Generation and analysis of sicXer and boXer

Production of xerogels was carried out as previously published [17, 25]. Homogeneous suspensions of 30 mg/mL collagen were obtained by dialysis (MWCO 12–14 kDa, Carl Roth) of bovine tropocollagen type I (GfN) against deionized water followed by fibrillation in 30 mM neutral sodium phosphate buffer solution, lyophilization (Christ Alpha1-4 laboratory freeze-dryer), and resuspension in 0.1 M TrisHCl pH 7.4 (Roth) [17]. For sicXer and boXer lyophilized collagen was gamma sterilized at 25 kGy. Silicic acid was prepared by hydrolysis of tetraethoxysilane (TEOS, 99%, Sigma; molar ratio TEOS/water = 1/4) under acidic conditions (0.01 M HCl). Silicic acid and collagen suspension were vigorously stirred by using a vortex mixer to form 800 µL hydrogels with the final composition of 30% collagen and 70% silica. Bortezomib (VELCADE®) was previously added to the collagen suspensions to generate final concentrations of 100 µg (boXer-100), 500 µg (boXer-500), and 2500 µg (boXer-2500) per 1 g silica/collagen xerogel. Hydrogels were stabilized for three days and dried at 37 °C until mass constancy. Monolithic silica/collagen xerogels (diameter: 5 mm, height: 3 mm) were powdered and classified according to suitable particle sizes. siXer-iaf and boXer-iaf (irradiation after fabrication) were gamma sterilized at 25 kGy after fabrication.

Sample preparation for degradation studies

For in vitro degradation studies (Supplementary Fig. S1), monoliths of sicXer and sicXer-iaf were incubated in 1.2 mL alpha-MEM medium (αMEM) containing 10% fetal calf serum (FCS), 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin (all from Biochrom). Medium was changed completely three times a week at days 2, 5 and 7. Supernatants were stored at − 20 °C until analysis via inductively coupled plasma mass spectrometry (ICP-MS).

Quantification of collagen, silica, calcium, and phosphate

For quantification of collagen in the supernatant, an ο-phthaldialdehyd assay was used. In brief, 90 µL of cell culture supernatant was incubated with 100 µL 0.5 mg/mL collagenase from Clostridium histolyticum (Sigma Aldrich) at 37 °C overnight. Then, 20 µL of this were mixed with 200 µL fluoroaldehyd (ThermoScientific) and fluorescence was measured at an excitation and emission wavelength of 340/440 nm (Infinite® M200Pro, Tecan). Different concentrations of bovine collagen in αMEM containing 10% FCS, 2 mM glutamine, 100 U/mL penicillin and 100 µg/mL streptomycin were used for calibration. For quantification of silica, calcium, and phosphate in the supernatant, ICP-MS (IRIS Intrepid II XUV, ThermoFisher Scientific) was used. For analysis preparation, supernatants were diluted in water and HNO3 (Carl Roth). Using to a calibration curve measured from element standards (High Purity), ion concentrations were determined.

Sample preparation for bortezomib, sicXer and sicXer-iaf release studies

For release studies of bortezomib from boXer-100 and boXer-500, individual fractions of < 125 µm, 125–250 µm and 250–710 µm, were investigated. Here, 9 mg of material were incubated in 700 µL of αMEM containing 10% FCS, 2 mM glutamine, 100 U/mL penicillin and 100 µg/mL streptomycin (all from Biochrom). After 1 h, 2 h, 4 h, 8 h, 24 h, 48 h, 72 h, 96 h, 168 h, 336 h, 504 h, and 672 h supernatants were removed completely and stored at −20 °C until analysis via ultrahigh-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS). Quantification of collagen, silica, calcium, and phosphate was performed as described above.

