The supplier and reference ID of all the reagents and equipment used in this study are listed in Table 1. The methods described here focus on a small selection of myeloma and osteoblastic cells, but the model can be readily adapted to accommodate other cell lines/types.

PolyHIPE scaffolds were prepared in-house as previously described. ^{20, 22} Briefly, surfactant was dissolved by heating in a mixture of 4-arm polycaprolactone methacrylate, then cooled prior to the addition of a photoinitiator and a solvent blend (40% toluene, 60% chloroform). Full details of the quantities used for a single batch are provided in Protocol 1 below; however, these can be scaled proportionally as required by users. Using a magnetic stirrer, the mixture was mixed at 400 rpm for 3 min at 37 °C, and then 2 ml water was added dropwise over 3 min. The resulting emulsion was mixed for an additional 5 min and then cured in a syringe for 5 min on both sides to produce polyHIPE tubes. This was then washed in 100% methanol for three days, then in water for a further day in order to remove contaminants. The washed polyHIPEs were then dried in a vacuum oven at room temperature overnight prior to sectioning to the desired thickness using a vibratome and then maintained in a dry environment at room temperature until required ( Fig. 1). We observed that a thickness of 250 μm was ideal; however, this thickness could be tailored by users.

Three myeloma cell lines (U266, JJN3, and 5TGM1) and three osteoblastic cell lines (MG-63, hFOB 1.19, and MC3T3) were used ( Table 2). Myeloma cell lines JJN-3 (DSMZ, Germany), U266 (LGC Standards, UK), and 5TGM1 ¹¹ were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (P/S), 1% non-essential amino acids (NEAA; Corning, UK), and 1% sodium pyruvate (SP).

MG-63 (ATCC) and MC3T3-E1 subclone 4 (ATCC) cells were maintained in minimum essential medium (MEM) alpha supplemented with 10% FBS and 1% P/S. Human fetal osteoblasts (hFOB 1.19; ATCC) were maintained in Dulbecco’s Modified Eagle’s medium/nutrient mixture F-12 medium supplemented with 10% FBS, 1% P/S, and 0.3 mg/ml G418 sulphate.

All cells were maintained at 37 °C with 5% CO ₂ in a humidified incubator (95 ± 5% relative humidity), except hFOB 1.19 monocultured cells, which were maintained at 34 °C. The co-cultures were maintained in complete RPMI medium for myeloma cell maintenance.

All cells were routinely tested for mycoplasma contamination prior to use and were confirmed to be mycoplasma-free. Human cell lines were authenticated by short tandem repeat (STR) profiling using ATCC profiling services. Murine cell lines were obtained from established academic or commercial sources. Cells were expanded from early passage stocks and used within 30 passages, relative to the passage number provided by the supplier, to minimise phenotypic drift. Unless stated otherwise, cells were maintained under standard culture conditions and used during the logarithmic phase of growth.

Scaffolds were sterilised by immersion in 100% methanol for 24 h, followed by washing in either phosphate-buffered saline (PBS) or complete media for 24 h, with a minimum of three changes. The scaffolds were then transferred to 48-well plates, and excess liquid was removed. Osteoblast cells were seeded onto the surface of scaffolds in 20 μl of RPMI media, at a density of 50,000 cells, and then incubated for 2 h at 37 °C to allow cell attachment. Following incubation, media was added to the wells containing scaffolds, and the models were maintained as required. Osteoblasts were cultured for 3 days, then the scaffold was transferred to a new 48-well plate, and excess liquid was removed. Myeloma cells were subsequently seeded onto the polyHIPE scaffold surface in the same manner as osteoblasts ( Fig. 2).

