patients with multiple myeloma (n = 46) from the MD Anderson Cancer Center with different translocations involving chromosome 14 were analyzed for CD70 mRNA using the pseudobulk approach on single cell (sc)-mRNA. The control group included samples from n = 3 healthy control donors. The fourth data point in the control group represents nontumor or polyclonal plasma cells (PC) from patients with multiple myeloma identified by the expression of B-cell receptor VDJ using scRNA-seq (see “Methods”); monoclonal plasma cells were similarly defined based on BCR VDJ sequences, and each remaining data point corresponds to an individual patient with multiple myeloma: no translocation (none; n = 20), t(11;14) (n = 16), t(14;16) (n = 2), and t(4;14) (n = 8). F, The sc-mRNA transcriptome data from all samples projected onto a Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP). Individual CD70 mRNA was mapped at single-cell resolution. G, Translocations involving the immunoglobulin heavy chain locus identified by FISH are mapped onto a UMAP. Box and whisker plots show the median and IQR (25th and 75th percentiles), and the whiskers extend to ±1.5 times the IQR. The P values were determined by the Wilcoxon test [A, B, D (left panel) and E]. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant. UMI, unique molecular identifier.

Paired analysis of samples from 61 patients from the CoMMpass dataset also demonstrated that CD70 mRNA expression was significantly higher at the time of relapse compared with the time of diagnosis (Fig. 1C; P = 0.0028). When we stratified all NDMM patients by CD70 mRNA levels, patients in the top third of expression had a significantly lower probability of survival compared with those in the lowest third of expression (Fig. 1D; P ≤ 0.001). Together, these data suggest an association between CD70 mRNA expression and the clinical outcome of patients with multiple myeloma, particularly in patients with high-risk cytogenetic features and relapsed disease.

Validation of CD70 mRNA Expression in Multiple Myeloma

To further validate the findings from the CoMMpass dataset, we collected bone marrow aspirates from healthy donors (n = 3) and patients with multiple myeloma (n = 46) with immunoglobulin heavy chain–related translocations: t(11;14), t(14;16), and t(4;14). Single-cell RNA-sequencing (scRNA-seq) was performed on CD138⁺ myeloma cells isolated from patient samples and CD138⁺ plasma cells from healthy donors. Using pseudobulk counts of CD70 mRNA, we observed significantly elevated CD70 mRNA levels in patients with t(4;14) compared with those with t(11;14) (Fig. 1E). Given that not all myeloma cells from patients harbor the t(4;14) translocation, we further constructed Uniform Manifold Approximation and Projection for Dimension Reduction plots of all single cells collected. These plots visualized CD70 expression (Fig. 1F) for the different translocations based on FISH (Fig. 1G), confirming the strong correlation with t(4;14). We also performed pseudobulk counts of nuclear receptor–binding SET domain 2 (NSD2), showing a high expression level in patients with t(4:14) (17) (Supplementary Fig. S2A and S2B). Interestingly, CD70-positive clusters strongly correlated with clusters showing high NSD2 expression (Supplementary Fig. S2C).

CD70 Surface Expression in Myeloma

BCMA is preferentially expressed on plasma cells and memory B cells at a relatively low level, and its overexpression and activation are closely associated with multiple myeloma, making it a well-established therapeutic target (18). To investigate the potential of CD70 as an alternative or complementary target, we analyzed bone marrow aspirates from 30 sequential patients with multiple myeloma (7 NDMM and 23 relapsed) by flow cytometry (Fig. 2A). In addition to CD70 and BCMA, we characterized the expression of other myeloma-associated markers, including CD38, CS1, CD56, and CD45 within the CD138⁺, CD3⁻, CD19/20⁻, and CD14⁻ populations (Fig. 2B) for all patient samples. The gating strategy for these analyses is shown in Supplementary Fig. S3. A notable proportion of patients demonstrated high CD70 surface protein levels, supporting the feasibility of targeting CD70 in multiple myeloma. Although a trend toward higher CD70 expression was observed in patients with relapsed disease compared with NDMM, the difference was not statistically significant (CD70 mean fluorescence intensity in the CD138⁺ population: 2,524 in newly diagnosed and 4,446 in relapsed; P = 0.3447), likely due to the small sample size. These findings suggest that CD70 is broadly expressed at the protein level in multiple myeloma, including in relapsed disease. To enhance visualization of marker expression within myeloma cell subpopulations, we generated optimized stochastic neighbor embedding plots from a representative patient, depicting marker distribution within the CD138⁺, CD3⁻, CD19/20⁻, and CD14⁻ subset (Fig. 2C).

Figure 2.

Phenotypic characterization of CD70 expression in patients with multiple myeloma. A, Bone marrow aspirates from 30 patients with multiple myeloma were stained with anti-CD70 and anti-BCMA antibodies and assayed by flow cytometry. The percentage of positive cells from the FSC/SSC/Single cells/Live-Dead/CD3⁻/CD19⁻/CD20⁻/CD14⁻/CD138+ population is plotted, with each dot representing one patient sample. The gating strategy is shown in Supplementary Fig. S3. B, The percentage of positive expression (circle size) and the robust z-score calculated from the mean fluorescence intensity (color) for CD70, BCMA, CD38, CS1, CD56, and CD45 proteins from the same population were plotted in a dot plot for each of the patient samples in (A). C, An optimized stochastic neighbor embedding (opt-SNE) plot was generated for each marker in the CD138⁺, CD3⁻, CD19/20⁻, and CD14⁻ populations from an example patient with amp 1q and loss of one copy of TP53. Representative images of IHC CD70 staining of the gingiva (D) and bone marrow (E) clot of a patient with high-grade myeloma. F, CD70 staining of the humerus of a patient who had not responded to the BCMA antibody–drug conjugate belantamab mafodotin. G, CD70 staining from the bone marrow of a patient without multiple myeloma as a control. Scale bar, 100 μm.

