Daratumumab in combination with urelumab to potentiate anti-myeloma activity in lymphocyte-deficient mice reconstituted with human NK cells

ABSTRACT Daratumumab is an anti-CD38 fully human IgG1 mAb approved for multiple myeloma treatment. One of the proposed mechanisms of action is the induction of antibody-dependent cellular cytotoxicity (ADCC) mediated by NK cells. NK cells acquire surface CD137 expression in the presence of solid-phase-attached daratumumab and when encountering a daratumumab-coated CD38+ tumor cell line. In this setting, addition of the agonist anti-CD137 mAb urelumab enhances NK-cell activation increasing CD25 expression and IFNɣ production. However, in vitro ADCC is not increased by the addition of urelumab both in 4h or 24h lasting experiments. To study urelumab-increased daratumumab-mediated ADCC activity in vivo, we set up a mouse model based on the intravenous administration of a luciferase-transfected multiple myeloma cell line of human origin, human NK cells and daratumumab to immuno-deficient NSG mice. In this model, intravenous administration of urelumab 24h after daratumumab delayed tumor growth and prolonged mice survival.


Introduction
CD38 is a transmembrane protein expressed on all malignant plasma cells in multiple myeloma. 1 The anti-CD38 IgG1 human monoclonal antibody daratumumab is used in combination with other drugs in relapsed or refractory multiple myeloma patients 2 and has been recently approved in combination with bortezomib, melphalan, and prednisone for newly diagnosed multiple myeloma. 3 Its clinical activity is contingent on antibody dependent-cellular cytotoxicity (ADCC) as mediated by NK cells and macrophages 4,5 and probably the additional poorly understood effects of CD38 ligation on immune system cells and on myeloma cells themselves. 6 Complement-mediated cytotoxicity and reductions of myeloid-derived suppressor cells in the myeloma microenvironment have also been observed. 4,7 CD137 (4-1BB) is a TNFR family surface glycoprotein expressed on activated T and NK cells. 8 Agonist monoclonal antibodies directed to CD137, such as urelumab, enhance T-cell mediated antitumor immunity 9 and have been shown to enhance anti-tumor ADCC as mediated by rituximab, 10 cetuximab, 11 and trastuzumab. 12 The mechanism proposed is that CD16 (FcRɣIIIa) ligation by a cell surface-bound IgG1 Fc leads to the induction of CD137 on the surface of NK cells. Once on the NK surface, if CD137 is stimulated by a CD137 agonist antibody, it enhances the survival and cytotoxic performance of NK cells in terms of more efficient ADCC. [10][11][12] These preclinical studies support the combined use of daratumumab with urelumab in multiple myeloma including evidence in NKhumanized mice.