Osteoblastic (MC3T3) and myeloma (5TGM1) cells were seeded onto scaffolds as described above and co-cultured for 21 days post-myeloma cell seeding. Scaffolds were then washed in room temperature PBS x3, fixed in 4% PFA for 15 min at room temperature, then washed in PBS x3. Since the scaffolds are auto-fluorescent when imaged using a 405 nm laser, to increase the fluorescent signal of the material, scaffolds were then stained with DAPI (1:1000) for 10 min at room temperature and washed in PBS x3. The scaffolds were then dipped in OCT and placed into a cryomold containing a layer of OCT. This was then submerged in liquid nitrogen for 15 s and quickly transferred to a steel chuck in a cryostat chamber set to −20 °C. Sections were cut to a thickness of 10–20 μm, with 10 μm representing the minimum thickness required to preserve structural integrity, and 20 μm the maximum thickness to ensure sufficient optical resolution and signal clarity for confocal microscopy. Sections were mounted on SuperFrost Plus glass slides at room temperature and air-dried for a minimum of 3 h.

Fluorescent imaging of the scaffolds was performed using a Zeiss LSM980 confocal microscope. Images were captured at x10 or x20 magnification using Plan-Apochromat 10x/0.3 M27 air or 20x/0.8 M27 air objectives, respectively. DAPI, GFP, and DiD channels were excited with 405, 488, and 639 nm lasers, respectively. The laser power, detector gain, and pinhole settings were optimised to maximise the signal while avoiding detector saturation and were kept consistent across samples. Images were acquired using Zeiss Zen 3.5 (blue edition) software.

Osteoblastic (MG-63 or hFOB) and myeloma (JJN3 or U266) cells were seeded onto scaffolds as described above, 24-hours post myeloma cell seeding a drug was added to the culture well and scaffolds cultured for a further 3 days. For JJN3 co-cultures, melphalan (5.8 μM), bortezomib (1.5 nM), or lenalidomide (5 μM) were used at concentrations representing 2D IC25 values. For U266 co-cultures, bortezomib (3 nM), lenalidomide (50 μM), or a combination of both were used at concentrations representing 2D IC50 values. The study design incorporated consistent seeding densities and repeated biological replicates to ensure reproducibility. Randomisation and blinding were not applied in this workflow, as all handling and measurements were performed using standardised protocols to maintain consistency across samples. Randomisation and blinding were implemented for scaffold-based flow cytometry assays, where individual scaffolds were processed and measured separately.

AlamarBlue ^® assessment

For JJN3 cultures, mixed-population cell viability was assessed using the alamarBlue ^® assay ( Fig. 4A). 2D controls were maintained and treated in a similar manner in 6-well plates. Following the seeding of cells on scaffolds, at the indicated time points, scaffolds were moved to new wells containing fresh medium. The movement of scaffolds to new wells was necessary to ensure that measurements were reflective only of cells growing on and within the polyHIPE scaffolds, rather than those that failed to attach/fall off the scaffold. Once removed, 10% (v/v) alamarBlue ^® reagent was added to wells, and the scaffolds were incubated at 37 °C for 4 h. Following this, scaffolds were removed from the wells, and the fluorescence produced by the reduction of the alamarBlue ^® reagent in the remaining media was measured using an EnSight Multimode Plate Reader. Relative fluorescence intensity, commonly used as a proxy for cell viability in drug response assays, was used as an indicator of cell metabolic activity. Values were normalised to those of vehicle controls and expressed as percentages.

Flow cytometry assessment

For U266 co-cultures, cells were enzymatically removed from scaffolds following a regimen of 10 min wash in PBS, followed by 30 min incubation at room temperature on a shaker with accutase, and 10 min incubation with trypsin-EDTA at 37 °C on a shaker. PBS, accutase, and trypsin suspensions were collected after each step, and the cells were collected by centrifugation at 800xg for 5 min. As a further step to maximise the total number of cells isolated from scaffolds, the scaffolds were cut into segments and centrifuged in PBS at 800xg for 5 min.

Whole myeloma cell and DMC (DiD ^Hi) cell population viability was assessed by flow cytometry (LSRII flow cytometer) following 30 min of incubation with live/dead violet at room temperature.