IHC analysis further confirmed high CD70 expression in various myeloma tissue samples, including in the gingiva of a patient with del17p and gain 1q (Fig. 2D), a bone marrow clot from a patient with plasma cell leukemia and t(14;16) (Fig. 2E), and the humerus of a patient who had progressed on the BCMA antibody–drug conjugate belantamab mafodotin (Fig. 2F) compared with a control bone marrow sample (Fig. 2G). Collectively, these data demonstrate a high level of CD70 protein in myeloma samples, including in high-risk and relapsed disease, reinforcing its potential as a therapeutic target for multiple myeloma.

CD70-Targeting CAR NK Cells Show Potent Anti-Myeloma Activity In Vitro

We previously showed (19) that CB-derived NK cells engineered to express CAR27/IL-15 (Fig. 3A) displayed enhanced tumor control in multiple preclinical models in vitro and in vivo. In that study, we also established that CAR27/IL-15 NK cells did not mediate cytotoxicity against normal hematopoietic stem cells, T cells, NK cells, or monocytes. Building on this work, we evaluated the anti-multiple myeloma efficacy of CAR27/IL-15 NK cells against CD70-positive (MM.1S) and CD70-negative (NCI-H929) multiple myeloma cell lines (Fig. 3B). Using a chromium-51 (⁵¹Cr) release assay, we compared the cytotoxicity of CAR27/IL-15 NK cells with nontransduced (NT) NK cells at multiple effector-to-target (E:T) ratios, measured after 4 hours of coculture. Both NT and CAR27/IL-15 NK cells effectively lysed NCI-H929 targets, consistent with recognition through their innate receptors. However, against MM.1S cells, CAR27/IL-15 NK cells exerted significantly greater cytotoxicity, confirming their specificity for CD70-positive targets (Fig. 3C and D). These findings were further validated in a long-term real-time cytotoxicity assay using Incucyte live-cell imaging. CAR27/IL-15 NK cells rapidly and effectively eliminated MM.1S cells compared with NT or IL-15–transduced (no CAR) NK cell controls (Fig. 3E and F), with significant reductions in target cell viability observed within a few hours of coculture. Similarly, CAR27/IL-15 NK cells exerted substantially greater cytotoxicity against the U266B1 myeloma cell line (Fig. 3G), which also expresses CD70. Furthermore, flow cytometry and intracellular cytokine staining revealed that CAR27/IL-15 NK cells displayed increased degranulation (CD107a) and higher production of effector cytokines (IFN-γ and TNF-α) when cocultured with MM.1S cells compared with cells cultured alone (Supplementary Fig. S4). These data collectively underscore the robust antimyeloma activity of CD70-targeting CAR NK cells in vitro, driven by their ability to selectively target and kill CD70-expressing myeloma cells.

Figure 3.

Armoring NK cells with CAR27/IL-15 results in enhanced antitumor activity against multiple myeloma. A, Schematic of the retroviral vector used to transduce CB-NK cells. B, Flow cytometry histogram assay of CD70 surface protein levels in NCI-H929 (blue) and MM.1S (magenta) cells. C,⁵¹Cr release assay of NT or CAR27/IL-15 NK cells against CD70⁺ MM.1S or (D) CD70⁻ NCI-H929 tumor cells at different E:T ratios and measured 4 hours after coincubation (n = 3). E, MM.1S tumor cells transduced to express mKATE were cultured alone or with NT, IL-15, or CAR27/IL-15 NK cells. The number of MM.1S cells was measured over time with sequential images from the Incucyte machine. Representative images from 0 Hours (baseline) and after 24 Hours after coculture are shown. White bar, 400 μm. F, The percentage of initial MM.1S cells from panel (E) is plotted for each group over 30 hours. Bar graph (right) shows the AUC analysis. G, The Incucyte killing assay was repeated for NT and CAR27/IL-15 NK cells against the CD70-positive U266B1 myeloma cell line. Bar graph (right) shows the AUC analysis. Data are represented as mean ± SD. Data for the AUC graphs are represented as mean ± SEM. The P values were determined by two-way ANOVA with Sidak’s multiple comparisons (C), one-way ANOVA with Tukey multiple comparison (F), and unpaired t test (G). ***, P ≤ 0.001. ICD, intracellular domain; LTR, long terminal repeats; TMD, transmembrane domain.

In Vivo Activity of CAR27/IL-15 NK Cells in a Multiple Myeloma Mouse Model

To evaluate the in vivo anti-myeloma activity of CD70-targeting CAR NK cells, we used a xenograft mouse model of multiple myeloma engrafted with ffLuc+ MM.1S cells (Fig. 4A). In this model, NOD/SCID/IL2rγ^null (NSG) mice received a single intravenous infusion of MM.1S cells (0.5 × 10⁶ cells/mice) on day −3 to allow tumor development. On day 0, mice received a single intravenous infusion of 3 × 10⁶ CAR27/IL-15 NK cells. Control groups included animals treated with NT, CAR27 (no IL-15), or IL-15 (alone) NK cells (n = 5 mice per group). Tumor progression was monitored longitudinally by bioluminescence imaging (BLI). Although NK cells from the control groups exhibited some tumor control, consistent with their innate ability to target multiple myeloma, rapid tumor progression was still observed. In contrast, mice treated with CAR27/IL-15 NK cells had a marked and sustained reduction in tumor burden (Fig. 4B and C). In addition to controlling tumor growth, treatment with CAR27/IL-15 NK cells resulted in prolonged persistence and significantly improved survival compared with control groups (Fig. 4C–E). Importantly, although the use of NSG mice might not reveal all potential immune-related toxicities, no appreciable toxicity was observed in animals treated with CAR27/IL-15 NK cells based on weight after NK cell infusion (Fig. 4F). These findings highlight the robust anti-multiple myeloma activity and safety of CAR27/IL-15 NK cells in vivo, supporting their therapeutic potential for treating multiple myeloma.