Impact of sicXer and boXer on in vitro osteoblastogenesis and osteoclastogenesisOsteoblastogenesis

Human mesenchymal stromal cells (hMSC) were isolated from bone marrow aspirates kindly provided by Prof. Bornhäuser et al., Medical Clinic I, Dresden University Hospital. hMSC were expanded in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% FCS, 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin (Biochrom). Osteoblastogenesis on sicXer and sicXer-iaf was investigated. Twenty-four well polystyrene culture plates were seeded with 1.7 × 104 hMSC in passage 5 in 800 µL αMEM containing 10% FCS, 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin. sicXer and boXer granules (15 mg) in ThinCerts (1 µm) were immersed in 1 mL αMEM containing 10% FCS for equilibration overnight. The next day, medium supplemented with 50 µM ascorbate (Sigma Aldrich) was added to the adherent hMSC on their initial day (day 0) in presence of xerogel granules in ThinCert (negative control). For osteogenic differentiation, hMSC were treated with 50 µM ascorbate, 5 mM β-glycerophosphate, and 10 nM dexamethasone (all Sigma Aldrich) by day 3. Medium was changed three times a week. For biochemical analysis (see below), cells were washed twice with phosphate-buffered saline (PBS) and frozen at − 80 °C.

OsteoclastogenesisMonocyte preparation

Peripheral blood mononuclear cells (PBMC) were isolated from buffy coats of five healthy donors (German Red Cross). Buffy coats were diluted with equal amount of PBS supplemented with 2 mM ethylendiamintetraacetate (EDTA) and 0.5% bovine serum albumin (BSA; both from Sigma Aldrich) (PBS E/B). Diluted buffy coats were centrifuged for 20 min at 836 g over density gradient 1.077 g/mL (NycoPrep™ 1.077, Progen, Germany) using Leucosep™ tubes (Greiner) without brake. Platelets were removed using density gradient centrifugation over 1.063 g/mL (obtained by dilution of NycoPrep™ 1.077 with PBS E/B) for 15 min at 353 g. Cells were washed with PBS E/B and centrifuged for 8 min at 301 g. Monocytes were isolated from PBMC using negative magnetic separation (Monocyte Isolation Kit II, Miltenyi Biotec) according to the manufacturer’s instructions. Monocytes at a concentration of 1 × 107 were cryo frozen in αMEM containing 10% dimethylsulfoxid (Sigma Aldrich) and 40% heat inactivated FCS until usage.

Osteoclastogenesis

Monocytes were thawed and counted (Scepter™, Millipore, Germany). Monocytes (1.5 × 105) in 800 µL αMEM containing 7.5% heat inactivated FCS, 7.5% human A/B serum (ccpro), 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin, additionally supplemented with 50 ng/mL macrophage colony stimulating factor (M-CSF; R&D Systems) were seeded in twenty-four well polystyrene culture plates. sicXer or sicXer-iaf granules (15 mg) in ThinCerts (1 µm, Greiner) were equilibrated overnight using 1 mL αMEM containing 10% heat inactivated FCS. The next day (day 0), by washing twice with PBS, non-attached cells were removed. Moreover, medium supplemented with 25 ng/mL M-CSF and 25 ng/mL receptor activator of nuclear factor kappa B ligand (RANKL; R&D Systems) was added to the adherent monocytes on their initial day (day 0) in presence of xerogel granules in ThinCert. Medium was changed every third day. For biochemical analysis, cells were washed twice with PBS followed by freezing at − 80 °C.

Biochemical analysis

Cell lysis was performed with 1% Triton X-100 (Sigma Aldrich) in PBS for 60 min with additional sonication for 10 min. For osteoblastogenesis, activity of lactate dehydrogenase (LDH) and alkaline phosphatase (ALP) was measured from lysates. For osteoclastogenesis, activity of tartrate-resistant acid phosphatase (TRAP) 5b was measured from lysates.

LDH activity was used for quantification of hMSC proliferation. Therefore, 50 µL cell lysate was incubated with 50 µL substrate solution of LDH Cytotoxicity Detection Kit (Takara) in the dark. After 15 min, the reaction was stopped using 50 µL 0.5 M HCl (Roth). Absorbance was measured at 492 nm (Infinite® M200Pro, Tecan). LDH activity was correlated with cell number using a calibration curve of cell lysates with defined number of cells.