This study did not involve human participants or human tissue samples requiring ethics approval. All human cell lines used in this study were obtained from commercial or academic sources and were used in accordance with institutional guidelines.

Unless otherwise stated, all experiments were performed in which a single scaffold represented an experimental unit (n). Data were analysed using GraphPad Prism v9.0 (GraphPad Software, USA) and are expressed as the mean ± standard deviation. Pairwise comparisons were performed using two-tailed Student’s t-tests. Statistical significance was defined as P < 0.05.

PolyHIPE scaffolds were synthesised according to the protocols of Aldemir Dikici et al. ¹⁹ and Jackson et al., ²⁰ by forming a water-in-oil high internal phase emulsion (HIPE), with a 40:60 aqueous-to-oil phase ratio to ensure high porosity, which is then photopolymerised to a polyHIPE. Solvents and uncured materials were extracted prior to sectioning the discs and sterilisation for cell culture.

Steps: 1.

Prepare PCL-M emulsion (~ 30 minutes) a.

In a glass vial, 0.40 g of PCL-M and 0.04 g of surfactant were mixed together. The surfactant was gently heated and allowed to cool to room temperature.

Osteoblast cells (hFOB 1.19) were pre-seeded onto the scaffolds to generate the osteoblast component of the 3D model. Pre-seeding prior to myeloma cells facilitates sufficient attachment and migration into the scaffolds such that myeloma cells do not outcompete osteoblasts because of their shorter doubling times. Osteoblasts were cultured for three days prior to the addition of myeloma cells, but osteoblasts could be cultured for longer periods (we cultured osteoblasts alone for up to 28 days).

Steps: 1.

Scaffold preparation (~15 minutes) a.

Scaffolds were removed from the wash medium 10–15 minutes prior to seeding hFOB 1.19 cells onto scaffolds, then placed into individual wells of a 48-well tissue culture plate.

Vybrant DiD cell labelling solution (DiD) is used to indirectly track cell proliferation, since non-proliferative cells retain a strong dye signal, whereas proliferating cells do not. Therefore, we used DiD to identify DMCs (e.g. JJN3), since these cells retain a strong DiD signal. We used DiD, but other similar dyes could also be used. The myeloma cells were labelled prior to seeding onto scaffolds; we did not observe DiD transfer onto the polyHIPE material, but it is possible for DiD to transfer to osteoblasts.

Steps: 1.

Myeloma cell preparation (~15 minutes)

Following osteoblast pre-seeding on the scaffold for three days, (DiD labelled) myeloma cells (JJN3) were added to the model system. We have cultured myeloma cells with osteoblasts for up to 21 days and advise leaving myeloma cells for at least 24 h for shorter study periods.

Steps: 1.

Scaffold preparation (~10 minutes)

Several methods can be used to process the scaffolds for analysis. Below are examples of how we processed the scaffolds; however, this is not an exhaustive list. 1.

Dissociation of cells from scaffolds (e.g. for flow cytometry)

1.

Uneven cell distribution or poor seeding efficiency

Issue: Cells failed to adhere evenly or settle onto the plastic well plate instead of the scaffold.

Steps to troubleshoot: •

Ensure that the scaffolds are dry on the surface before seeding to help the droplets stay in place. This can be achieved by placing scaffolds at 37 °C if previously performed only at room temperature. Alternatively, the drying time could be increased.

Steps to troubleshoot: •

Add medium slowly down the side of the well after the initial 2-hour seeding phase.

Steps to troubleshoot: •

Cells were allowed to settle by initially seeding in a smaller volume (10 μl) before adding larger volumes of media.

Steps to troubleshoot: •

Optimise enzymatic digestion protocols (e.g., use of collagenase/accutase or increased incubation length). This will be cell dependent, but the use of accutase at 30 min, followed by trypsin for 5 min, has shown no effect on cells in our hands.