Figure 4.

In vivo anti-myeloma activity of CD70-targeting CAR NK cells. A, Mice were first injected with MM.1S tumor cells and subsequently treated (day 0) with no additional cells (tumor alone), NT NK cells, CAR27/IL-15 NK cells, CAR27 NK cells without IL-15 (CAR27), and NK cells transduced to express only IL-15 (IL-15 NK). B, BLI was obtained weekly until day 54. Different radiance scales were used for early (days 0–14) and late (days 21–54) imaging due to signal intensities spanning several orders of magnitude. An ‘X’ denotes that all mice in the group have died. C, BLI signal for individual mice plotted over days post-CAR NK cell treatment. D, Percentage of human CD45⁺ cells (NK cells) detected in the blood at day 12 after NK cell infusion. Only three of the five mice per group were bled for this analysis. E, Kaplan–Meier survival plot. P value significance is denoted for the following comparisons: blue asterisks: NT NK vs. CAR27/IL-15 NK; green asterisks: CAR27 NK vs. CAR27/IL-15 NK; and magenta asterisks: IL-15 NK vs. CAR27/IL-15 NK. F, Weight of the mice in grams over time. Data are represented as mean ± SD. P values were calculated using the log-rank test (E). **, P ≤ 0.01.

CAR27/IL-15 NK Cells Retain Their Cytotoxicity against Multiple Myeloma Cells with BCMA Loss

Given the eventual relapse of many patients following anti-BCMA therapy, we next investigated whether CAR27/IL-15 NK cells could mediate activity against myeloma cells with BCMA loss. To model BCMA treatment failure, TNFRSF17, the gene encoding BCMA, was knocked out in MM.1S cells using CRISPR/Cas9. The efficiency of TNFRSF17 KO was high (95%) as confirmed by western blotting and flow cytometry (Fig. 5A and B). Importantly, MM.1S TNFRSF17 knockout (KO) cells maintained CD70 expression (Fig. 5C), indicating that the loss of BCMA does not affect CD70 expression and supporting the rationale for CD70 as a therapeutic target in BCMA-resistant disease. In vitro cytotoxicity assays using a range of E:T ratios confirmed that CAR27/IL-15 NK cells effectively killed MM.1S TNFRSF17 KO cells compared with NT NK cells (Fig. 5D). These findings were further validated in a long-term Incucyte cytotoxicity assay, in which CAR27/IL-15 NK cells consistently showed robust killing of MM.1S TNFRSF17 KO cells compared with NT or IL-15 NK cell control counterparts (Fig. 5E).

Figure 5.

Loss of BCMA does not affect the anti-myeloma activity of CAR27/IL-15 NK cells. A, Whole-cell lysates from K562 (negative control), MM.1S wild-type (WT), and MM.1S TNFRSF17 KO cells (KO) were analyzed by Western blotting for BCMA expression. β-actin served as the loading control. A representative blot is shown. B,TNFRSF17 KO efficiency analyzed by flow cytometry. C, CD70 expression of MM.1S WT and MM.1S TNFRSF17 KO tumor cells. D,⁵¹Cr release assay of NT NK cells or CAR27/IL-15 NK cells against MM.1S TNFRSF17 KO cells. E, Incucyte real-time killing assay of MM.1S TNFRSF17 KO tumor cells by NT NK, IL-15 NK, and CAR27/IL-15 NK cells. Bar graph (right) shows the AUC analysis. Data are represented as mean ± SD. Data for the AUC graph are represented as mean ± SEM. The P values were determined by two-way ANOVA with Sidak’s multiple comparisons (D) and one-way ANOVA with Tukey multiple comparisons (E). ***, P ≤ 0.001.

To confirm these findings in vivo, NSG mice were engrafted with ffLuc+ MM.1S TNFRSF17 KO and, 3 days later, were either left untreated or treated with 3 × 10⁶ NT, CAR27/IL-15, CAR27 (no IL-15), or IL-15 NK cells (Fig. 6A). BLI was used to measure tumor burden. Mice treated with CAR27/IL-15 NK cells showed markedly reduced tumor burden by BLI compared with untreated or control NK cell groups (Fig. 6B and C). Treatment also resulted in prolonged persistence, with CAR27/IL-15 NK cells detectable in both blood (days 12 and 27) and bone marrow at day 27, correlating with lower tumor burden (Fig. 6D; Supplementary Fig. S5A and S5B). Survival was significantly extended, with three of five mice alive at day 60 and no evidence of toxicity (Fig. 6E and F). These findings underscore the efficacy of CD70-targeting CAR NK cells against myeloma, even in the absence of BCMA expression, highlighting their potential as a therapeutic option for patients who relapse or progress following BCMA-directed therapies.

Figure 6.