Results and discussion
We hypothesized that daratumumab could synergize with urelumab and perhaps other CD137 agonists in its activity against multiple myeloma. We first found that plastic-bound daratumumab, as any other human IgG1 antibody, induces in 16 hours surface expression of CD137 on resting NK cells freshly isolated from peripheral blood of healthy donors (Figure 1(a)). Of note, such an activity was not further increased by incubation with known NK-stimulating cytokines such as IL-2, IL-15 or IL-12 (Figure 1(a)). Consistent with this, if the CD38 + multiple myeloma-derived cell line MM1S-GFP-Luc (Supp. Figure 1) was coated with daratumumab but not with irrelevant human IgG, it induced CD137 surface expression on cocultured resting NK cells (Figure 1(b)). This effect was not enhanced by the addition of cytokines to the 16-hour cocultures. These results were recapitulated with the multiple myeloma cell line KMS28-BM coated with daratumumab (Supp. Figure 2A). NK cells that upregulated CD137 on the membrane were mostly CD56 dim . Moverover, NK cells positive for CD25, CD94, NKG2D and NKG2A expressed more frequently CD137 on the membrane than those NK cells that did not expressed these markers (Figure 1(c)). However, CD16 + cells expressed less frequently CD137 than CD16 − NK cells, as had been previously descrided. 12 Importantly, gated NK cells in seven bonemarrow aspirates from multiple myeloma patients upregulated CD137 in short-term cultures when daratumumab was added (Figure 1(d)). CD38 expression on normal and clonal plasma cells as well and on other lymphoid and myeloid populations of one of the patient bone marrow was studied, confirming expression not only on plasma cells but also on a variety of lymphoid and myeloid cells including NK cells (Supp. Figure 3).
To ascertain if CD137 ligation by urelumab costimulated NK cells, we exposed freshly isolated human NK cells to plate-bound human IgG1 in order to upregulate surface CD137 and harvested these cells 24h later to be exposed to MM1S-GFP-Luc cells coated with daratumumab in the presence or absence of (c) CD137 co-expression with CD56, CD16, CD25, CD94, NKG2D and NKG2A on NK cells cultured during 16h in plates coated with human IgG. Paired dots in the graphs indicate data from each subject. (d) FACS-Gating strategy and CD137 expression on NK cells from bone marrow aspirates taken from multiple myeloma patients, that were cultured during 16h with daratumumab or irrelevant human IgGs. A representative contour plot and a bar graph with data from 7 subjects are shown. * = p < 0.05, ** = p < 0.01 and *** = p < 0.001 in a paired T-test or one-way ANOVA test.
soluble urelumab that was added to these 24 and 48h cultures. Figure 2(a) shows that urelumab increased the level of CD25 surface expression, as well as the intracellular content of IFNɣ (Figure 2(c)). In these co-cultures, CD137 was co-expressed with CD25 on NK cells (Figure 2(b)), and the fraction of NK cells expressing a higher level of intracellular IFNɣ, co-expressed more frequently CD137 (Figure 2(e)). Co-expression of other NK cell markers was also studied in this setting (Supp. Figure   4A-F). NK cells that expressed more PD-L1, LAG-3, NKG2D, CD94 and NKG2A on the membrane, expressed more frequently CD137 than their negative counterparts. In the same way, NK cells containing more Granzyme B exhibited a higher proportion of CD137 + cells. Moreover, the supernatants of such cultures showed higher concentrations of secreted IFNɣ in the presence of urelumab (Figure 2(d)). These results were also seen in independent series of coculture experiments with the KMS28- BM cell line (Supp. Figure 2B-D) that also express daratumumab-recognizable CD38 (Supp. Figure 1). The anti-SLAMF7 mAb elotuzumab, approved for multiple myeloma treatment 13 showed similar effects upregulating CD137 expression upon coculture of NK cells with multiple myeloma cell lines expressing SLAMF7 (Supp. Figure 5A-B) resulting in enhanced of CD25 expression and IFNɣ production when urelumab was added to the cultures (Supp- Figure 5C-D).
We repeatedly tried to see if such CD137-stimulated NK cells would perform better in ADCC assays. [10][11][12] However, in these types of cultures such an enhancement of the cytotoxicity triggered by daratumumab could not be observed (Figure 3(a)). In line with this, CD16 engagement by myeloma cell-coated daratumumab resulted in downregulation of surface CD16 expression but again, this effect was not modified by urelumab (Figure 3(b)). NK-cell degranulation was also studied in such co-cultures, where only minor increases of CD107a surface expression on NK cells could be observed under the influence of urelumab (Figure 3(c)). Thinking that this negative result was due to insufficient CD137 induction, we performed experiments with NK cells that had been co-cultured for a week in the presence of an irradiated lymphoblastoid cell line plus recombinant IL-2, which showed high levels of CD137 expression and overall activation (Supp. Figure 6A). The purity of isolated NK cells was verified in each case (Supp. Figure 7B). ADCC mediated by activated and expanded NK cells was of far greater intensity than that performed by freshly harvested NK cells (Supp. Figure 7C). However, even when using these activated NK cells, the effect of adding urelumab to ADCC cytotoxicity cultures against daratumumab-coated MM1S-GFP-Luc cells was again negligible (Supp. Figure 6B-C). Our data indicate that at least in this myeloma model, enhancement of NK mediated ADCC by urelumab is not mediating any prominent effect, albeit CD137 ligation clearly augmented NK-cell production of IFNɣ.
To model in vivo the effects of urelumab on NK antitumor activity upon daratumumab treatment, we used a model of NSG mice reconstituted with human NK cells. In this setting, mice intravenously injected with luciferase-transfected MM1S-GFP-Luc cells were given 2 × 10 6 human NK cells and daratumumab i.v. on the same day. Twenty-four hours later, mice were dosed with urelumab, once having given time for CD137 induction. A purity check of the administered NK cells is shown in supplementary figure 7A. Twenty four hours after administration, MM1S-GFP-Luc and NK cells were present in bone marrow. Human NK cells could also be found in blood, liver and spleen of the mice (Supp. Figure. 8).
Comparing the different treatment groups by measuring tumor-emitted bioluminescence and survival, it was clear that the daratumumab and NK + daratumumab groups showed disease control over the untreated group and the group receiving NK cells. Importantly, the group given triple treatment with NK transfer, daratumumab and urelumab had significantly better tumor control (Figure 4(a-b)) and survival advantage over the daratumumab + NK group (Figure 4(c)). Similar experiments were carried out with the lymphoblastoid cell line+IL-2 activated NK cells. In these experiments, the group co-treated with urelumab again showed a certain degree of survival advantage (Supp. Figure 9A). Of note, daratumumab as a single agent without NK cell transfer exerted some antitumor effects that could be related to ADCC performed by murine macrophages, but this effect is not enhanced by urelumab in ausence of human NK cells (Supp. Figure 9B).
Considering these results as a whole, we conclude that even in the absence of effects on CD8 T cells, 14 addition of urelumab to daratumumab treatment may improve therapeutic outcomes. Given the fact that urelumab at full doses causes liver inflammation in a subset of patients, 15 testing other safer or targeted CD137 agonists in the same setting will be important. It is surprising that the pharmacological interaction mechanism between daratumumab and urelumab seems not to be a mere enhancement of the effector phase of ADCC as proposed by H. Kohrt et al. [10][11][12] but probably involves a higher level of intricacy. This could involve the immunoregulatory and proinflammatory functions of NK cells. We cannot exclude agonist effects of CD38 ligation on NK cells themselves that could potentially further foster their activation. 16 Indeed, we confirmed the binding of daratumumab to a large fraction of NK cells (data not shown).
Furthermore, in immunocompetent patients, anti-CD137 and anti-CD38 mAb may act on their target molecules on T lymphocytes thereby potentially enhancing and consolidating therapeutic efficacy through amplification of adaptive immunity. In this regard, we have previously reported enhancement of CD8 + T-cell immunity by agonist anti-CD 137 mAbs against syngeneic myeloma mouse models. 17 Our model system in NSG mice permits us to focus on NKmediated mechanisms since complement activation and macrophage functions are deficient in this mouse model.  Figure 2(a) corresponding to five different healthy donors. * = p < 0.05, ** = p < 0.01 and *** = p < 0.001 in a paired t-Test or one-way ANOVA test.
Additionally, urelumab decreases antibody responses through a route dependent on T follicular helper cells. 18 This may help reduce anti-drug antibodies.
Clinical trials are ongoing testing the combination of lowdose urelumab with elotuzumab, another anti-myeloma ADCC-inducing antibody (NTC02252263). Effects of the daratumumab + urelumab combination in a xenografted multiple myeloma model advocate for a clinical trial to sustain and enhance with CD137 agonists the outstanding effects produced by daratumumab against myeloma.