Steps to troubleshoot: •

Include scaffold-only controls to assess background autofluorescence.

Confocal imaging of the scaffold cross-sections confirmed that the scaffolds successfully supported the co-culture and spatial organisation of myeloma and osteoblast-lineage cells. After 21 days of co-culture of 5TGM1 and MC3T3 cells on the scaffolds, both cell types were able to establish networks throughout the scaffold structure. Myeloma cells grew in direct contact with osteoblasts ( Fig. 3A), adjacent to osteoblast clusters ( Fig. 3B), or near osteoblast populations without direct contact ( Fig. 3C). These observations indicate that the scaffolds permit both direct contact-dependent and indirect paracrine interactions, resembling the cell-to-cell interactions observed within the endosteal niche in vivo. ^{6, 23}

Having confirmed the co-culture architecture within our scaffolds, we next assessed drug responses on the scaffolds and compared this with standard 2D in vitro cultures. JJN-3 myeloma cells were co-cultured with either hFOB 1.19 or MG-63 osteoblastic cells and treated with 2D IC25 concentrations of melphalan, bortezomib, or lenalidomide: three standard-of-care myeloma therapies with known mechanisms of action. Whole scaffold populations were assessed using alamarBlue® assays ( Fig. 4A) and showed distinct responses between the osteoblast types and culture methods ( Fig. 4B). In hFOB co-cultures, cells were significantly more sensitive to melphalan on our scaffolds (3D) compared to 2D, whereas MG-63 co-cultures displayed a similar but non-significant trend. Bortezomib treatment had greater cytotoxicity in 3D MG-63 co-cultures, while hFOB co-cultures showed no difference. In contrast, lenalidomide sensitivity was reduced in 3D cultures for both osteoblast co-cultures. Despite this, for any given drug and culture format (e.g., melphalan treatment in 2D), there were no significant differences in cytotoxicity between MG-63 and hFOB co-cultures. This suggests that the sensitivity of each drug was altered by microenvironmental cues, particularly 2D vs. 3D environments, whereas the osteoblast type contributes more subtle effects that become apparent only when comparing across culture formats. Notably, alamarBlue® measures the conversion of resazurin to resorufin in all viable cells (myeloma and osteoblasts); therefore, these experiments do not allow definitive conclusions to be drawn for a specific cell type.

To investigate the effects of the drugs on distinct myeloma populations, U266 myeloma cells were treated in 2D and 3D co-cultures (with hFOB or MG-63 cells) with bortezomib, lenalidomide, or a combination of both ( Fig. 5). Flow cytometric analysis again showed that in 3D cultures, viability was significantly reduced following bortezomib or combination treatment compared with vehicle controls, whereas no significant differences were observed in 2D cultures ( Fig. 5A-D).

When DMCs (DiD ^Hi) were analysed separately in 2D and 3D cultures, some differences were observed. In 2D hFOB co-cultures, combination treatment led to significantly fewer DiD ^Hi cells compared to the vehicle ( Fig. 5E), while no significant differences were observed in 2D MG-63 co-cultures ( Fig. 5F). However, in 3D cultures, both hFOB and MG-63 co-cultures had significantly fewer DiD ^Hi cells in bortezomib-treated samples compared with vehicle ( Fig. 5G-H), demonstrating reproducibility on our scaffolds with different osteoblastic cells, although it should be noted that fewer cells were retrieved from hFOB co-cultures on scaffolds than MG-63 co-cultures.

The polyHIPE scaffold provided a 3D, highly porous environment suitable for modelling interactions between osteoblasts and myeloma cells under physiologically relevant conditions. Its architecture supports high-resolution imaging and fluorescent tracking of proliferative and DMCs, visualisation of osteoblast networks, and co-localisation studies using confocal microscopy. Sequential seeding allows investigation of dormancy induction and osteoblast-myeloma cell interactions within a bone-lining niche-like context. The scaffold also serves as a platform for functional assays, including pharmacological testing, thereby supporting early-stage drug evaluation. Its modular design allows adaptation to different cell types, co-culture combinations, and assay formats, bridging the gap between conventional 2D culture and in vivo models, while supporting early-stage therapeutic evaluation. Below are examples of how we used some of these methods in our study.