CAR27/IL-15 NK cells improve tumor control in a MM.1S TNFRSF17 KO mouse model. A, Mice were first injected with MM.1S TNFRSF17 KO tumor cells on day −3. The mice were subsequently treated (day 0) with no additional cells (tumor alone), NT NK, CAR27/IL-15 NK, CAR27 NK, and IL-15 NK cells. B, BLI was obtained weekly until day 54. Different radiance scales were used for early (days 0–14) and late imaging due to signal intensities spanning several orders of magnitude. An ‘X’ denotes that all mice in the group have died. C, BLI signal for individual mice plotted over days post-CAR NK cell treatment. D, Percentage of human CD45⁺ cells (NK cells) detected in the blood at day 12 after NK cell infusion. Only three of the five mice per group were bled for this analysis. E, Kaplan–Meier survival plot. Blue asterisks: NT NK vs. CAR27/IL-15 NK; green asterisk: CAR27 NK vs. CAR27/IL-15 NK; and magenta asterisk: IL-15 NK vs. CAR27/IL-15 NK. F, Weight of the mice over time. Data are represented as mean ± SD. P values were calculated using the log-rank test (E). *, P ≤ 0.05; **, P ≤ 0.01.

To investigate mechanisms of relapse in mice with residual disease, we sacrificed animals and analyzed surviving MM.1S cells by flow cytometry. Compared with untreated controls, MM.1S cells persisting after CAR27/IL-15 NK treatment displayed marked downregulation of CD70 expression, suggesting antigen loss as a potential mechanism of escape (Supplementary Fig. S5C).

To benchmark against T cell–based approaches, we generated CAR27/IL-15 NK and T cells from the same CB donor. In vitro, both cell types displayed potent and comparable cytotoxicity against MM.1S cells, outperforming their NT counterparts (Fig. 7A). Notably, NT NK cells demonstrated superior anti-myeloma activity compared with NT T cells, reflecting their innate anti-myeloma activity. In an MM.1S xenograft model, both CAR27/IL-15 NK and T cells significantly reduced tumor burden and improved survival compared with controls, with no weight loss or signs of toxicity (Fig. 7B–E). Finally, in the MM.1S TNFRSF17 KO xenograft model, CAR27/IL-15 NK and T cells again exhibited equivalent tumor control, confirming that CD70 targeting provides effective anti-myeloma activity even in the absence of BCMA (Supplementary Fig. S6A–S6D). Together, these findings demonstrate that CAR27/IL-15 NK cells selectively and effectively target CD70-positive myeloma, retain activity against BCMA-negative disease, and perform comparably with CAR T cells, supporting their potential role as an alternative or complementary therapy for patients who relapse after BCMA-directed treatment.

Figure 7.

CAR27/IL-15 NK cells display comparable anti-multiple myeloma activity to CAR27/IL-15 T cells. A, Cytotoxicity of NT NK, CAR27/IL-15 NK, NT T, and CAR27/IL-15 T cells against MM.1S tumor cells was assessed using an Incucyte assay. Bar graph (right) shows AUC analysis. B, NSG mice were injected with MM.1S tumor cells on day −3 and treated on day 0 with CAR27/IL-15 NK cells or CAR27/IL-15 T cells or left untreated (tumor alone). C, Tumor burden was monitored by BLI through day 34. D, BLI signal for individual mice plotted over time. E, Mouse body weight was measured during the study as an indicator of treatment tolerability. Data are represented as mean ± SD. Data for the AUC graph are represented as mean ± SEM. The P values were determined by one-way ANOVA with Tukey multiple comparison (A). ***, P ≤ 0.001.

Discussion

Advances in multiple myeloma treatment, including CD38 antibodies, immunomodulatory drugs, and proteasome inhibitors, have significantly improved patient outcomes, with overall survival exceeding a decade for many. However, patients with high-risk cytogenetics continue to experience earlier relapses and achieve less durable responses compared with those with standard-risk disease, even with intensive approaches such as autologous stem cell transplantation (20), maintenance treatment (21), or quadruplet induction regimens (22). As a consequence, these patients often exhaust treatment options more rapidly. Cellular therapies, including BCMA–CAR T cells such as ciltacabtagene autoleucel, have demonstrated remarkable efficacy in these patients, yet relapse remains common (23), underscoring the need for additional therapeutic strategies.

Currently, there is no standardized treatment for patients with multiple myeloma who relapse following BCMA-directed therapy. In this study, we identified CD70 as a promising therapeutic target for multiple myeloma particularly in patients with high-risk disease or those who have progressed after BCMA-directed treatments. CD70 is expressed in a number of solid (24, 25) and hematologic cancers (26) but is largely absent from normal tissue, making it a clinically safe target. A CD70-targeting antibody in clinical development demonstrated minimal drug-related side effects, further supporting the feasibility of targeting this antigen (27).

Although CD70 has not been extensively studied as a therapeutic target in multiple myeloma, prior staining studies reported its expression in 42% of myeloma samples (28). Expanding on this, our analysis of the CoMMpass dataset revealed that CD70 is broadly expressed in myeloma cells, with particularly elevated levels in patients with t(4;14) translocation, amplification of 1q21, del17p, ISS stage 2 or 3, and relapsed patients. Furthermore, CD70 expression was strongly associated with poorer survival, with patients in the highest CD70 expression tertile experiencing significantly shorter survival than those in the lowest tertile.

We validated these findings in our own dataset at MD Anderson Cancer Center. scRNA-seq confirmed that CD70 mRNA expression is higher in myeloma cells that harbor the t(4;14) translocation. Flow cytometry analysis of 30 sequential bone marrow aspirates from patients with multiple myeloma demonstrated that CD70 protein expression is highly expressed in myeloma cells. IHC further demonstrated CD70 expression in high-risk and relapsed disease, including in a patient who progressed on belantamab mafodotin, a BCMA antibody–drug conjugate. Interestingly, the elevated CD70 expression in high-risk multiple myeloma [e.g., t(4;14) cases] may be a downstream effect of NSD2-mediated transcriptional and epigenetic reprogramming (29).