Isolation and activation of NK cells
Blood samples were obtained from healthy volunteers through the Banco de Sangre y Tejidos de Navarra while the fulfilling the requirements of the Clinical Research Ethics Committee of the Navarre Government following the 2016.143 protocol. PBMCs from buffy coats were isolated after gradient centrifugation with Ficoll-Paque (GE Healthcare), and NK cells were isolated by negative selection using magnetic beads and LS selection columns following the manufacturer's protocol (Miltenyi Biotec).
When indicated, NK cells were cultured during 16h in plastic culture plates coated by daratumumab or irrelevant human IgG. For this, a solution of antibody (10µg/mL) in DPBS without calcium or magnesium (Gibco, Invitrogen) was added to culture plates and maintained at least 4h at 37ºC. Then, plates were washed at least three times with fresh DPBS. NKs were cultured in antibody-coated plates with the different treatments in NK MACS Medium (Miltenyi Biotec) supplemented with 5% human male AB serum (Sigma-Aldrich) and 500 IU/mL of IL-2 (Aldesleukin, Novartis) for 16h. Then, cells were harvested and used for study by flow cytometry or for subsequent experiments.
For NK long-term activation and expansion, NK cells were cultured with allogeneic 20Gy irradiated PBMCs in a 10:1 ratio and 20Gy irradiated RPMI8866 lymphoblastoid cells in a 4:1 ratio in RPMI 1640 medium with glutamax (Gibco, Invitrogen). The culture medium was also supplemented with 10% of heat-inactivated FBS, 10% human male AB serum, 1mM non-essential aminoacids, 1mM sodium pyruvate, 2x10 −5 M beta-mercaptoethanol, 50U/mL of IL-2, 1µg/ mL PHA (Sigma-Aldrich), 100 IU/mL penicillin and 100 g/ mL streptomycin. Cultures were maintained during 7-14 days and resulting activated NK cells were freshly used or cryopreserved. NK purity was checked in every experiment and found to be over 98%.