This protocol describes the step-by-step use of a novel 3D scaffold model that replicates the key features of the BM microenvironment to study myeloma cell dormancy and drug responses. It offers a scalable, reproducible, and adaptable platform that minimises reliance on animal models in line with 3Rs principles. The validation of this model against existing in vivo data is ongoing. However, we demonstrated the compatibility of our model with human and mouse osteoblastic cells as well as human and mouse myeloma cells. We also demonstrate a range of downstream analysis methods using our model, which widens the potential applications of this model. For instance, confocal imaging allowed us to validate that the polyHIPE scaffolds were capable of recreating the spatial organisation, facilitating direct and indirect contact between osteoblasts and myeloma cells, which is characteristic of the BM niche. These interactions are important and difficult to model simultaneously in 2D in vitro systems. Thus, the ability to visualise these interactions on our scaffolds underscores their suitability for studying cell-to-cell interactions and relationships.

Through drug testing on our model compared to 2D in vitro cultures, we were able to confirm the influence of the 3D structure, which largely increased drug sensitivity. Importantly, the use of a range of cell lines in our model reduces the need for animal-derived materials and increases the adaptability of our model. Using a range of osteoblast cell lines for co-cultures, we were able to confirm the ability to derive similar conclusions across several osteoblast phenotypes, which are important to consider as they may influence therapeutic responses. The use of hFOB and MG-63 cells, which are two commonly used osteoblastic cell lines in research laboratories, was interchangeable, even with different myeloma cell lines. Importantly, the similarities observed on our scaffold when using different osteoblast types were not consistently replicated in 2D culture, highlighting the detrimental effect of 2D culture compared with 3D culture. At the cellular level, flow cytometry confirmed that DMCs could be targeted on our scaffolds, mirroring observations made previously using in vivo models ⁶ and emphasises our model’s capacity to study dormancy-associated resistance mechanisms. Ongoing work aims to further validate our 3D model using in vivo data. We hope to further explore the integration of patient-derived samples to enhance the clinical relevance of the model.

Compared with other 3D models, our polyHIPE scaffold offers several practical and biological advantages. Hydrogel-based or collagen scaffolds often lack the structural rigidity and pore interconnectivity needed for long-term co-culture or imaging, while microfluidic “bone marrow-on-a-chip” systems are technically complex and difficult to scale for parallel experiments. PolyHIPE scaffolds bridge this gap, offering a comparatively higher-throughput system that combines stability, tunability, and optical accessibility with compatibility for imaging, biochemical, and flow cytometric analyses. Thus, it is a suitable model for preclinical application. Importantly, the model aligns with NC3Rs 3Rs principles. By reproducing key aspects of the BM microenvironment in vitro, the scaffold enabled the evaluation of cellular dormancy and drug responses without relying on murine models.

Based on the work presented here, we suggest that scaffold-based assays could replace small pilot in vivo studies, typically with ~10–20 animals per study. Widespread adoption of this model has the potential to reduce the use of hundreds of animals annually across the field, minimising the experimental burden of research while maintaining translational relevance.

Finally, the adaptability of the model extends its potential beyond myeloma. By varying cell types, ECM coatings, or scaffold composition, it can easily be adapted for studying broader aspects of myeloma (aside from dormancy) or for studying other cancers such as breast or prostate dormancy. Adoption by other researchers can be aided by sourcing pre-made scaffolds from our in-house facility; however, the potential to manufacture the scaffolds within separate laboratories requires specialised equipment (e.g., vibratome), which could limit the uptake of the model.