NK cells offer several advantages for multiple myeloma treatment due to their unique receptor profile and mechanism of action. NK cells express activating receptors such as NKG2D and DNAM-1, which recognize stress ligands on tumor cells, as well as CD16, which mediates antibody-dependent cellular cytotoxicity (30). These receptors allow NK cells to efficiently target myeloma cells, especially in the presence of mAbs such as daratumumab (anti-CD38) and elotuzumab (anti-SLAMF7; refs. 31, 32). Furthermore, NK cells represent an attractive source of cells for cellular therapy. They have demonstrated minimal toxicity, such as cytokine release syndrome or immune effector cell–associated neurotoxicity syndrome, and can be used as allogeneic, “off-the-shelf” therapies due to the absence of GVHD, even when HLA mismatched (10, 11). This makes them particularly advantageous for patients with high-risk disease who cannot wait for the manufacturing of conventional autologous CAR T-cell products.

In this study, we generated CAR27/IL-15 CAR NK cells and demonstrated their enhanced killing ability in vitro, which led to superior tumor control and prolonged survival in mice. To model resistance to BCMA-directed therapies, we generated TNFRSF17 KO myeloma cell lines and confirmed that they retained CD70 expression and remained sensitive to CAR27/IL-15 NK cell–mediated killing. Importantly, direct comparisons showed that CAR27/IL-15 NK cells achieved tumor control comparable with CAR T cells, underscoring their potential as an alternative or complementary therapeutic strategy, particularly in patients relapsing after BCMA-directed therapy. Although these findings highlight the promise of CD70 CAR NK cells, additional studies are warranted to define mechanisms of immune escape and optimize strategies for durable disease control.

Thus, CD70 represents an attractive target for patients with multiple myeloma who have progressed on BCMA-directed therapies. Based on these findings, we have initiated a phase I/II clinical trial to evaluate the safety and efficacy of CD70-targeting CAR NK cells in patients with CD70-positive multiple myeloma (Clinical Trial # NCT05092451).

Methods

Analysis of CoMMPass Dataset

CD70 expression in multiple myeloma was analyzed using the MMRF CoMMpass dataset. This comprehensive dataset includes genomic and transcriptomic profiling of hundreds of patients with NDMM and relapsed multiple myeloma. Tumor-enriched samples were analyzed using long-insert whole-genome sequencing, whole-exome sequencing, and RNA sequencing. Longitudinal survival data were collected, and when available, additional samples for whole-genome sequencing, whole-exome sequencing, and RNA sequencing were obtained at the time of progression. The CoMMpass dataset is publicly accessible through the MMRF Researcher Gateway (https://research.themmrf.org).

All genomic profiling and patient metadata were extracted from the MMRF CoMMpass study portal (https://research.themmrf.org/, version IA21). Transcriptomic expression data were processed by importing the STAR gene count matrix into the R package DESeq2 and normalized using the variance-stabilizing transformation algorithm. Specific translocations, including t(4;14), t(6;14), t(11;14), t(14;16), and t(14;20), were identified using NSD2_CALL, CCND3_CALL, CCND1_CALL, MAF_CALL, and MAFB_CALL, respectively, from the igtx_pairoscope file. Genomic alterations del17p and amp1q were determined by the SeqExome_Cp_17p13_20% and SeqExome_Cp_1q21_20% calls according to the seqFISH algorithm.

Patient demographics (i.e., gender, race/ethnicity, disease stage) were extracted from the patient metadata file, D_PT_gender, D_PT_race, DEMOG_WHITE, DEMOG_BLACKORAFRICA, DEMOG_ASIAN, and DEMOG_OTHER. Disease staging information was retrieved from the D_PT_iss variable to stratify patients by ISS stage.

For survival analysis, the overall survival data were derived from the ttcosw and censos variables and fitted using the survfit2() function of the R package ggsurvfit. All visualizations and statistical tests, including survival curves, were conducted using the R packages ggpubr, ggplot2, and ggsurvfit.

Analysis of scRNA-seq Data

Bone marrow aspirates were collected from healthy donors (n = 3) and patients with NDMM (n = 20) and RRMM (n = 26) under protocols approved by the Institutional Review Board. Written informed consent was obtained from all participants in accordance with the Declaration of Helsinki. CD138⁺ plasma cells from healthy donors and CD138⁺ myeloma cells from patients with multiple myeloma were isolated using the EasySep CD138-positive selection kit (STEMCELL Technologies) and processed as described previously (33). 5′ Gene expression and B-cell receptor VDJ (variable, diversity, joining) libraries were prepared using the 10x Genomics platform versions 1.0 and 1.1. Data analysis was performed as detailed previously (34). Briefly, FASTQ files were aligned, and unique molecular identifier counting was performed using Cell Ranger (version 3.1.0; ref. 35). Datasets from all samples were merged, normalized, and scaled using Seurat (version 4.3.0; ref. 36). VDJ sequences were used to distinguish between monoclonal and polyclonal plasma cells within each sample. Cells sharing identical heavy and/or light VDJ sequences were classified as monoclonal plasma cells, whereas those with low-frequency VDJ sequences (<0.1%) were classified as polyclonal plasma cells. Differential gene expression analysis was conducted using a pseudobulk approach by DESeq2 (version 1.34; ref. 37).

Flow Cytometry Analysis

Sequential bone marrow aspirates were collected from patients and processed for flow cytometry analysis. Red blood cells were lysed (BD Biosciences, cat. #555899), and the remaining cells were washed and stained with a viability dye (Tonbo, Ghost UV450, cat. #13-0863-T500). After a second wash, samples were incubated with an Fc blocking antibody (BioLegend, cat. #422302, RRID: AB_2818986) to minimize nonspecific binding. An antibody mixture was added, and samples were incubated at room temperature for 20 minutes. Following incubation, cells were washed, resuspended in 2% paraformaldehyde, and stored at 4°C until analysis. Details of the antibodies used for flow cytometry are described in Supplementary Table S1.

