If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Challenges in αCD38-chimeric antigen receptor (CAR)-expressing natural killer (NK) cell-based immunotherapy in multiple myeloma: Harnessing the CD38dim phenotype of cytokine-stimulated NK cells as a strategy to prevent fratricide
Correspondence: Evren Alici, Center for Hematology and Regenerative Medicine, Department of Medicine-Huddinge, Karolinska Institutet, Alici group, ANA Futura Alfred Nobels allé 8, 14152, Huddinge, Stockholm, Sweden.
Adoptive cell therapy with chimeric antigen receptor (CAR)-expressing natural killer (NK) cells is an emerging approach that holds promise in multiple myeloma (MM). However, the generation of CAR-NK cells targeting CD38 is met with obstacles due to the expression of CD38 on NK cells. Knock-out of CD38 is currently explored as a strategy, although the consequences of the lack of CD38 expression with regards to engraftment and activity in the bone marrow microenvironment are not fully elucidated. Here, we present an alternative approach by harnessing the CD38dim phenotype occurring during long-term cytokine stimulation of primary NK cells.
Methods
Primary NK cells were expanded from peripheral blood mononuclear cells by long-term IL-2 stimulation. During expansion, the CD38 expression was monitored in order to identify a time point when introduction of a novel affinity-optimized αCD38-CAR confered optimal viability, i.e. prevented fratricide. CD38dim NK cells were trasduced with retroviral vectors encoding for the CAR trasngene and their functionality was assessed in in vitro activation and cytotoxicity assays.
Results
We verified the functionality of the αCD38-CAR-NK cells against CD38+ cell lines and primary MM cells. Importantly, we demonstrated that αCD38-CAR-NK cells derived from patients with MM have increased activity against autologous MM samples ex vivo.
Conclusions
Overall, our results highlight that incorporation of a functional αCD38-CAR construct into a suitable NK-cell expansion and activation protocol results in a potent and feasible immunotherapeutic strategy for the treatment of patients with MM.
Multiple myeloma (MM) is a highly heterogenous plasma cell malignancy representing 10% of all hematological cancers and approximately 1% of all malignant diseases [
], MM is still considered an incurable disease due to the high relapse rate. Lately, the treatment paradigm has changed due to the identification of molecules with differential expression on MM cells, compared with physiological cells, that act as promising immunotherapeutic targets. Examples of these molecules are CD38, CD138 (SYND1), B-cell maturation antigen (BCMA), and signaling lymphocyte activation molecule F7 (SLAMF7), G protein–coupled receptor, class C, group 5, member D (GPRC5D) and FcRH5 [
CD38 is a multifunctional transmembrane glycoprotein that plays a key role in cell metabolism, catalyzing the hydrolysis of NAD+ and NADP, and the synthesis of the second messengers cyclic adenosine diphosphate ribose, adenosine diphosphate ribose and nicotinic acid adenine dinucleotide phosphate, which are potent Ca+2-mobilizing compounds [
CD38, a receptor with multifunctional activities: from modulatory functions on regulatory cell subsets and extracellular vesicles, to a target for therapeutic strategies.
]. Under physiological conditions, CD38 is ubiquitously expressed on the surface of cells of myeloid and lymphoid lineage, including plasma cells, natural killer (NK) cells, and B and T cells [
], CD38 is found upregulated in a significant number of patients with acute myeloid leukemia, B-cell chronic lymphoblastic leukemia and other disorders of hematopoietic origin. Daratumumab [
] are two regulatory-approved human IgG1κ monoclonal antibodies that trigger anti-MM immune responses via targeting the CD38 protein. Both therapeutic antibodies achieved impressive response rates in patients with relapsed/refractory MM during clinical trials, having limited adverse effects that derived mainly from the elimination of CD38+ cell populations and the subsequent enhanced risk for infectious complications [
]. CD38-targeting antibodies are now established as standard of care for newly diagnosed, transplant-eligible and transplant-ineligible patients, where they are administered either as single-agents or in combination with proteasome inhibitors and immunomodulatory agents [
Daratumumab, lenalidomide, and dexamethasone versus lenalidomide and dexamethasone alone in newly diagnosed multiple myeloma (MAIA): overall survival results from a randomised, open-label, phase 3 trial.
Bortezomib, thalidomide, and dexamethasone with or without daratumumab before and after autologous stem-cell transplantation for newly diagnosed multiple myeloma (CASSIOPEIA): a randomised, open-label, phase 3 study.
Impact of performance status on overall survival in patients with relapsed and/or refractory multiple myeloma: real-life outcomes of daratumumab treatment.
Specific-antigen targeting can also be achieved via the adoptive transfer of genetically modified effector cells, T or NK, expressing chimeric antigen receptors (CARs) [
]. The potential of this approach in MM was highlighted by the recent Food and Drug Administration approvals of two αBCMA-CAR-T therapy products, namely idecabtagene vicleucel and ciltacabtagene autoleucel [
]. CD38-targeting CAR-T cell therapy has also been investigated, although yet only at a preclinical level. Specifically, a fully human second-generation affinity-optimized αCD38-CAR construct was recently introduced to donor and patient T cells, demonstrating increased elimination of CD38+ targets in vitro and in vivo [
]. Adoptive CAR-T cell therapy has shown significant clinical benefit, although it is also frequently associated with life-threatening side effects, such as cytokine release syndrome and neurotoxicity [
]. NK cells can serve as an alternative effector cell in CAR-based therapy. They are susceptible to genetic engineering and have an overall safer profile, as NK cell infusions have been reported to reduce graft-versus-host–related complications [
]. In addition, alloreactivity of NK cells is enhanced under KIR receptor–HLA ligand mismatch, which further attenuates their efficacy under allogeneic conditions and reinforces the need for development of readily available and cost-effective allogeneic off-the-shelf products [
]. Regarding CD38-targeting treatments, NK cells have the additional advantage of being able to exert cytotoxicity without specific antigen recognition and are, thus, suitable in case of CD38 downregulation during the course of the disease.
Despite the theoretical appeal, the combination of αCD38-targeting antibodies and NK-cell therapy is limited by the intrinsic expression of CD38 on NK cells, which could trigger fratricide. Strategies are, therefore, focusing on either knocking-out (KO) CD38 from NK [
CD38-specific chimeric antigen receptor expressing natural killer KHYG-1 cells: a proof of concept for an "off the shelf" therapy for multiple myeloma.