Cocultures of NKs and tumor cells
Activated NK cells were cocultured with MM1S-GFP-Luc or KMS28-BM cells in a 1:1 ration in the presence or absence of daratumumab and urelumab (10µg/mL). At 24h, part of the supernatants was collected and frozen for subsequent IFNɣ concentration determination by commercial ELISA (R&D). Part of the cells was used to study the tumor cell lysis percentage by flow cytometry, and the rest of the culture was maintained with brefeldin A during 4h. Cytotoxicity was measured as the percentage of non-viable (ZombieNir + and/or active caspase 3 + ) versus viable (ZombieNir − and active caspase 3 − ) cells among GFP + myeloma cells. At this point, cells were harvested, and CD25, IFNɣ and other molecules expression were studied by flow cytometry. Daratumumab was acquired from Janssen. Elotuzumab and urelumab was obtained from Bristol-Myers Squibb. Beriglobin P (CSL Behring) was used as irrelevant human IgG. were purchased from Biolegend. Anti-CD56 (NCAM16.2), CD16 (3G8) and active Caspase 3 (C92-605) were obtained from BD. Anti-CD3 (UCHT1), anti CD25 (BC96) and anti-IFNɣ (4SB3) were from eBioscience and anti-NKG2A (131411) was purchased from R&D System. To study the expression of CD137 on NK cells previously treated with urelumab, a biotilinated mAb developed by our group was used (5D1). This mAb does not compete with urelumab in binding to CD137. 22 The anti-CD38 (Cy38F2) mAb was from Cytognos. Intracellular staining (to study active Caspase 3 and anti-IFNɣ expression) was performed with the BD Cytofix/ Cytoperm™ solution following the manufacturer's protocol. Dead cells were identified using the ZombieNIR™ kit (Biolegend). FACSCanto II cytometer was used for cell acquisition, and data analysis was performed using FACS DiVa (BD Biosciences) and FlowJo 7.2.1 (Tree Star Inc., San Carlos, CA).

In vivo model
NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice were purchased from The Jackson Laboratory and bred in the animal facility of CIMA of the University of Navarra under the guidelines of Ethics Committee of the center following the 001-17 protocol. For the xenograft model, 3 × 10 6 MM1S-GFP-Luc cells were inoculated i.v. via the caudal vein together with 2 × 10 6 freshly isolated NK cells and 50µg of daratumumab per mice. The following day, 100µg of urelumab were injected i.v. In the activated NK cell model, 5 × 10 6 MM1S-GFP-Luc cells were injected together with 4 × 10 6 activated NK cells, 50µg of daratumumab and 100µg of urelumab per mice at the same time.
Six weeks following tumor inoculation, bioluminescence was measured once every week. For this purpose, mice were anesthetized using ketamine/xylazine and injected with 100 μl D-luciferine i.p. (Xenogen) at a concentration of 30 mg/ml. Mice were placed in the imaging chamber of the PhotonIMAGER Optima system (Biospace labs). A colorscale photograph of the animals was acquired, followed by bioluminescent acquisition. Regions of interest were drawn over the animal image, as well as over regions of no signal, which were used as background readings. Light intensity was quantified using photons/s (RLU). The color-scale photograph and data images from all studies were superimposed using M3 Vision software (Biospace labs).

Statistical analysis
Prism software (GraphPad Software) was used for statistical analysis. For in vitro experiments, differences between groups were verified with t-student, paired T tests or one-way ANOVA tests. Linear regression was used to compare photon emission in bioluminescence experiments. In vivo luciferase expression was analyzed by differences in linear regression after logarithmic transformation of the data. Mice survival was studied with Log-Rank (Mantel-Cox) test. *p < 0.05, **p < 0.01, ***p < 0.001.