Multiparametric data were acquired using the ID7000 spectral flow cytometer (Sony) and analyzed with the Kaluza software (Beckman Coulter). Dot plots were generated using Morpheus (https://software.broadinstitute.org/morpheus), with a robust z-score applied to create a color gradient for data visualization. The optimized stochastic neighbor embedding plots were generated using OMIQ (Dotmatics) to facilitate dimensionality reduction and clustering analysis.

IHC Staining

Tissues were fixed in 10% formalin, paraffin-embedded, and sectioned at 4 μm. Bone samples were first decalcified with formic acid. Sections were then stained using the Leica BOND-RX automated stainer (Leica Biosystems). Antigen retrieval was performed with BOND ER solution #2 for 20 minutes at 100°C (ER2 pH 9.0 from Leica Biosystems AR9640). The CD70 antibody (Cell Signaling Technology, clone e3Q1A, cat. #69209S) was used at a 1:50 dilution, incubated for 15 minutes at room temperature, and subsequently detected using the BOND Polymer Refine Detection kit (Leica Biosystems, cat. #DS9800) with diaminobenzidine as the chromogen and counterstained with hematoxylin. Sections from a patient with newly diagnosed classic Hodgkin’s lymphoma with no detectable disease in the bone marrow served as a negative control.

Intracellular Cytokine Staining

CAR27/IL-15 NK cells were cultured alone or cocultured with MM.1S cells at a 1:1 E:T ratio for 6 hours at 37°C. Monensin (BioLegend, cat. #420701), brefeldin A (BioLegend, cat. #420601), and CD107a antibody were added at the time of culture. After coculture, cells were washed and stained for CD45, CD138, IFN-γ, and TNF-α in PBS containing 2% FBS. Fixation and permeabilization were performed using a Fixation/Permeabilization Kit (BD Cytofix/Cytoperm, cat. #554714, AB_2869008) according to the manufacturer’s instructions. Following incubation, cells were washed, resuspended in 2% paraformaldehyde, and stored at 4°C until analysis. Details of the antibodies used are described in Supplementary Table S1.

Cell Lines and Culture Conditions

Human myeloma cell lines MM.1S (cat. #CRL-2974, RRID: CVCL_8792) and U266B1 (cat. #TIB-196, RRID: CVCL_0566) were obtained from the ATCC. NCI-H929 human myeloma cell lines (ATCC, cat. #CRL-3580, RRID: CVCL_1600) were provided by R.Z. Orlowski, MD Anderson Cancer Center. K562 cells (cat. #CCL-243, RRID: CVCL_0004) were obtained from the ATCC. MM.1S, U266B1, and NCI-H929 cells were cultured in RPMI 1640 medium supplemented with 10% FBS, 1% GlutaMAX, and 1% penicillin/streptomycin (Pen/Strep).

Both MM.1S and MM.1S TNFRSF17 KO cells were further engineered to express firefly luciferase for in vivo bioluminescence detection. All cell lines were authenticated with short tandem repeat DNA fingerprinting by the Cytogenetics and Cell Authentication Core at MD Anderson Cancer Center. All cell lines were tested for Mycoplasma contamination using the MycoAlert Mycoplasma Detection Kit (Lonza Bioscience), and all samples consistently tested negative.

Generating TNFRSF17 KO Tumor Cells

TNFRSF17 was knocked out of MM.1S cells using CRISPR/Cas9. First, crRNA (BCMA from IDT) and tracrRNA (IDT, cat. #1072533) were annealed together in a thermocycler. Cas9 (IDT, cat. #1081059) was added to the complex and incubated at room temperature. The subsequent complex was mixed with MM.1S and electroporated using the Neon electroporation system (Invitrogen) at 1,600 V, 10 ms, 3 pulses. Cells were placed in culture for 2 weeks. Remaining BCMA-positive MM.1S cells were subsequently removed by negative selection. The cells were first incubated with an anti-BCMA biotin antibody (BioLegend, cat. #357514, RRID: AB_2629517), washed, and incubated with anti-biotin microbeads (Miltenyi Biotec, cat. #130-090-485, RRID: AB_244365). After incubation, the cells were placed over an LD Column (Miltenyi Biotec, cat. #130-042-901), and the MM.1S TNFRSF17 KO cells were collected as the eluate.

Western Blotting

Cells were lysed using IP Lysis Buffer (Thermo Fisher Scientific, cat. #87788) supplemented with protease inhibitors (Thermo Fisher Scientific, cat. #78443) and incubated on ice for 30 minutes. Equal amounts of whole-cell lysates were resolved by SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Bio-Rad). Membranes were blocked with 5% milk in PBST for 30 minutes, followed by overnight incubation at 4°C with primary antibodies: TNFRSF17/BCMA (Thermo Fisher Scientific, cat. #MA5-31673, RRID: AB_2787297) and β-actin (Sigma-Aldrich, cat. #A5441, RRID:AB_476744). After three washes with PBST (5 minutes each), membranes were incubated with horseradish peroxidase–conjugated sheep anti-mouse IgG secondary antibody (Genesee Scientific, cat. #84-848) for 1 hour at room temperature. Protein signals were detected using enhanced chemiluminescence substrate (Amersham) according to the manufacturer’s instructions.