CD38-specific chimeric antigen receptor expressing natural killer KHYG-1 cells: a proof of concept for an "off the shelf" therapy for multiple myeloma.
], or monoclonal antibodies. Specifically in the context of αCD38-CAR-NK therapy, however, the clinical applicability of the assessed cell sources (CRISPR-edited NK cells and NK-cell lines) is questioned. For instance, the use of immortal cell lines carries safety concerns, even after their uncontrolled proliferation is prevented by irradiation. Similarly, KO strategies potentially result in unpredictable genetic alterations due to off-target events. It is therefore clear that although αCD38-CAR-NK therapy is a promising immunotherapeutic strategy with potential in treating MM, the therapy is not yet adequately feasible.
Here, we aim to extend the feasibility of αCD38-CAR-NK therapy by providing an alternative CD38dim primary NK-cell source. As it has been previously shown, different NK-cell expansion protocols can lead to different phenotypic profiles on the generated cells [
] and establish a strategy to generate functional CD38dim CAR-NK cells by exploring the fine line between optimal transduction efficacy and minimal fratricide events. We then propose an autologous NK-cell approach to further increase safety and assess its feasibility in in vitro functional studies.
Materials and Methods
Primary cells
Peripheral blood mononuclear cells (PBMCs) were obtained from buffy coats of anonymous healthy donors in accordance with institutional and national guidelines. Ethical permits were granted by the Swedish Ethical Review Board for research on patient-derived material. To summarize, patient PBMCs and bone marrow (BM)-derived mononuclear cells (MNCs) were isolated from PB and BM aspirates respectively, by Ficoll-gradient separation using LymphoPrep (STEMCELL Technologies, Vancouver, British Columbia, Canada), according to standard procedures. Magnetic separation of the malignant CD138+ fraction from the MNC samples was performed using MACSprep Multiple Myeloma CD138 MicroBeads (Miltenyi Biotec, San Diego, CA, USA) where indicated. All patient samples were provided by Karolinska University Hospital Biobank in a cryopreserved form. BM-derived samples were used in functional assays immediately after thawing. Information on patient characteristics is summarized in supplementary Table 1.
Cell lines
PG13 (ATCC CRL-10686), HEK293 (ATCC CRL-1573) and PhoenixGP (ATCC CRL-3215) cells were cultured in Dulbecco's Modified Eagle Medium+ Glutamax (DMEM-Glutamax) with the addition of 10% fetal bovine serum (FBS; Gibco, Billings, MT, USA). K562 (CCL-243), RPMI-8226 (ATCC CCL-155) and MM.1S (CRL-2974) were cultured in Roswell Park Memorial Institute (RPMI) 1640 media (Invitrogen, Carlsbad, CA, USA) enriched with 10% FBS. NK-92 cells (ATCC CSC-C0499) were maintained in Good Manufacturing Practice (GMP)-grade SCGM media (Sartorius CellGenix, Freiburg, Germany) supplemented with 20% FBS and 500 IU/mL recombinant interleukin-2 (IL-2; R&D Systems, Minneapolis, MN, USA) every 2–3 days. Cells were cultured in antibiotic-free conditions and were verified to be mycoplasma free with regular testing. The cells were maintained in a 37°C, 5% CO2 humidified incubator and split every 2–3 days.
NK-cell expansion
A GMP-compatible protocol was used for the expansion of NK cells (expNK), initially described by Alici et al. [
]. In summary, PBMCs were isolated from buffy coats of healthy donors using Ficoll density-gradient centrifugation (Lymphoprep; STEMCELL Technologies). PBMCs were cultured in a cell density of 0.5 × 106 cells/mL, in GMP-grade SCGM media (Sartorius CellGenix) supplemented with 5% human serum (Access Biologicals, Vista, CA, USA), in antibiotic-free conditions. On day 0 of the NK-cell expansion, the medium was supplemented with 10 ng/mL CD3 antibody (clone OKT3; Miltenyi Biotec) and 50 0IU/mL IL-2. From days 1–5, 500 IU/mL IL-2 was added daily, followed by IL-2 addition five times per week until the end of the expansion.
Generation of a stable retroviral vector (RV)-producing research cell bank (RCB)
RV-producing RCBs for the CAR and the control virus were produced according to the method described by Loew et al. [
]. In summary, VSV-G pseudotyped RVs were produced in PhoenixGP cells following transient transfection using the calcium phosphate transfection kit (Sigma-Aldrich, St. Louis, MO, USA). PhoenixGP cells were co-transfected with the envelope plasmid pMDG (Addgene #12259) and either pMFG-CD38A2 (kindly provided by Sorrento Therapeutics Inc., San Diego, CA, USA), encoding the αCD38-CAR, or the control plasmid pMFG-GFP encoding green fluorescent protein. Following the PhoenixGP cell transfection, produced RVs were used to stably transduce the packaging cell line PG13. Single-cell clones from the transduced cells were selected by limiting dilution and RCBs were established. Final RCB selection was done by direct comparison of the virus titer contained in freshly harvested supernatants.
Generation of αCD38-CAR–expressing NK cells
Cells of the NK-92 cell line or activated NK cells on day 13 of the expansion were transduced with RVs encoding the CAR or the GFP transgene, using RetroNectin-coated dishes according to manufacturer's instructions (Takara Bio, Kusatsu, Japan). To summarize, 1.5–2 mL of freshly harvested retrovirus containing solution was added to the RetroNectin-coated plates and centrifuged at 1000g for 2 h at 32°C to ensure binding of the RV particles to the RetroNectin reagent. The supernatant was then removed and 0.2 × 106 expNK cells were added to each well. The cells were incubated at 37°C, 5% CO2 for 3–4 days and analyzed for transduction efficacy by flow cytometry.
