Multiple myeloma (MM) cells home to the bone marrow (BM) and adhere to extracellular matrix (ECM) proteins and to bone marrow stromal cells (BMSCs). The close cross talk between MM cells and cells of the non-tumor compartment within the BM has a key role in supporting tumor growth, survival and development of drug resistance. This biological scenario has led to a growing interest in novel drugs, targeting MM cells and/or interfering with their human BM milieu (HuBMM).1, 2 Based on this, appropriate in vivo models that recapitulate the complex interactions occurring between MM and its HuBMM are required for preclinical evaluation of new anti-MM agents. To date, the in vivo study of MM pathobiology and the validation of therapeutic anti-MM agents has been carried out using a variety of models of murine MM or human MM xenografts in immunocompromised mice.3 These models, however, do not replicate the HuBMM. The development of the SCID (severe combined immunodeficiency)-hu model has been an important advance, as it was the first model to recapitulate a HuBMM in mice.4, 5, 6 However, although the SCID-hu system remains a highly relevant model for preclinical investigation, it does have important limitations: (i) restricted availability of human fetal bone chips; (ii) the allogeneic nature of the fetal BM milieu versus MM cells; and (iii) the significant heterogeneity of implanted human bone chips, collected from different individuals at different gestational age, that may produce experimental variability.

With the aim to overcome the limitations of the SCID-hu system, here, we report the development of a novel in vivo human MM model based upon the implantation into a SCID mouse of a three-dimensional (3D) bone-like poly-ɛ-caprolactone polymeric scaffold (PCLS). In this model, we demonstrated efficient coating of the 3D scaffold internal surface with mouse or human BMSCs and successful engraftment of human primary explanted MM cells within their autologous HuBMM in vivo. Using a previously described methodology,7 we synthesized PCLSs cylinders (7 mm in length and 3 mm in diameter) that, at scanning electron microscope (SEM), demonstrated interconnected large (100–300 μm) and small pores (1–10 μm) (Figure 1A), resembling the microarchitecture of a normal human femur adult bone (Figure 1B). PCLSs were stored at 4 °C in ethanol 70% and then were maintained overnight in 50 ml of sterile water before use. In these PCLSs, we first seeded OP9 cells, a mouse BMSC line, to set up the experimental loading conditions. As OP9 cells efficiently adhered to 3D surfaces (Figure 1C), we next evaluated the suitability of these PCLSs to host human BMSCs derived from MM patients. OP9 mouse BMSCs were cultured in α-MEM (GIBCO, Invitrogen, San Giuliano Milanese (MI), Italy) containing 10% heat-inactivated fetal bovine serum. Human BMSCs were obtained with long-term culture (28 days) of bone marrow mononuclear cells (BMMCs) from Ficoll gradient separation of BM aspirates.8 Dynamic seeding of BMSCs into PCLSs was performed in vitro using a suspension of 106 cells in 500 μl of growth medium. A 22-gauge needle on a 2.5 ml syringe was threaded into two ending face of the cylindrical scaffold. A medium flow rate of 500 μl/min and three drawing cycles were carried out on both scaffold ends. Before implantation, PCLSs were incubated in complete medium at 37 °C in 5% CO2 for 24 h to allow cell adhesion on 3D surfaces. Then, the PCLS was surgically implanted subcutaneously into a SCID mouse flank. Chloralium hydrate anesthesia (400 mg/kg) was used during all surgical procedures. Figure 1G shows the flow chart diagram of experimental procedures. As shown in Figures 1D–F, cultured human BMSCs adhered and spreaded among 3D interconnected surfaces, starting 1 week after cell injection, as demonstrated by SEM, hematoxylin and eosin (H&E) staining and confocal laser scanning microscopy. Based on these achievements, we next evaluated the suitability of these BMSCs-coated PCLSs for engraftment of primary MM cells and interleukin-6 (IL-6)/BM-dependent INA-6 MM cell line. CD138+ immunoselected primary MM cells from BM aspirates of patients with newly diagnosed (n=5) or plasma cell leukemia (PCL, n=1), and from peripheral blood of PCL patients (n=2), as well as INA-6 cells were injected in vivo in previously implanted PCLSs that, 3 weeks before, were coated ex vivo with BMSCs. Importantly, although both primary MM cells and IL-6/BM-dependent INA-6 MM cells successfully engrafted in human BMSCs-coated PCLSs, they did not engraft into scaffolds previously coated with murine BM stromal cell line OP9 or empty PCLSs. This finding is in agreement with previous observations in the SCID-hu model, where primary MM or INA-6 cells did not infiltrate mouse bone or other murine organs.5, 6 These observations suggest that species-specific interactions allow MM in vivo growth in our system.

Figure 1
figure 1

(A) SEM analysis of a PCLS ( × 65 (1), × 500 (2), × 650 (3) and × 1200 (4) magnification) shows interconnected large and small pores (white bars=1 mm (1), 200 μm (2), 100 μm (3) and 50 μm (4)). (B) Comparative SEM analysis ( × 150 magnification) between a synthetic PCLS (upper panel) and a surgical sample of human femur adult bone (lower panel) shows similar microarchitecture (white bars=500 μm). Surface morphology was studied by a Leica Cambridge (Stereoscan S440) SEM (Cambridge, UK) at an accelerating potential of 20 kV. (C) H&E staining of adherent OP9 mouse stromal cells at 3 weeks after scaffold implant in mice. (DF) SEM analysis ( × 3500 magnification) (d), H&E staining (e) and confocal laser scanning microscopy (f) show adherence within a PCLS of human BMSCs, obtained from two different patients (pt #3, (a); pt #7, (b) See Table 1), 1 week after implant in mice. (G) Flow chart diagram. Schematic representation of experimental methodology.