NK Cell Expansion and CAR Transduction

NK cells were isolated from CB mononuclear cells using a negative selection protocol with the NK cell isolation kit (Miltenyi Biotec, cat. #130-092-657) according to the manufacturer’s instructions. Glycophorin A microbeads (Miltenyi Biotec, cat. #130-050-501, RRID:AB_3076160) were added to ensure the removal of residual red blood cells. Isolated NK cells were activated with a cytokine cocktail containing IL-12 (10 ng/mL), IL-15 (50 ng/mL), and IL-18 (50 ng/mL) for 16 hours as described previously (38). Following activation, the cells were cultured with universal antigen-presenting cells in Click’s/RPMI media (50:50) supplemented with 10% FBS, 1% Pen/Strep, 1% GlutaMAX, and IL-2 (Proleukin, 200 U/mL; ref. 39).

For CAR transduction, NK cells were transduced on day 5 after isolation from CB with a retroviral construct encoding CAR27/IL-15, which includes the CD27 extracellular domain that binds CD70 linked to the CD28 transmembrane domain, a CD28 costimulatory domain, a CD3ζ signaling domain, and IL-15, as shown in Fig. 3A (19). Transduction was performed as previously described (40). Control constructs, including CAR27 without IL-15 and IL-15 alone, were generated and transduced under the same conditions for comparative analyses.

T-cell Expansion and CAR Transduction

For experiments involving both NK and T cells from the same CB, T cells were first isolated from CB mononuclear cells using a Pan T Cell Isolation Kit (Miltenyi Biotec, cat. #130-096-535). The remaining cells were subsequently processed for NK cell isolation as described above. The isolated T cells were stimulated with CD3/CD28 beads (Gibco Dynabeads, cat. #11131D) and cultured in RPMI media supplemented with 10% FBS, 1% Pen/Strep, 1% GlutaMAX, and IL-2 (Proleukin, 50 U/mL). Two days later, T cells were transduced using the same protocol described above for NK cells. Details of the antibodies used for these experiments are provided in Supplementary Table S2.

NK Cell Cytotoxicity Assays

Chromium release assay and Incucyte real-time assay were used to assess NK cell cytotoxicity. In the chromium release assay, NT or CAR27/IL-15 NK cells were incubated with ⁵¹Cr -labeled MM.1S (CD70⁺) or NCI-H929 (CD70⁻ control) cells at different E:T ratios in V-bottom plates. Cytotoxicity was determined by measuring the amount of ⁵¹Cr released after a 4-hour incubation (41). To evaluate NK cell cytotoxicity over time, an Incucyte live-cell analysis was performed. NT, CAR27/IL-15, CAR27, or IL-15 NK cells were cocultured with either MM.1S or U266B1 at a 1:1 E:T ratio in flat-bottom 96-well plates, with each well containing 100,000 NK cells and 100,000 myeloma cells. MM.1S cells were transduced with the Nuclight mKate lentivirus (Sartorius, cat. #4625) as per the manufacturer’s instructions. U266B1 cells were prelabeled with GFP. Four images per well were captured with a 10× objective using the Sartorius IncuCyte S3 Live-Cell Analysis System (RRID: SCR_023147) at defined time intervals. The number of fluorescent cells was counted in each well, and percentage cytotoxicity was calculated by dividing the number of fluorescent cells at each time point by the number of fluorescent cells at time zero.

Multiple Myeloma Mouse Model

Female NSG mice, ages 10 to 12 weeks (The Jackson Laboratory, RRID: IMSR_JAX:005557), were used to establish the myeloma xenograft model. Mice were irradiated with 300 cGy total body irradiation to facilitate tumor engraftment, as shown in Figs. 4A, 6A, and 7B and Supplementary Fig. S6A. The following day (day −3), 0.5 × 10⁶ firefly luciferase–labeled MM.1S or MM.1S TNFRSF17 KO cells were administered intravenously by tail vein injection. On day 0, mice received either 3 × 10⁶ NK cells or 1 × 10⁶ T cells (experimental groups) or no NK cells (tumor alone controls) by intravenous infusion. Mice were monitored for tumor progression by weekly BLI using the IVIS 200 Imaging system (PerkinElmer) or Ami HT (Spectral Instruments Imaging) and were weighed three times per week. All animal experiments were performed in accordance with NIH recommendations under protocols approved by the Institutional Animal Care and Use Committee.

Statistical Analysis

All statistical analyses were performed using GraphPad Prism version 10.0.3 (RRID: SCR_002798). For comparisons involving multiple groups, data were analyzed by one-way ANOVA, two-way ANOVA, or Wilcoxon signed-rank test, as appropriate. Two-group comparisons were conducted using unpaired t tests. Differences were considered statistically significant at *, P < 0.05; **, P < 0.01; and ***, P < 0.001. Data are presented as the mean ± SD or mean ± SEM. Overall survival was analyzed by Kaplan–Meier estimates with log-rank tests and 95% confidence intervals. The specific statistical tests used are indicated in each figure legend.

Supplementary Material

Table S1

Supplementary Table S1. List of antibodies used for flow cytometry and intracellular cytokine staining.

Table S2

Supplementary Table S2. Additional antibodies used for CAR-NK vs CAR-T cell mouse experiments.

Supplementary Fig S1

Supplementary Fig. S1. CD70 mRNA expression of patients with multiple myeloma regarding sex and race.

Supplementary Fig S2

Supplementary Fig. S2. Expression of NSD2 and CD70 overlap.

Supplementary Fig S3

Supplementary Fig. S3. Flow cytometry gating strategy.

Supplementary Fig S4

Supplementary Fig. S4. Flow cytometry and intracellular cytokine staining of CAR27/IL15 NK cells co-cultured with or without MM.1S target cells.

Supplementary Fig S5

Supplementary Fig. S5. In vivo persistence of CAR27/IL-15 NK cells in the blood and bone marrow.

Supplementary Fig S6

Supplementary Fig. S6. In vivo activity of CAR27/IL-15 NK and CAR27/IL-15 T cells against BCMA-negative (TNFRSF17 KO) MM.1S cells.