CRISPR/Cas9 KO of CD38
CD38 KOs of the MM target cell lines RPMI-8226 and MM.1S were generated using the CRISPR-Cas9 technology, according to a well-established protocol [
] and ordered from Integrated DNA Technologies (IDT, Coralville, IA, USA). The oligos were cloned into the lentiCRISPR.v2 plasmid (Addgene #52961). Lentiviral vectors (LVs) were produced in Poly-D-lysine coated plates (Corning, Corning, NY, USA) by calcium-phosphate-based transfection (CAPHOS; Sigma-Aldrich) of HEK293 cells treated with chloroquine (25μΜ, Sigma). The plasmids used were pMDLg/pRRE (Addgene #12251), pRSV-Rev (Addgene #12253), pCMV-VSV-G (Addgene #8454) and lentiCRISPR.v2-gRNA. After overnight transfection the media was replaced with fresh DMEM media complete with 10% FBS, 1% L-glutamine, 1% sodium pyruvate, 1% non-essential amino acids and 2% HEPES buffer. The LV-containing supernatant was collected 24 h and 48 h post-media change and was filtered through a 0.45-μm syringe filter before storing at –80°C. MM cell lines were transduced with the LVs using Polybrene (final c = 8 µg/mL; Sigma-Aldrich). Following assessment of the transduction efficacy by flow cytometry, cells were treated for 1 week with puromycin (final c = 1 μg/mL; Sigma-Aldrich) to allow for the selective survival of the transduced cells. If needed, flow cytometry-assisted cell sorting (FACS) sorting of the CD38-negative population was performed.
Flow cytometry
CAR-expressing cells were labeled with recombinant human CD38-Fc protein (Creative BioMart, Shirley, NY, USA), followed by a secondary mouse anti-hFc-PE antibody (clone HP6017; BioLegend, San Diego, CA, USA). Cell surface markers were labeled with the following anti-human antibodies: CD16-BUV737 (clone 3G8), CD56-BUV563 (clone NCAM16.2) and HLA-C PE (clone DT-9) (all from BD Biosciences, Franklin Lakes, NJ, USA); CD3-APC Cy7 (clone SK7), CD38-BV421 (clone HIT2), CD112-PE-cy7 (clone TX31), CD155-PerCP-Cy5.5 (clone SKIL4), CD324-FITC (clone 67A4), CD325-APC (clone 8C11), HLA-ABC-APC-Cy7 (clone W6/32), MICA/B-AlexaFluor488 (clone 6D4) and PCNA-AlexaFluor647 (clone PC10) (all from BioLegend). The antibody staining was done according to standard procedures [
]. All antibodies were titrated before use. Staining was performed in FACS buffer (2% FBS in phosphate-buffered saline) or Brilliant Stain Buffer (BD Biosciences) when two or more brilliant dyes were used. BD fixation/permeabilization kit (BD Biosciences) was used, according to manufacturer's instructions, for intracellular staining. In all experiments, Near-IR, Far Red or Violet LIVE/DEAD staining (Invitrogen) was added to the samples for analysis of the live cell population. Samples were fixed with 1% paraformaldehyde (HistoLab, Brea, CA, USA) before acquisition on CytoFLEX S (Beckman Coulter, Brea, CA, USA) or BD Symphony A5 (BD Biosciences) flow cytometers. Flow cytometry analyses were performed using FlowJo software, version 10.0 (TreeStar Inc., San Francisco, CA, USA).
In vitro NK-cell activation
Target cell lines were labeled with CellTrace Violet (CTV) (Invitrogen) following manufacturer's instructions, to facilitate experiment's analysis. CAR-NK and control NK cells were co-cultured with targets at an effector-to-target ratio (E:T) of 1:1 for 4 h, at 37°C and 5% CO2. Effector cells were chemically stimulated with phorbol 12-myristate 13-acetate (Sigma-Aldrich) and ionomycin (Sigma-Aldrich) or kept without targets as controls. After the first hour of co-culture, GolgiStop protein transport inhibitor (BD Biosciences) was added to promote the accumulation of cytokines in the Golgi apparatus and facilitate their intracellular detection. Degranulation was assessed by measuring surface CD107a expression with a PE-Cy7-labeled anti-hCD107a antibody (clone H4A3; BioLegend), whereas intracellular cytokine production was measured with anti-hIFNγ-APC (clone B27; BD) and anti-TNF-BV605 (clone MAb11; BD).
NK-cell cytotoxicity
The cytotoxic activity of CAR-NK/NK cells was assessed in a standard 51Cr-release assay [
Quantitative assay of the lytic action of immune lymphoid cells on 51-Cr-labelled allogeneic target cells in vitro; inhibition by isoantibody and by drugs.
]. In brief, 3 × 103 target cells labelled with 51Cr were co-cultured for 4 h with effector cells in triplicates, at E:T ratios of 10:1, 3:1, 1:1 and 1:3. To determine the minimum and maximum release of radioactive chromium, 51Cr-labeled target cells were cultured alone or in 0.5% Triton X-100, respectively. After a 4-h incubation, 20 μL of supernatant from each sample was transferred to Luma plates (PerkinElmer, Waltham, MA, USA) and left to dry overnight. The radioactive signal was measured using a MicroBeta 2450 Microplate Counter (PerkinElmer). The percentage of specific cell killing was calculated as follows: (chromium release from co-culture with effectors – minimum chromium release)/ (maximum chromium release – minimum chromium release) × 100. For assessing the NK-cell cytotoxicity during repeated tumor cell challenges, wild-type (WT) RPMI-8226 target cells were labelled with CTV (Invitrogen) for the first challenge, or CellVue NIR815 (Thermo Fisher Scientific) for the second, according to manufacturer's instructions. Effector and target cells were co-cultured in a 1:1 ratio and cell death was assessed 2 days after each challenge using LIVE/DEAD stain (Invitrogen).
Proliferation assay
Effector cells transduced with RV-CAR or RV-GFP were labeled with the proliferation marker CTV (Invitrogen), according to manufacturer's instructions. Effector cells were then cultured with targets at E:T of 1:1, or kept alone. A sample of CTV-labeled cells was analyzed at day 0 of the assay to determine the maximum CTV signal, which corresponds to the signal of undivided cells. Following 5 days of co-culture, samples were acquired by flow cytometry.
Statistical analysis
Statistical analysis was performed using the GraphPad Prism 9.0 software (GraphPad, San Diego, CA, USA). Student's t-test or analysis of variance was used to compare quantitative differences between two or more groups, respectively. All data points represent mean ± standard deviation, whereas values of *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 were considered significant.
Results
Retroviral transduction of αCD38-CAR to NK-92 cells has high efficacy without impairing viability
RVs carrying the αCD38-CAR (Figure 1A) or the control GFP transgene were harvested by the respective RV-producing RCBs and used to transduce NK-92 cells in order to evaluate the expression and stability of the CAR on the NK-cell surface. Transduction efficacies greater than 60% were achieved for both αCD38-CAR and GFP as determined by flow cytometry staining (Figure 1B). To assess the functionality and specificity of the CAR construct, transduced NK-92 cells underwent enrichment by FACS and a pure population was obtained. αCD38-CAR-NK-92 cells showed similar viability to GFP+ NK-92 and WT NK-92 in culture (supplementary Figure 1A) and had stable CAR expression above 97% over time (supplementary Figure 1B). Notably, assessment of CD38 expression showed no significant differences between CAR and WT NK-92 (supplementary Figure 1C).
Fig. 1Generation and functional assessment of αCD38-CAR-NK-92 cells against CD38+ and CD38− targets. (A) Schematic representation of αCD38-CAR construct. (B) Representative flow cytometry plots showing cell surface expression of CAR (left) and control GFP (right) protein on NK-92 cells, 3 days after retroviral transduction (N = 2). (C) Surface CD38 expression on MM cell lines RPMI-8226 and MM.1S. (D) CD38 expression of WT and CD38 KO target cell lines compared to unstained (US) controls. CD38 expression was assessed following enrichment of the CD38 KO population by puromycin selection and FACS sorting. (E, F) In vitro responsiveness of αCD38-CAR-NK-92 and WT NK-92 cells following 4-h co-incubation with CD38+ or CD38-KO MM cell lines, showing frequency of (E) degranulating and (F) IFNγ+ cells. Results represent three independent experiments. (G) 51Cr release assay of WT and CD38-KO RPMI-8226 and MM.1S targets after 4-h co-culture with αCD38-CAR-NK-92 cells. Results from three independent experiments. Statistics determined by Student's t-test, two tailed. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.
αCD38-CAR increases the in vitro response of NK-92 cells against CD38+ targets
To evaluate the cytotoxic potential of the αCD38-CAR-NK-92 cells, MM cell lines with differential cell surface expression of CD38 were used as targets, namely the CD38high RPMI-8226 and CD38dim MM.1S (Figure 1C). The selectivity of the CAR against CD38+ targets was controlled using CD38 KO target cells, generated by CRISPR/Cas9 (Figure 1D; supplementary Figure 2A). The phenotype of the CD38 KO RPMI-8226 and MM.1S populations was further assessed for the expression of NK-cell activating and inhibitory ligands to ensure that no bias was introduced by the genetic manipulation. None of the receptors analyzed changed in expression, except for the frequency of CD112+ cells which was found higher in CD38 KO RPMI-8226 cells (supplementary Figure 2B,C).
In vitro, αCD38-CAR-NK-92 cells showed significantly higher degranulation and IFNγ production against CD38+ MM cells compared with WT NK-92 and GFP+ NK-92 cells (Figure 1E,F). Importantly, increased response was not observed against CD38 KO targets. In addition, compared with WT NK-92 and NK-92-GFP cells, αCD38-CAR-NK-92 cells induced CD38-specific target cell lysis (Figure 1G). Collectively, these results suggest a functional and selective CAR construct, capable of increasing the in vitro response of NK-92 cells against CD38-expressing targets.
Expansion of primary NK cells induces gradual CD38 downregulation
Following the proof-of-concept that NK-92 cells are amenable to genetic engineering with this αCD38-CAR to confer target specificity, we sought to introduce the αCD38-CAR construct in primary NK cells. Peripheral blood (PB)-derived NK cells from healthy individuals were expanded using the GMP-compatible ex vivo NK cell expansion protocol as previously described [
]. Expansion of PB-NK cells showed exponential increase in NK cell numbers, specifically between days 5 and 15, reaching up to 2 × 103-fold expansion (Figure 2A). At day 16, the cultures were enriched for NK cells, comprising 65.3±11.9% of the total cells (supplementary Figure 3A). Notably, phenotypic analyses revealed a progressive decrease of surface CD38 expression (Figure 2B), both in terms of the frequency of positive cells (Figure 2C) and the mean fluorescence intensity (MFI) of the CD38+ NK cells (Figure 2D). These observations were consistent, indicating that CD38 downregulation occurs irrespective of the individual healthy donor characteristics and the day of flow cytometry analysis. Similar CD38 downregulation was reported in expansion of PB-NK from patients with MM [
]. Importantly, CD38dim NK cells were able to evade daratumumab-induced fratricide (supplementary Figure 3B) and presented an activated phenotype (supplementary Figure 3C,D).
Fig. 2Assessment of CD38 expression dynamics during ex vivo PB-NK cell expansion. (A) NK-cell fold expansion (N = 5 donors). (B) CD38 expression on NK cells of one representative donor at different time points of ex vivo NK cell expansion, as determined by flow cytometry. (US, unstained sample) (C) Frequency and (D) intensity of CD38 expression on NK cells (N = 6 donors) during expansion. (E) Comparison of transduction efficacies (TX) resulted after NK-cell transduction with RV-CAR or control RV-GFP at different time points (day 5, 7, 13, 16) of ex vivo expansion (N = 4 donors). The transduction efficacy was measured by flow cytometry on day 4 after transduction. Bars represent the median. (F) Activation of non-transduced expNK, NK-GFP and αCD38-CAR-NK cells after 4-h contact with K-562 target cells (N = 7 donors). Results from three independent experiments. Statistics determined by Student's t-test, two tailed. (n.s., not significant)
We next studied the impact of NK-cell expansion rate to the efficacy of retroviral transduction. PB-NK cell transduction was compared between four different expansion time points, ranging between day 5 and 16 (Figure 2E). In general, NK cells showed higher transduction rates for GFP than αCD38-CAR transduction, which, however, equalized towards the later time points. Transduction on day 16 resulted in the lowest efficacy for both transgenes, reflecting the decline in NK expansion rate, which can negatively impact RV transductions. Overall, day 13 was considered the optimal transduction time point, reaching a mean of 62.8 ± 3% for the GFP transgene and 57.4 ± 2.9% for the CAR. Moreover, transduced CD38dim NK cells maintained viability greater than 80%, as determined by flow cytometry (supplementary Figure 3E), and samples showed no significant differences in their NK-cell frequency (supplementary Figure 3F). Importantly, transduced CD38dim NK cells showed potent effector function in co-cultures with the NK-sensitive cell line K562, which was comparable between unmodified expNK, GFP and CAR transduced samples (Figure 2F).
CD38dim CD38-CAR-NK cells efficiently activate against CD38+ target cells
Taking our findings from Figure 2 into consideration, we developed an optimized protocol that incorporates NK-cell expansion, RV transduction and NK-cell function assessment steps (Figure 3A). Based on this, we aimed to test whether primary CD38dim αCD38-CAR-NK cells display specificity against CD38+ target cells. Specifically, CAR or GFP-expressing NK cells were assayed for the expression of CD107a (Figure 3B), production of interferon γ (IFNγ) (Figure 3C) and tumor necrosis factor (Figure 3D) after co-culture with CD38+ MM cell lines or their CD38 KO counterparts. The flow cytometry gating strategy that we followed, as well as the assay's controls, are displayed in supplementary Figure 4. Indeed, increased frequency of degranulating, IFNγ and tumor necrosis factor–producing CAR-expressing NK cells was observed against CD38+, but not CD38−, MM cell lines (Figure 3B-D). In addition, CAR-expressing NK cells showed proliferative advantage over nontransduced and GFP-transduced expanded NK cells upon contact with wild type RPMI-8226 and MM.1S cells (supplementary Figure 5).
Fig. 3Expression of αCD38-CAR increases the activation of primary NK cells against CD38+ targets. (A) Schematic representation of the protocol followed for the expansion and transduction of healthy donor-derived PB-NK cells (B-D) 4-h in vitro responsiveness assay showing frequency of (B) CD107a (C) IFNγ and (D) TNF expressing effector cells (unmodified expNK, NK-GFP and αCD38-CAR-NK), in co-cultures with CD38+ and CD38 KO MM cell lines. Cells were transduced on day 13 of the ex vivo expansion and assessed for their in vitro activation 4 days later. A total of 7 donors were used, in 3 independent experiments. Statistics determined by Student's t-test, two tailed. Bars indicate the median. (E) 51Cr release assay showing % of specific cell lysis induced by the αCD38-CAR-NK cells of 2 different donors against the WT or CD38 KO target cell lines RPMI-8226 and MM.1S. NK cells were FACS-sorted for CD3−CD56+ expression. Results from two independent experiments. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.
To assess whether CAR-expressing NK cells also induce CD38-specific target cell lysis we performed killing assays. Similar to our observations in NK-92 cells, PB-derived αCD38-CAR-NK cells of two representative donors showed increased lysis of CD38+ compared with CD38− MM cells (Figure 3E), whereas GFP-transduced PB-NK cells did not differ in their ability to kill CD38+ and CD38− targets (supplementary Figure 6A). A comparison between the killing induced by αCD38-CAR-NK and NK-GFP cells across four healthy donors revealed that αCD38-CAR-NK showed 2–17% higher specific killing of RPMI-8226 cells and 3–28% higher specific killing of MM.1S cells compared with GFP+ NK cells (supplementary Figure 6B).
Since CD38 is a known activation marker and can be upregulated during the exertion of NK cell cytotoxicity, we investigated the CD38 expression levels before and after repeated challenges with WT RPMI-8226 cells. CD38 expression was indeed found upregulated after contact with target cells in all donors and was significantly higher in CAR+ NK cells (supplementary Figure 7A–C). In the same experiment, we monitored the serial killing potential of αCD38-CAR-NK cells by counting dead target cells after each challenge, which revealed a marked advantage of CAR-expressing cells during the second challenge (Supplementary Figure 7D). After 5 days of co-culture and two challenges, death of NK cells in all conditions was below 20%, although it was significantly higher in the CAR-expressing cells (Supplementary Figure 7E). Altogether, the results establish the feasibility of the developed protocol and demonstrate the targeted cytotoxic potential of CAR-expressing CD38dim PB-NK cells.
αCD38-CAR-NK-92 cells activate against BM-derived cells from patients with MM in a CD38-dependent manner
To increase the translational value of our study, modified NK-92 cells were tested for CD38-specific recognition of primary MM cells from five different patients (supplementary Table 1). The included patients had undergone up to 9 rounds of treatment and relapse prior to the collection of the BM aspirate. Importantly, all but one patient had received αCD38-antibody treatment (daratumumab), which ultimately led to disease progression. αCD38-CAR-NK-92 cells were first tested for reactivity against CD138+ unselected BM-derived MNC samples from three patients. Phenotypic analysis showed malignant CD138+ cell content of below 18% (Figure 4A) in all samples and high CD38 expression in the tumor cells of patients MM.1 and MM.2 (Figure 4B). In in vitro stimulation experiments, αCD38-CAR-NK-92 showed a higher frequency of CD107a+ (Figure 4C) and IFNγ-producing cells against sample MM.2 that had the highest expression of CD38 (Figure 4D).
Fig. 4αCD38-CAR-NK-92 cells activate against primary MM cells. (A) Frequency of malignant (CD138+) cell content within the BM-derived mononuclear cell samples of three patients with MM. (B) Flow cytometry plots indicating the CD38 expression of the malignant cells. Intensity of CD38 expression (MFI) is depicted within the plots. (C, D) In vitro activation of αCD38-CAR-NK-92 and NK-92-GFP cells after 4-h contact with BM-derived MNC samples from three patients with MM. Results show (C) degranulation and (D) IFNγ release. (E) CD38 surface expression of CD138-selected MM cells from two patients as determined by flow cytometry. Intensity of CD38 expression (MFI) is depicted within the plots. (F) Degranulation of αCD38-CAR-NK-92 and NK-92-GFP cells following 4-h co-culture with primary CD138+-selected MM cells of two patients. (G) Combined results of the degranulating capacity of αCD38-CAR-NK-92 compared with NK-92-GFP cells after contact with primary CD138+-selected or CD138+-unselected (MNC) MM cell samples.
To further validate our results, CD138+-selected samples from two patients with MM were used. The samples were assessed for their CD38 expression where patient MM.5 appeared to have considerably higher frequency and MFI of CD38-expressing cells, compared with patient MM.4 (Figure 4E). This difference was expected, as MM.4 had undergone CD38-targeting antibody therapy before the BM sample collection (supplementary Table 1). The activation of αCD38-CAR-NK-92 cells, as measured by CD107a expression, was in both cases higher than that of NK-92-GFP cells (Figure 4F). Notably, 40% more CAR+ NK-92 cells degranulated against the CD38high MM.5, compared with the CD38low MM.4. The combination of results from the unselected and selected samples further supported the fact that αCD38-CAR-NK-92 are effectively activating against primary MM samples (Figure 4G). Together the data suggest that CAR potentiates NK-92 cells against CD138+-selected and CD138+-unselected BM samples of patients with MM.
We next sought to assess the reactivity of αCD38-CAR-NK cells against autologous MM samples. For this, NK cells from PBMCs of three patients were expanded according to our previously optimized protocol (Figure 3A). In contrast to PBMC expansion from healthy donors, NK-cell frequency remained overall low throughout the expansion, reaching 40 ± 24.7% on day 17 of expansion (Figure 5A). In parallel, phenotypic analysis revealed that frequency of CD38+ NK cells remained stable, while MFI CD38 showed decrease (Figure 5B) [
]. Further phenotypic analyses confirmed that expanded NK cells expressed high levels of CD16, NKp30 and NKp46, which did not change after transduction with either CAR or GFP transgenes (supplementary Figure 8).
Fig. 5αCD38-CAR-expressing NK cells of patients with MM effectively activate against autologous tumor cell-containing BM samples in vitro. (A) Ex vivo expansion of PB-derived NK cells from three patients with MM depicted as frequency of NK cell in culture. NK cells were identified as CD3−CD56+ by flow cytometry. (B) Frequency of CD38+ NK cells (left) and intensity of surface CD38 expression
on PB-derived NK cells from three patients with MM during expansion. Cells were transduced with RV-GFP or RV-CAR on day 13 of expansion. CD38 expression of the transduced cells was assessed again on day 4 after transduction (day 17). (C) Transduction efficacy of patient-derived expanded NK cells on day 4 after retroviral transduction. (D) Degranulation and (E) IFNγ release of CAR-NK and NK-GFP cells after 4-h co-incubation with autologous CD138+-containing BM-derived MNC samples at E:T = 1:1.
Transduction of expanded NK cells on day 13 of expansion resulted in transduction efficiencies of 9–15% for αCD38-CAR and 12–65% for the GFP control transgenes (Figure 5C). Following the transduction, NK cells were assessed for in vitro responsiveness against autologous BM-derived mononuclear cell samples, where CAR-expressing NK cells showed higher degranulation and IFNγ production compared with GFP-transduced NK cells (Figure 5D,E). Altogether, the data suggest that patient PB-derived NK cells expand in culture, are amenable to genetic engineering with CARs, sustain a high expression of NK cell activating receptors and have the capacity to specifically recognize autologous MM cells.
Discussion
Encouraged by the successful application of CD38-targeted antibody therapy, the recent approval of CAR-T immunotherapy in MM, and the promising results of NK cell-based therapy in hematological malignancies, we sought to assess the potential of αCD38-CAR-NK cells in MM. Although αCD38-CAR-NK therapy has been investigated in a preclinical setting, the inherent expression of CD38 on the surface NK cells and the potential fratricide concerns question the feasibility of the approach [
]. Here, we introduce an affinity-optimized second-generation CAR construct targeting a unique epitope on the CD38 protein. We demonstrate the CAR's ability to be stably expressed on the surface of NK cells and to be functional and selective for its target CD38. The αCD38-CAR construct significantly enhanced the effector function of NK-92 cells against MM cell lines in vitro. In addition, αCD38-CAR-NK-92 cells exerted higher cytotoxicity towards the CD38high RPMI-8226 cells, compared with the CD38low MM.1S cells, suggesting a potential positive correlation between effector induced cytotoxicity and target cell CD38 surface expression.
Despite NK-92 cells being a useful proof-of-concept platform for NK cell immunotherapy, the immortal nature of the cell line, the required irradiation step before clinical use, and the related safety concerns shifted our attention towards primary NK cells [
]. PB is an easily accessible NK-cell source of minimal invasion, from which large amounts of activated NK cells can be produced by ex vivo manipulation [
IL-2-based immunotherapy after autologous transplantation for lymphoma and breast cancer induces immune activation and cytokine release: a phase I/II trial.
], showing impressive activity in vitro and adequate safety profiles in vivo. However, due to the limited clinical response that the initial trials reported, the autologous setting remained largely under-studied and did not follow the progress of the NK cell therapy field. Given the fact that NK cell expansion methods have since been optimized, and genetic manipulations (i.e., CARs) can confer additional or enhanced effector properties, autologous NK-cell therapy may still hold potential.
Specifically in the context of MM, we have previously shown that patient-derived PB-NK cells can be efficiently expanded ex vivo, reaching an average of 511-fold increase at the end of the 20-day expansion protocol [
]. Moreover, in a recent clinical study we demonstrated that, upon infusion, the autologous NK-cell product is detected for about 4 weeks in PB without causing severe adverse events [
]. Here, we apply the same GMP-grade feeder free-expansion protocol, based on IL-2 support. In healthy donor PBMCs, we report the highest NK cell proliferation rate between day 5 and 15 and observe a sharp decrease both in the percentage of CD38+ NK cells and in the intensity of surface CD38 expression within 5 days after the initiation of the expansion. Given the fact that CD38+ NK cells have been associated with fratricide events when combined with CD38-targeting treatments [
], we found the spontaneous generation of CD38dim NK cells desirable for our application. By transducing primary NK cells on day 13 of the expansion, when the CD38 expression is low and the expansion rate peaks, we additionally achieved high transduction rates in line with RVs’ ability to infect cells undergoing mitosis. Importantly, we carefully considered the translational aspect of the study, ensuring all steps of our optimized protocol, i.e., the NK-cell expansion protocol, the generation of RVs by viral vector-producing cell bank and the RetroNectin-mediated transduction method can be easily scalable and GMP-compliant.
Interestingly, and opposite to our findings, feeder cell expansion systems are associated with significant CD38 upregulation in NK cells [
]. Phenotypic variations between expanded NK cells of different protocols are generally expected, though not easily predicted, or understood. Due to this, different explanations have been considered. According to Wang et al. [
], PB-NK cells sorted for CD38−/low expression show a significant proliferative advantage over CD38+ NK cells, during feeder cell-based ex vivo expansion. Thus, it is plausible that the initial small percentage of CD38−/low NK cells in the PBMC pool rapidly overgrow the expanding CD38+ NK and prevail in culture. However, the same study also reported that despite the initial CD38−/low NK cell phenotype, cells quickly acquire surface CD38 during expansion. Therefore, we believe that our findings are less likely related to the population of CD38−/low NK cells, but rather to differential biological inputs affecting CD38 regulation derivative to our expansion conditions and the presence of other cytokine-releasing cell populations, like T cells. Further investigation on the CD38 regulation in NK cells would be beneficial not only for deciphering such observations but also to advance NK cell-based immunotherapy.
Despite the significance of CD38 in cellular processes, loss of surface expression does not impair NK cell effector functions. This was evident in our current study, as well as in previous studies of our group [
] demonstrated that CD38−/low NK cells had significantly higher anti-tumor activity in vitro and in vivo compared to CD38+ NK cells. It has been additionally shown that CD38− NK cells undergo favorable metabolic alterations as a result of higher NAD+ concentration and improved glycolysis [
]. Although the CD38−/low have been extensively compared with CD38+/high NK phenotype, the same does not apply for CD38 KO and CD38dim NK cells, especially regarding their in vivo anti-tumor activity. Based on the key role of CD38 in lymphocyte trafficking via the CD38/CD31 axis, and its additional roles in participation in signaling events and immune synapse formation [
], it is likely that dim CD38 expression is a well-reasoned alternative to complete KO. Although a direct comparison was beyond the scope of this study, we believe that future directions should focus on identifying the favored NK cell phenotype, since there is also increasing interest for the use of CD38 KO NK cells in combinatorial treatments with CD38-targeting monoclonal antibodies [
The CD38dim NK cells were used as the effector cell-basis for the validation studies of the αCD38-CAR-NK approach. In primary NK cells, we show that CAR expression provides an activating advantage and promotes the selective manifestation of cytotoxicity towards CD38+ MM cell lines. The functionality of the CAR was further corroborated in in vitro experiments of αCD38-CAR-NK-92 against primary MM targets. Notably, among the different mononuclear cell targets, CAR+ NK-92 cells displayed more potent effector responses against the sample with the highest MM cell content. The same was observed against the CD38high CD138-selected sample, compared to the CD38low sample of the patient that was treated with daratumumab. These observations provide evidence regarding the selectivity of the CAR for CD38high targets.
We next assessed the feasibility of autologous αCD38CAR-NK cell immunotherapy in an in vitro setting. NK-cell expansion from PBMCs derived from patients with MM was characterized by a low expansion rate and less prominent CD38 downregulation, compared with healthy donor-derived NK cells. Since this was not observed when the same expansion protocol was applied in PBMCs of newly diagnosed patients [
], we deduce that the heavy treatment that the patients had undergone before blood sample collection was causative. The low expansion rate further resulted in low transduction efficacies with our retroviral approach. To tackle both of these issues, we suggest collection of blood sample at diagnosis and use of lentiviral vectors that would allow for transduction of both dividing and non-dividing cells [
]. Nevertheless, CAR-expressing patient NK cells showed increased in vitro responsiveness against the autologous tumor-containing BM samples, compared with the GFP-transduced, in all patients. The fact that these patients were daratumumab-refractory strengthens our argument regarding the potential of the approach as an alternative therapy.
Taking everything into consideration, this study aims to provide a novel and feasible approach to αCD38-CAR-NK therapy for MM. By harnessing the CD38dim NK cell phenotype that occurs across healthy donor and newly diagnosed patient PB-NK cells over the course of our expansion protocol, we have bypassed the need for genetic CD38 knock-out while generating adequate amounts of highly reactive and clinically safe primary NK cells [
]. In addition, we have provided evidence that combination with an affinity optimized αCD38-CAR construct leads to increased and selective reactivity against CD38+ MM cells. This alternative to existing CD38-targeting cell-based approaches could benefit patients with MM found to be refractory to previous lines of treatment and/ or are unsuitable for allogeneic adoptive cell therapy. It could further be considered for treating patients of other diseases where CD38-targeting has proven effective, such as acute myeloid leukemia and B-cell chronic lymphoblastic leukemia. Overall, our study emphasizes the potential of integrating NK-cell expansion protocols to targeted immunotherapies, while our findings aim to inspire novel therapeutic approaches where genome editing could be replaced by the selection of a suitable expansion protocol.
Author Contributions
Conception and design of the study: MK, AL, AKW and EA. Acquisition of data: MK, MVM, KHS, MG and AKW. Analysis and interpretation of data: MK, MVM, KHS, AH, YZ, JDG, JL, GK, HGL, HJ, AL, AKW, WG and EA. Drafting or revising the manuscript: MK, MVM, AL, AKW and EA. All authors have approved the final article.
Declaration of Competing Interest
YZ would like to disclose employment at Sorrento Therapeutics Inc. JDG, HJ and WG would like to disclose equity ownership and shared patents and royalties in Sorrento Therapeutics Inc. GK is disclosing equity and options in Sorrento Therapeutics, Inc. EA is scientific advisor to Sorrento Therapeutics. All other authors have no commercial, proprietary or financial interest in the products or companies described in this article.
Funding
The study was performed with support from VINNOVA (Sweden's Innovation Agency; grant number: 2019-00056), Castenbäcks stiftelse för cancer research (EA), Radiumhemmets Forskningsfonder (EA; grant number: 191063), KI fonder (EA) and the collaborative agreement with Sorrento Therapeutics Inc.
Supplementary Fig. 1. Genetically modified NK-92 cells exhibit similar viability to wild type cells. (A) Representative flow cytometry plots of WT NK-92, NK-92-GFP and αCD38-CAR-NK-92 cell viability in culture (N = 3). Transduced cells were FACS sorted for GFP or CAR expression respectively, before the assessment. (B) Surface expression of αCD38-CAR and GFP transgenes on NK-92 cells over time in culture. (C) Frequency (left) and MFI (right) of CD38 expression on WT NK-92, NK-92-GFP and αCD38-CAR-NK-92 cells (N = 3).
Supplementary Fig. 2. Phenotypic characterization of WT and CD38 KO target cells. (A) Efficacy of CD38 KO on day 4 after lentiviral transduction with the CRISPR-Cas9 system. The cells were subsequently subjected to puromycin selection and FACS sorting to acquire a pure CD38 KO population. (B, C) Expression of NK cell activating and inhibitory ligands on WT and CD38 KO (B) RPMI-8226 and (C) MM.1S target cell lines. WT and CD38 KO populations were examined for expression of E-cadherin, N-cadherin, CD112, CD155, MIC-AB, HLA-ABC, HLA-C and PCNA in 3 independent flow cytometric analyses. Statistics determined by Student's t-test, two tailed (**P ≤ 0.01).
Supplementary Fig. 3. Frequency and phenotype of NK cells before and after transduction with RVs. (A) Changes in NK-cell frequency during ex vivo expansion (N = 9 donors). (B) Frequency of NK-cell death induced by treatment with 10 µg/mL daratumumab for 4 h, in relation to the NK cell expansion day (N = 2 donors). (C) Heat map showing frequency of NK-cell activation markers and activating receptors before (Day 0) and after (Day 20+) ex vivo expansion (N = 2 donors). (D) Heat map of the median fluorescent intensity (MFI) of the NK cells positive for the respective marker. The values are normalized according to the function: (MFI day20+/ MFI day 0) × 100 and presented in a log scale (N = 2 donors). (E) NK-cell viability at day 4 after retroviral transduction (day 17 of NK cell expansion) with RVs encoding for GFP or CAR proteins. (F) Comparison of NK-cell frequency at day 4 after retroviral transduction (day 17 of NK cell expansion) with RV-GFP or RV-CAR. Statistics determined by Student's t-test, two tailed (n.s., not significant).
Supplementary Fig. 4. Analysis of in vitro responsiveness assay (A) Flow cytometry gating strategy. Target cells were labelled with CTV staining before co-culture for ease of discrimination during analysis. NK cells were identified as CD3−CD56+. (B) Unstimulated (negative) and PMA/ionomycin (positive) controls of the in vitro activation assay, showing spontaneous and maximum degranulation, IFNγ and tumor necrosis factor release respectively of unmodified expNK, GFP-NK and αCD38-CAR-NK cells (N = 7 donors). Results from three independent experiments.
Supplementary Fig. 5. αCD38-CAR-NK cells have increased proliferation after contact with CD38+ target cells. Proliferation of non-transduced, GFP- and CAR-expressing NK cells after 5 days in culture with (A) no target, (B) K-562, (C) RPMI-8226 WT and KO and (D) MM.1S WT and KO cells. NK cells were identified as CD3−CD56+ by flow cytometry. Results from one healthy donor.
Supplementary Fig. 6. αCD38-CAR-NK cells exhibit increased cytotoxicity against CD38+ targets compared with NK-GFP cells. (A) 51Cr release assay showing % of specific cell lysis induced by NK-GFP cells of 2 healthy donors against WT or CD38 KO RPMI-8226 and MM.1S cell lines. (B) Difference between the specific cell lysis induced by CAR- and GFP-expressing cells in killing CD38+ targets (N = 4 donors). Results from three independent 51Cr release assays experiments.
Supplementary Fig. 7. CD38 expression is induced on αCD38-CAR-NK cells following long-term co-culture with RPMI-8226 target cells. (A) Representative flow cytometry histograms showing CD38 expression on NK cells before and after repeated challenges with RPMI-8226 target cells. Target cells were added at a 1:1 ratio and CD38 expression was assessed by flow cytometry 2 days after each challenge (N = 3 donors). (B) Frequency and (C) median fluorescent intensity (MFI) of CD38-positive untransduced (untx) and CAR-expressing NK cells in relation to the rounds of stimulation with WT RPMI-8226 cells (N = 3 donors). (D) Frequency of dead RPMI-8226 cells and (E) dead NK cells after repeated rounds of stimulation.
Supplementary Fig. 8. Expression dynamics of NKp30, CD16 and NKp46 on PB-NK cells from 3 MM patients during ex vivo NK cell expansion and after retroviral transduction. Cells were transduced with RV-GFP or RV-CAR on day 13 of expansion. Receptor expression was assessed on day 4 following viral transduction (day 17).
CD38, a receptor with multifunctional activities: from modulatory functions on regulatory cell subsets and extracellular vesicles, to a target for therapeutic strategies.
Daratumumab, lenalidomide, and dexamethasone versus lenalidomide and dexamethasone alone in newly diagnosed multiple myeloma (MAIA): overall survival results from a randomised, open-label, phase 3 trial.
Bortezomib, thalidomide, and dexamethasone with or without daratumumab before and after autologous stem-cell transplantation for newly diagnosed multiple myeloma (CASSIOPEIA): a randomised, open-label, phase 3 study.
Impact of performance status on overall survival in patients with relapsed and/or refractory multiple myeloma: real-life outcomes of daratumumab treatment.
CD38-specific chimeric antigen receptor expressing natural killer KHYG-1 cells: a proof of concept for an "off the shelf" therapy for multiple myeloma.
Quantitative assay of the lytic action of immune lymphoid cells on 51-Cr-labelled allogeneic target cells in vitro; inhibition by isoantibody and by drugs.
IL-2-based immunotherapy after autologous transplantation for lymphoma and breast cancer induces immune activation and cytokine release: a phase I/II trial.
00068-3&id=gr1.jpg){kind=link}
00068-3&id=gr2.jpg){kind=link}
00068-3&id=gr3.jpg){kind=link}
00068-3&id=gr4.jpg){kind=link}
00068-3&id=gr5.jpg){kind=link}