With the aim to overcome one of the important limitations associated with the SCID-hu model, we further attempted to engraft primary MM cells within their autologous HuBMM. To achieve this aim, we injected unselected BMMCs, obtained by Ficoll gradient separation of BM aspirate of MM patients, into PCLSs. This unselected BMMC suspension, containing both primary CD138+ MM cells and their autologous BMSCs, was freshly seeded ex vivo in uncoated PCLSs before in vivo implantation of the scaffold in SCID mouse flank (Figure 1g, flow chart diagram). Notably, we achieved a successful engraftment of all patient samples (Figure 2, Table 1 ). Clinical features of donor MM patients and the actual rate of MM engraftment in SCID-synth-hu mice are reported in Table 1. As shown in Figure 2a, retrieved PCLSs from SCID-synth-hu mice injected with CD138+ immunoselected primary MM cells or unselected BMMCs containing primary MM cells, or INA-6 MM cells demonstrated engraftment and filling of 3D spaces by tumor cells, as confirmed by both H&E and CD138 or κ/λ staining. As we observed slow progressive degradation of PCLSs in mice, long-term survival of MM cells and active proliferation were demonstrated by Ki-67 staining 3–4 months after cell injection (Figure 2b). Following 3–4 weeks, MM cell growth was monitored in vivo by enzyme-linked immunosorbent assay detection of human heavy or light chains in mouse sera (Figures 2c and d). Moreover, to assess the suitability of this model for preclinical evaluation of anti-MM agents, we next treated MM-bearing SCID-synth-hu mice intraperitoneally with bortezomib (on days 1, 4, 8 and 11) and dexamethasone (on days 1–4). Such regimen dramatically inhibited MM cell growth in vivo, as demonstrated by the fall of paraprotein levels in mouse sera (Figure 3a) and by the detection of MM cell apoptosis in retrieved PCLSs (Figure 3b), as compared with untreated control mice. These results clearly demonstrated that this model is actually suitable for large-scale in vivo preclinical screening of investigational anti-MM drugs. An additional important observation was the detection of vasculogenetic events in retrieved PCLSs. In fact, H&E and immunohistochemical staining demonstrated vasculogeneic activity together with the presence of a neosynthesized ECM (Figure 3c). Both in BMSCs-coated PCLSs injected with MM cells and in scaffolds directly implanted with BMMCs, we observed vessels of variable size and human-derived endothelial cells as demonstrated by reactivity of a specific anti-human CD31 mAb. The vasculogenetic activity mostly occurred within areas of MM infiltration (Figure 3 c3). These findings suggest a role of human vasculogenesis in our system.

Figure 2
figure 2

(a) Histological and immunohistochemical (anti-CD138/MI15 (Dako, Dako Itala S.p.A., Milano, Italy), anti-κ (Dako) and anti-λ (Dako)) analysis of retrieved PCLSs. Line 1 shows unselected BMMCs; line 2 shows selected CD138+ on heterologous BMSCs; line 3 shows selected CD138+ from PB of a PCL patient on heterologous BMSCs; and line 4 shows IL-6-dependent INA-6 MM cells on heterologous BMSCs. Anti-human κ-light chains and anti-human λ-light chains were used (Dako). Dilutions were 1:15 000 for κ-light chains and 1:10 000 for λ-light chains. (b) Ki-67 staining of MM-bearing PCLSs retrieved 3–4 months after cell injection. (c, d) Kinetics of appearance of paraproteins in sera from representative SCID-synth-hu mice engrafted with samples from MM patients. A code for each patient sample is provided according to Table 1. Left panel includes seven samples from patients with whole paraprotein (ELISA, GenWay Biotech Inc., San Diego, CA, USA); right panel includes two patients with light chain MM (ELISA, Bethyl Laboratories Inc., Montgomery, TX, USA).

Table 1 Clinical features of MM patients and PCLS engraftment
Figure 3
figure 3

(a) SCID-synth-hu mice were treated intraperitoneally with bortezomib (1 mg/kg)+dexamethasone (1 mg/kg). The black arrows indicate the time of treatment. (b) H&E staining of a MM-bearing PCLS, engrafted with primary MM cells, following Bort+Dex treatment showed diffuse stromal and intracellular calcification demonstrating massive apoptosis. (c) (1) H&E staining of a retrieved PCLS showed vessels of variable size (black circles). (2) Immunohistochemistry of a PCLS coated with BMSCs demonstrated reactivity with an anti-CD31/JC70A (Dako) specific for human CD31 (black arrow). Appropriate positive control tissues were added on each automated immunohistochemistry run to confirm antibody specificity. Negative control section included normal rabbit serum as a substitute for the primary antibody. (3) H&E staining of a retrieved PCLS engrafted with BMMCs shows an active site of vasculogenesis within an area of MM cell growth (yellow arrow). (4) H&E staining of a retrieved PCLS coated with BMSCs revealed the presence of a neosynthesized ECM inside the pores.

All together, our findings demonstrate that the SCID-synth-hu is a novel model for in vivo engraftment of human primary MM cells within their autologous adult HuBMM. This model offers several potential advantages as compared with the SCID-hu system that includes the unlimited availability of PCLSs, the ability to dissect biological events within the HuBMM and, most importantly, the engraftment of primary MM cells in a non-fetal adult autologous HuBMM. Moreover, the present model represents the first system for in vivo expansion of human MM cells on a synthetic platform. We conclude that the SCID-synth-hu is a unique tool for large-scale in vivo preclinical evaluation of novel agents targeting MM in an autologous HuBMM and a novel resource for translational research in the experimental treatment of this still incurable disease.