Data Availability

The CoMMpass dataset analyzed in this study was obtained from the MMRF and is publicly accessible through the MMRF Researcher Gateway (https://research.themmrf.org). The scRNA-seq data analyzed in this study were obtained from the European Genome-phenome Archive at https://ega-archive.org/dacs/EGAC50000000271. All other data generated in the study are available from the corresponding author upon reasonable request.

Authors’ Disclosures

Pa. Lin reports other support from Takeda Pharmaceutical outside the submitted work. S. Acharya reports other support from Takeda Pharmaceutical outside the submitted work and a patent for the CD27-extracellular domain CAR to target CD70-positive tumors pending. R. Basar reports other support from Takeda and Affimed outside the submitted work. N. Uprety reports personal fees from Takeda outside the submitted work and a patent for CAR27/IL-15 pending. P.P. Banerjee reports other support from Takeda Pharmaceutical outside the submitted work and that he and the University of Texas MD Anderson Cancer Center have an institutional financial conflict of interest with Takeda Pharmaceutical for the licensing of the technology related to CAR NK cell research. MD Anderson has implemented an Institutional Conflict of Interest Management and Monitoring Plan to manage and monitor the conflict of interest with respect to MD Anderson Cancer Center’s conduct of any other ongoing or future research related to this relationship. M. Daher reports personal fees from Takeda Pharmaceutical outside the submitted work, a patent for CAR27/IL-15 NK cells for cancer immunotherapy pending, and personal fees from Cellsbin, Aurigene and Bruker Cellular Analysis outside the submitted work. M.H. Qazilbash reports grants from Johnson & Johnson and Sanofi/Genzyme outside the submitted work. K.K. Patel reports personal fees from Johnson & Johnson, Bristol Myers Squibb, AstraZeneca, Legend Biotech, Caribou Biosciences, Poseida Therapeutics, Kite Pharma, Arcellx, and Oricel outside the submitted work, as well as grants from Pfizer, Takeda Pharmaceutical, Sanofi, Genentech, and AbbVie. H.C. Lee reports grants and personal fees from Bristol Myers Squibb, Takeda Pharmaceutical, Alexion Pharmaceuticals, Janssen, Regeneron Pharmaceuticals, Menarini, and GlaxoSmithKline; personal fees and other support from Allogene and Takeda Pharmaceutical; and personal fees from Pfizer, and Sanofi outside the submitted work. D. Marin reports a patent for CD70 CAR NK pending and support from Takeda and Affimed outside of the submitted work. E.J. Shpall reports other support from NY Blood Center, Adaptimmune, Navan, Celaid Therapeutics, Zelluna Immunotherapy, FibroBiologics, Axio, and Orca Biosystems outside the submitted work and patents for Takeda Pharmaceutical, Affimed, and Prana X licensed to a license agreement. R.Z. Orlowski reports support from AbbVie (advisory board), Adaptive Biotech (advisory board), Asylia Therapeutics Inc. (advisory board), Biotheryx (advisory board), Bristol Myers Squibb Pharmaceuticals (advisory board), DEM BioPharma Inc. (advisory board), Karyopharm Therapeutics (advisory board), Lytica Therapeutics (advisory board), Meridian Therapeutics (advisory board), Monte Rosa Therapeutics (advisory board), myeloma360 (advisory board), Nanjing IASO Biotherapeutics (advisory board), Neoleukin Corporation (advisory board), Oncopeptides AB (advisory board), Pfizer Inc. (advisory board), Regeneron Pharmaceuticals Inc. (advisory board), Sanofi (advisory board), and Takeda Pharmaceuticals (advisory board). K. Rezvani reports personal fees from AvengeBio, Virogin Biotech, NAVAN Technologies, Caribou Biosciences, Bit Bio Limited, Replay Holdings, oNKo-Innate, The Alliance for Cancer Gene Therapy, Innate Pharma, and Shinobi Therapeutics and nonfinancial support from Syena Therapeutics outside the submitted work and patents for Takeda Pharmaceutical and Affimed licensed and with royalties paid. No disclosures were reported by the other authors.

Authors’ Contributions

Pa. Lin: Conceptualization, data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, writing–review and editing. S. Acharya: Formal analysis, investigation. F. Reyes-Silva: Data curation, formal analysis, investigation. R. Basar: Formal analysis, investigation. N. Uprety: Investigation. L.Y. Moreno Rueda: Data curation, formal analysis, validation, investigation. Pe. Lin: Formal analysis, investigation. A.L. Gilbert: Formal analysis, investigation. P.P. Banerjee: Data curation, formal analysis, investigation. D. Fang: Formal analysis, investigation. C. Zhang: Data curation, formal analysis, investigation. A.K. Nunez Cortes: Formal analysis, investigation. L. Melo Garcia: Formal analysis, investigation. M. Daher: Formal analysis, investigation. L. Muniz-Feliciano: Writing–original draft, writing–review and editing. G.M. Deyter: Writing–original draft, writing–review and editing. V. Woods: Formal analysis, investigation. S. Rawal: Formal analysis, investigation. P. Li: Investigation. C.M. Jones: Formal analysis, investigation. R. Shrestha: Formal analysis, investigation. M.H. Qazilbash: Resources, formal analysis. K.K. Patel: Resources, formal analysis. H.C. Lee: Resources, formal analysis. R.E. Champlin: Resources, formal analysis. D. Marin: Resources, formal analysis. E.J. Shpall: Resources, formal analysis. R.Z. Orlowski: Resources, formal analysis, funding acquisition, methodology. K. Rezvani: Conceptualization, supervision, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing.