Expression of CD38 in myeloma bone niche: A rational basis for the use of anti-CD38 immunotherapy to inhibit osteoclast formation

It is known that multiple myeloma (MM) cells express CD38 and that a recently developed human anti-CD38 monoclonal antibody Daratumumab mediates myeloma killing. However, the expression of CD38 and other functionally related ectoenzymes within the MM bone niche and the potential effects of Daratumumab on bone cells are still unknown. This study firstly defines by flow cytometry and immunohistochemistry the expression of CD38 by bone marrow cells in a cohort of patients with MM and indolent monoclonal gammopathies. Results indicate that only plasma cells expressed CD38 at high level within the bone niche. In addition, the flow cytometry analysis shows that CD38 was also expressed by monocytes and early osteoclast progenitors but not by osteoblasts and mature osteoclasts. Indeed, CD38 was lost during in vitro osteoclastogenesis. Consistently, we found that Daratumumab reacted with CD38 expressed on monocytes and its binding inhibited in vitro osteoclastogenesis and bone resorption activity from bone marrow total mononuclear cells of MM patients, targeting early osteoclast progenitors. The inhibitory effect was not observed from purified CD14+ cells, suggesting an indirect inhibitory effect of Daratumumab. Interestingly, all-trans retinoic acid treatment increased the inhibitory effect of Daratumumab on osteoclast formation. These observations provide a rationale for the use of an anti-CD38 antibody-based approach as treatment for multiple myeloma-induced osteoclastogenesis.


INTRODUCTION
Multiple myeloma (MM) is an incurable plasma cell (PC) malignancy leading to osteolytic bone disease, due to an increased formation and activity of osteoclasts (OCs) and a decreased osteoblastogenesis [1][2][3]. The relationship between MM cells and the bone cells, OCs and osteoblasts (OBs), plays a critical role both in the progression of MM and in the development of osteolytic bone lesions. Malignant PCs highly express surface molecules involved in the adhesion to the bone marrow (BM) microenvironment cells and in the development of bone lesions [4,5].

Research Paper
Oncotarget 56599 www.impactjournals.com/oncotarget Among the adhesion molecules, CD38 is highly expressed by MM cells [6][7][8][9] although its pathophysiological role in MM and in MM-induced bone disease is apparently more complex. CD38 is a 45-kDa type II transmembrane glycoprotein, widely expressed by several cell types [8], which plays a dual role as a receptor and ectoenzyme [10]. It is involved in T cell activation and proliferation, B cell differentiation and cell adhesion through the non-substrate ligand CD31 [11]. Moreover, CD38 acts as metabolic sensor that converts NAD+ to cADPR and ADPR and NADP+ to NAADP+, according to pH status. [12] CD38 acts in conjunction with the other ectoenzymes CD73 and CD203a, in the alternative axis of extracellular production of the immunosuppressive factor adenosine (ADO), bypassing the canonical pathway mediated by CD39 [13]. In addition to the transmembrane arrangement, a CD38 soluble form also exists, probably as a result of enzymatic cleavage of the cell-surface protein, both in normal and pathological fluids [14] and in exosomes [15]. CD38 is also involved in the remodelling of the adult skeleton in mice [16], being expressed on murine OCs [1,17] and OBs [18]. Moreover, Sun L et al. showed that rabbit OCs expressed CD38 on plasma membrane, with ADP ribosyl cyclase activity, and an anti-CD38 agonist antibody inhibited bone resorption [17]. However, the expression of CD38 by human OCs and OBs has not yet been reported.
Daratumumab (DARA), a high-affinity human IgG1κ anti-CD38 monoclonal antibody (mAb), showed encouraging results in the treatment of patients with relapsed or refractory disease [19][20][21]. It has a broadspectrum killing activity that includes complementdependent cytotoxicity (CDC), antibody-dependent cell-mediated cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), apoptosis and, at least in part, modulation of CD38 enzymatic activity [22]. A study from Nijhof IS et al. [23] showed that DARA-induced CDC and ADCC were strictly associated to the level of CD38 and the pre-treatment with all-trans retinoic acid (ATRA) up-regulated CD38 expression in MM cells, enhancing DARA effects in a humanized MM mouse model [23].
Microvesicles (MVs) are a class of extracellular vesicles shed by different cell types under physiological conditions. Moreover, patients with different types of cancer were reported to have high number of circulating MVs suggesting their direct involvement in modulating cell communication within the tumor microenvironment. More importantly, the hypoxic microenvironment may accelerate the release of MVs into the surrounding environment promoting tumor growth, invasion and angiogenesis [24,25]. Recently, Horenstein AL et al. [26] have shown that DARA ligation on MM cells was followed by aggregation, polarization, and release of MVs, extrusions of cell membrane that bear CD38 on the surface, whose fate needs to be defined [26].
Currently, the expression of CD38 and its related ectoenzymes by OCs and OBs in MM BM niche and the effects of DARA on MM-induced bone remodelling alterations are still unknown and were investigated in this study.

Expression profiling of CD38 in the bone niche of MM patients
The expression pattern of CD38 was evaluated by flow cytometry on primary CD138+ purified from 16 MM patients, human myeloma cell lines (HMCLs) and microenvironment cell lines. All primary MM cells expressed CD38 at high level, as well as HMCLs (except for INA-6). The stromal cell line human stromal cell line (HS5) expressed CD38 weakly in contrast to human osteoblastic cell lines human pre-osteocytic cell line (HOB-01) and human osteoblast-like cells (HOBIT) which did not express CD38 at surface level ( Figure 1A). On the other hand, CD38 expression was present at cytoplasmic level in the same cell lines ( Figure 1B).
The immunohistochemistry analysis performed on bone biopsies of MM patients confirmed the immunophenotype data. Specifically, MM cells expressed CD38 at high level ( Figure 2A-2B), whereas OBs ( Figure 2B, 2ii), OCs ( Figure  2B, 2iii-iv) and endothelial cells did not. Moreover, the expression profiling of the nonsubstrate ligand CD31 and that of other ectoenzymes of the adenosinergic pathway, as CD39, CD73 and CD203a was also evaluated. Results are reported in Supplementary Tables 1, 2

CD38 is expressed by monocytes and early OC progenitors but lost during the in vitro osteoclastogenesis
We evaluated CD38 expression by monocytes and during in vitro osteoclastogenesis from CD14 + cells, either at cytoplasmatic levels by western blotting or surface levels by flow-cytometry. We found that both monocytes and mature OCs expressed CD38 at cytoplasmatic level ( Figure 3A). Conversely, the flow-cytometry analysis showed that monocytes were positive for CD38 and that its expression decreased on day 7, with a further reduction on day 14 of OC differentiation ( Figure 3B, left panels) suggesting that mature OCs lost CD38 surface expression. Interestingly, CD38 levels increased during in vitro osteoclastogenesis in the presence of high concentration (20 nM) of ATRA but not at low concentration (0.1 nM) ( Figure 3B).
Confocal analysis confirmed that mature OCs did not express CD38 on their surface ( Figure 4A). Lastly, we confirmed that OBs were negative for surface CD38 ( Figure 4B).

DARA inhibited OC formation, targeting early OC progenitors
Considering the CD38 expression profiling, we checked the ability of DARA to bind monocytes compared to other peripheral blood (PB) cell populations. We performed a flow cytometry assay with FITC-conjugated DARA on Healthy donor (HD) PB mononuclear cells (PBMCs) and demonstrated that DARA binds monocytes and NK cells. However, DARA binding was not detectable on T cells and B cells ( Figure 4C), consistent with the CD38 expression previously described on monocytes.
Thereafter, to determine the potential effect of DARA on osteoclastogenesis, we performed in vitro OC differentiation from mononuclear cells (MNCs) of 13 MM patients, in the presence of DARA (or isotype control IgG) at seeding (day 0) or after 10 days of culture (day 10). When compared with cultures with IgG alone, the presence of DARA reduced the number of OCs when added on day 0 (two-tailed, paired Student's t-test, DARA 10 µg/ml vs IgG and DARA 25 µg/ml vs IgG: p < 0.0001) ( Figure 5A-5B). On the contrary, DARA did not inhibit OCs formation when it was added on day 10, indicating the lack of effects on late OC progenitors and mature OCs ( Figure 5A-5B). The inhibitory effect of DARA on OCs formation was not observed when OCs were differentiated from purified CD14 + cells ( Figure 5C), consistent with DARA mechanisms of action in MM cells mediated by effector cells [22]. Treatment with DARA also decreased the area of osteoclast bone resorption pits and OC resorption activity ( Figure 6) after both 14 (two-tailed, paired Student's t-test, DARA 10 µg/ml vs IgG: p = 0.0004; DARA 25 µg/ml vs IgG: p = 0.0003) and 21 days of treatment (twotailed, paired Student's t-test, DARA 10 µg/ml vs IgG: p = 0.0002; DARA 25 µg/ml vs IgG: p = 0.0001).
In addition, we found that the inhibitory effect of DARA in OC differentiation was not mediated by soluble factors secreted by MM cells, since the conditioned medium (CM) of HMCLs previously pre-treated with DARA did not inhibited OC in vitro formation (data not shown). Lastly, we also investigated the potential effect of DARA-induced MVs on OC formation and we performed in vitro osteoclastogenesis from MM MNCs, in the presence of MVs isolated from the HMCLs RPMI-8226 and JJN3, treated with DARA 200 µg/ml or IgG. No significant differences in OC number were observed in the presence of DARA-induced MVs compared to IgGinduced MVs (Supplementary Figure 2).

ATRA treatment increased DARA effects on OC formation
Since literature data showed that ATRA improves DARA-mediated ADCC and CDC against MM cells [23], we investigated the effect of DARA in combination with ATRA on in vitro osteoclastogenesis. Initial experiments were performed to evaluate the effect of ATRA (0.01 nM-200 nM) on OCs differentiated from MNCs of MM patients for 21 days. ATRA caused a dose-dependent inhibition of osteoclastogenesis with a minimal effect at Oncotarget 56601 www.impactjournals.com/oncotarget 0.1 nM. Indeed, higher concentration induced a drastic reduction of osteoclasts (Supplementary Figure 3). In order to assess a possible combinatory effect, we used ATRA at 0.1 nM and 20 nM for further experiments.
OC number was significantly reduced in the combined treatment condition (DARA 25 µg/ml and ATRA 0.1 nM) compared to the single agent treatment (one-way ANOVA, DARA 25 µg/ml plus ATRA 0.1 nM vs DARA 25 µg/ml: p < 0.01; DARA 25 µg/ml plus ATRA 0.1 nM vs ATRA 0.1 nM: p < 0.01) ( Figure 7A-7B). On the other hand, the treatment with ATRA at 20 nM showed a drastic effect on OCs which made the study of the effects in combination with DARA impossible. Yet, DARA 25 µg/ml plus ATRA 0.1 nM was used to treat purified CD14 + cells under osteoclastogenic conditions. The treatment did not affect the number of OC compared with the single agent treatment confirming the importance of effector cells in the system.

DISCUSSION
The use of the CD38 molecule as a target for antibody-mediated treatment of MM [27] provided an opportunity to access evidence from in vivo applications. The anti-MM effect of DARA has been widely described. [19] Some issues, stemmed from the fact that the target molecule is expressed not only by the tumor but also by effectors and inhibitory cells, are still unresolved [28]. Several studies have reported the CD38 involvement in bone remodelling, in mice and rabbit models [16,17], where both OBs and OCs expressed CD38. However, reports on CD38 expression by human OBs and OCs are still very limited [18,[29][30][31].
It has been reported that CD38 is also involved, in conjunction with other ectoenzymes such as CD73 and CD203a, in the alternative production of ADO. Horenstein AL. et al. [32] recently have shown that this ectoenzymatic network is active even in MM bone niche and that ADO levels correlate with disease aggressiveness and ISS staging of MM patients [32].
The study firstly focused on the evaluation of the expression profiling of CD38 on primary CD138 purified from MM patients. All MM samples expressed CD38 as well as HMCLs. We observed that human stromal and osteoblastic cell lines did not express CD38. However, OBs expressed CD38 at cytoplasmatic levels consistent with CD38 localization in endoplasmic reticulum and nuclear membrane, as found in osteoblastic cell lines by Sun L et al. [18]. Moreover, a study from Romanello M et al. [29] showed that OBs display ADP-ribosyl cyclase/ CD38 activity and NAD+ stimulation inhibited cell growth, markedly altered cell morphology, and induced Oncotarget 56603 www.impactjournals.com/oncotarget significant increases in alkaline phosphatase activity and osteocalcin mRNA, indicating that the nucleotide may act as a differentiation signal [29]. The analysis of gene expression of CD38 by MM PCs did not show any correlation with the presence of osteolytic lesions in MM patients (Toscani D et al unpublished data).
In line with the literature [6,7], the immunohistochemistry data showed that CD38 was expressed at high levels by PCs of patients with monoclonal gammopathies. On the other hand, OBs and OCs were negative for CD38 expression. MM cells also expressed CD31, CD39 and CD73 at variable levels, as previously reporte [10,33,34]. Interestingly, we found that OBs were positive for CD73 but expressed low levels of CD39 and did not express CD31. Previous studies showed that osteoblastic differentiation of hMSCs was induced by ATP degradation mediated by CD73 and CD39 expression on hMSC membrane [35], and that osteoblastogenesis is characterized by a stability of surface CD73, an increased CD203a expression and a lack of cell surface expression of CD38 and CD157 [36]. Lastly, endothelial cells express high level of CD31 and CD39, CD73 at variable levels while express neither CD38 nor CD203a. The distribution of the analyzed ectoenzymes within the MM microenvironment could clarify the involvement of bone cells in the ADO-mediated stimulation of growth and survival of MM cells. Thereafter, we investigated the possible role of CD38 in human OC differentiation by the analysis of CD38 expression on monocytes and OC. Both cell types expressed CD38 in the cytoplasm in line with literature data [30,31]. Faust J et al. [30] found that CD38 was expressed by OC-like cells, whereas a more recent study reported that mature OCs express CD38 [31]. Moreover, we checked flow cytometry expression of CD38 during in vitro osteoclastogenesis from CD14 + cells. We found that monocytes expressed CD38 at seeding but it was lost during OC differentiation. Confocal microscopy confirmed    [37] reported on macrophage differentiation. Overall these results indicate that, other than in PCs, the cell surface CD38 expression is limited to early osteoclast progenitors. Consequently, these cell types could be the target of anti-CD38 therapeutic Abs.
DARA, a human anti-CD38 IgG1κ antibody, has shown encouraging results in the treatment of MM [19,22,38,39]. The main mechanisms by which DARA exerts its anti-myeloma effect are ADCC and CDC [22,38]. However, DARA effects on bone cells and on MM-induced bone remodelling alterations are still unknown. Thus, based on CD38 expression profiling, we evaluated the effect of DARA on osteoclastogenesis. The flow-cytometry analysis showed that DARA binds monocytes and our in vitro experiments indicated that DARA treatment significantly reduced the number of TRAP-positive multinucleated OCs compared to IgG isotype controls when it was added on day 0 but not on day 10, consistent with the expression profiling of Oncotarget 56607 www.impactjournals.com/oncotarget CD38 observed by flow cytometry. Furthermore, DARA significantly reduced the area of osteoclast bone resorption pits and OC activity. Interestingly, we showed that DARA inhibited OC formation only when OCs were differentiated from total MNCs but not from isolated CD14 + cells, suggesting that DARA effect on osteoclastogenesis was mediated by the effector cells, as shown for the anti-MM activity [22]. In support of our findings, a recent study from An G et al. [31] showed that a different humanized Ab specific for CD38, namely SAR 650984, had no effect on osteoclastogenesis from isolated CD14 + cells [31].
More recently, Horenstein AL et al. [26] have shown that DARA ligation on MM cells was followed by aggregation, polarization, and release of MVs, which express several molecules clustered in lipid domains on their surface, including both CD38 and DARA. However, the fate of MVs is still unknown. Chillemi A et al. [40] also showed a tendency of DARA-labeled MVs to cluster around NK cells and monocytes, since DARA shows a high affinity to Fc receptors of immune cell types [40]. Relying on these data, we investigated the involvement of DARAinduced MVs from MM cells in OC formation. We did not find any appreciable effects on in vitro osteoclastogenesis in the presence of DARA-induced MVs compared to IgGinduced MVs. Overall, our in vitro evidence indicates that the anti osteoclastogenic effect of DARA is mainly due to its ability to bind monocytes and OC progenitors, and, consequently, to the activation of effector cells. Since literature data showed that ATRA treatment increased DARA anti-MM effects, [23] we also tested the effect of DARA in combination with ATRA in osteoclastogenesis. High concentration of ATRA (20nM) enhanced CD38 expression on day 7 and 14 of OC differentiation, in line with that reported by Drach J et al. [41] on myeloid cells. On the other hand, low concentration of ATRA (0.1 nM) had no effect on CD38 expression on OC progenitors, but it decreased the number of TRAP positive multinucleate OCs when used in combination with DARA, compared with the single agent treatment supporting the possibility to use ATRA to enhance DARA effect. This combinatory effect is not likely to be correlated to the modulation of CD38 expression on OC progenitors by ATRA that occurs only at high concentration. As reported in the literature, [42] a direct inhibitory effect of ATRA on RANKLstimulated osteoclastogenesis was shown and this mechanism could potentiate the effect of DARA on OC formation. Indeed, the combination ATRA plus DARA did not affect the number of OCs differentiated from purified CD14 + cells thus confirming the importance of effector cells and elucidating the effect of the combination. In conclusion, our findings define the expression profiling of CD38 and a panel of selected ectoenzymes and molecules in the bone niche of MM patients. Moreover, consistent with the CD38 expression profiling, we show that therapeutic anti-CD38 DARA significantly inhibits OC formation by targeting OC progenitors with a potential clinical relevance. On the other hand, CD38 was expressed by OBs at detectable levels, which bona fide rules out the possibility of cytotoxic events delivered by DARA. Taken together, these findings highlight the possibility of a role of CD38 during OC formation, which further supports the use of DARA as a treatment for bone destruction in MM patients. Patient samples were obtained after informed consent, according to the Declaration of Helsinki. The study was included in a larger project on patients with monoclonal gammopathies, approved by the Institutional Ethical Review Board of our Hospital.

Drugs
DARA and isotype control IgG were provided by Janssen Pharmaceuticals (Spring House, PA, USA). ATRA was purchased from Sigma-Aldrich (Saint Louis, MO).

Cells and cell culture conditions PC purification
BM CD138 + cells were purified from total MNCs by an immuno-magnetic method using anti-CD138 mAb coated microbeads (MACS, Miltenyi Biotec; Bergisch-Gladbach, Germany), as previously described [43].

DARA binding on PBMCs
PBMCs isolated from 5 HDs were washed in PBS containing 1% BSA + NaN 3 and incubated with FITCconjugated DARA for 30 min at 4°C. T cells, B cells, NK cells and monocytes were identified by expression of surface markers using the following mAbs produced in the Lab of one of the authors (FM) and conjugated by AcZon (Bologna, Italy), respectively: anti-CD3-PerCP Cy5.5 (clone CBT3G), CD19-APC (clone CB19), CD16-APC (clone CB16) and CD14-APC (clone CB14). Cells were washed in PBS and acquired on FACS. Data were analyzed using FlowJo Software (TreeStar, Ashland, OR).
A sample was considered positive if the target antigen was detected at least in the 50% of cells. Images were captured by DP22 digital camera (Olympus; Hamburg, Germany) and analyzed with the OLYMPUS Stream software, adjusting tone and contrast to ensure the best image quality.

Western blot analysis
These methodologies were detailed in Supplemental Methods section of Supplementary Data.

Immunofluorescence confocal microscopy
Immunofluorescence analysis was carried out to quantify CD38 expression on mature OCs, obtained from HD CD14 + cells. Cells (1.2x10 6 /well) were cultured in 24-well plates on Ø 12mm cover glasses (Thermo Scientific), in αMEM with 10% FBS, rhM-CSF 25 ng/ ml and rhRANKL 60 ng/ml, for 21 days, replacing half medium every 2-3 days. At the end of culture period, cells were reacted at 4°C with the primary mAb anti-CD38 (clone IB4, produced in the Lab of one of the authors (FM) for 1 h. After rinsing with PBS, goat anti-mouse IgG (H+L) conjugated with Alexa Fluor 488 secondary Ab (code n. 115-545-003, Jackson ImmunoResearch Laboratories, West Grove, PA) was added for 30 min at 4°C. The negative control samples were not incubated with primary antibody. After extensive washings, the cells were fixed with paraformaldehyde (2% in PBS, 15 min at 4°C). DNA was visualized after permeabilization with 0.2% Triton X-100 detergent (5 min at 4°C) and successive staining with DAPI. The same protocol was used to quantify CD38 expression on HOBIT cell line. The confocal imaging was performed on a TCS SP5 laser scanning confocal microscope with 4 lasers (Leica Microsystems; Wetzlar, Germany).