ALP activity was used to quantify osteogenic differentiation, i.e., osteoblast activity. Cell lysates (25 µL) were mixed with 100 µL substrate solution consisting of 5.4 mM 4-nitrophenylphosphate disodium salt in substrate buffer containing 100 mM diethanolamin, 1 mM magnesium chloride, and 0.1% Triton X-100 (all Sigma Aldrich) adjusted to pH 9.8. After incubation for 30 min at 37 °C, reaction was stopped with 50 µL 0.5 M NaOH (Carl Roth). Absorbance was measured at 405 nm (Infinite® M200Pro, Tecan). For calibration, different concentrations of p-nitrophenol (Sigma Aldrich) in substrate buffer were used. ALP activity was normalized regarding cell number assessed via LDH activity.

TRAP 5b as osteoclast-specific enzyme was quantified using the substrate naphthol-ASBI phosphate. Cell lysates (10 µL) were added to 50 µL substrate buffer consisting of 2.5 mM naphthol ASBI phosphate (Sigma Aldrich) in 100 mM sodium acetate (Carl Roth), 50 mM sodium tartrate, 2% Nonidet™ NP 40, and 1% ethylene glycol monomethyl ether (all Fluka) adjusted to pH 6.1. The reaction was stopped with 125 µL 0.1 M NaOH (Roth) after incubation for 30 min at 37 °C. Fluorescence was measured at an excitation and emission wavelength of 335/405 nm (Infinite® M200Pro, Tecan). Different TRAP concentrations (BoneTRAP) were used for calibration.

Scanning electron microscopy

Samples were prepared on aluminum stubs and coated with carbon. An ESEM XL 30 scanning electron microscope (Philips) working at 3 kV and detecting secondary electrons was used for imaging.

Impact of sicXer and boXer on primary myeloma cells and myeloma cell linesPatients and samples

Consecutive patients (n = 8) with previously untreated, therapy-requiring (due to the presence of myeloma-defining CRAB-features [26]) or relapsed multiple myeloma were included in the study approved by the ethics committee of the Medical Faculty of the University of Heidelberg (#S-152/2010) after written informed consent.

Myeloma cells were purified from bone marrow aspirates by using anti-CD138 microbeads and an AutoMACS Pro Separator (Miltenyi Biotec); quality control was performed using flow cytometry [9, 27, 28]. The human myeloma cell lines AMO-1, KARPAS-620, KMS-11, KMS-12-BM, L363, OPM-2, RPMI-8226, and U266 were purchased from the German Collection of Microorganisms and Cell Cultures, American Type Cell Culture, or Japan Health Science Research Resources Bank; the HG-lines HG1 and HG9 were generated at the Myeloma Research Laboratory Heidelberg (Germany). Cell line identity was assessed by DNA-fingerprinting, mycoplasma-contamination excluded by PCR-based assays, and EBV-infection status by clinical routine PCR-based diagnostics.

Killing of myeloma cell lines

After 24 h pre-incubation of xerogels, myeloma cell lines were exposed to sicXer-iaf, boXer-20-iaf, boXer-100-iaf, and boXer-500-iaf, respectively, and cultured for 72 h. Cell viability was assessed by using a WST-1 based colorimetric assay (Roche) and referred to the medium control without xerogels. Each dose point was done in triplicates.

Survival of primary myeloma cells

Primary myeloma cells cultured together with their bone marrow microenvironment of eight myeloma patients were exposed to sicXer-iaf and boXer-500-iaf, respectively. After six days, cell viability was measured by CD138-FITC (IQ products, clone B-A38)/propidium iodide (PI; Pharmingen) staining and referred to the medium control as published [9, 27, 28]. In brief, remaining viable myeloma cells are identified as CD138+/PI−, remaining viable cells of the bone marrow microenvironment as CD138−/PI− cells. Each dose point was done in duplicates. Data analysis was performed using FACSDiva software (BD Biosciences).

Rat model (drill hole defect) bone healing

sicXer-iaf and boXer-100/500/2500-iaf (particle sizes: 250–710 µm) were implanted into a drill hole defect in the left femur of healthy female Sprague Dawley rats ([Crl:CD(SD)], age of 4 month; Charles River). Animal housing and experimental procedures were performed in full compliance with the institutional and German protection laws after approval by the local animal welfare committee (reference number: V54-19 c20/15-F 31/38). After general anesthesia, the distal femur was approached anterolaterally, and a drill hole defect was made in the metaphyseal area with a diameter of 2.5 mm and 4 mm depth. Post-operatively, animals were individually housed with free access to feed and water. Four weeks post-surgery, animals were euthanized under inhalation of CO2 after general anesthesia. Both operated and contralateral femora of each group were harvested and processed as mentioned below. For determination of the bortezomib release, samples from the drill hole itself, the surrounding muscle, and bone samples from the surrounding femoral condyle bone, the distal shaft region and the proximal shaft region were taken.

Histology and histomorphometryAssessment of new bone formation

Harvested rat femora were fixed in phosphate-buffered 4% paraformaldehyde and stored at 4 °C until processing. Samples were then embedded in Technovit® 9100 NEU according to the manufacturers protocol (HeraeusKulzer). After embedding, Technovit blocks were sectioned into 5 µm thick slices with the aid of Kawamoto´s film (Section-Lab) to prevent the loss of any biomaterial. While sectioning, the plane with both condyles being visible was maintained for all groups.

Histomorphometric analyses

Undecalcified bone sections stained with movat pentachrome and toluidine blue staining were used for qualitative and quantitative morphological analyses. Images were obtained with a 10 × objective using a light microscope (Axioplan 2 imaging with photomodule Axiophot 2, Carl Zeiss) and a Leica DC500 camera, acquired with Leica IM1000 software and processed using Adobe Photoshop version CS6.

For histomorphometric analysis of new bone formation (bone volume/trabecular volume; BV/TV) Adobe Photoshop CS6 was used. A single region of interest (ROI) comprising the initially created drill hole defect was chosen. This included both the new bone formation in the former created defect zone and the biomaterial-tissue interface. The ROI was kept constant for all groups. The measurements for ROIs, area of bone, implant, osteoid, and the void were considered to determine the extent of new bone formation.

Osteoblasts were traced on toluidine blue stained slides as blue cuboidal cells aligned in clusters at the bone surface. The osteoblast surface over bone surface (Ob.S/BS) was then determined by tracing directly on the osteoblast cells. The measurements were done blind folded with regards to the test groups.

Enzyme histochemical analysis

TRAP staining was used to investigate osteoclast activity. Samples were deplastified, followed by treatment with sodium acetate buffer and incubation in Napthol-AS-TR phosphate in N-N-dimethyl formamide (both from Sigma Aldrich) and sodium tartrate (Merck) with Fast Red TR salt (Sigma Aldrich) at 37 °C for 1 h.

Counterstaining was done with hematoxylin (Shandon). A count of TRAP-positive cells (osteoclasts) was done in the fixed ROI to determine the osteoclast count per trabecular area (Oc./Tb. Ar).

Immunohistochemistry and immune histomorphometry

The following antibodies were used: rabbit anti-BMP-2 polyclonal antibody (AP20597PU-N; Acris), rabbit anti-osteoprotegerin (OPG) polyclonal antibody (250,800; Abbiotec, USA), rabbit anti-CD254/RANKL polyclonal antibody (AP30826PU-N; Acris), rabbit anti-CD31 antibody (250,590; Abbiotec), and mouse anti-rat monocytes/macrophages monoclonal antibody ED1 (MAB1435; Chemicon), respectively.

Goat anti-rabbit (BA-1000, Vector) served as secondary antibody for BMP-2, OPG and RANKL followed by Vectastain ABC kit (Elite PK-6100, Standard, Vector Laboratories). Final visualization was done using Nova Red (SK4800, Vector Laboratories) and hematoxylin (Shandon) was used for counterstaining. ED1 antigen identification was done using DakoEnvision + System-HRP (DAB) for use with mouse primary antibodies (Dako, K4006).

Images were taken using Axioplan 2 Imaging system (Carl Zeiss) with a Leica DC500 camera, acquired with Leica IM1000 software and processed using Adobe Photoshop CS6.

Histomorphometrical analysis for OPG/RANKL ratio [%] was determined by manual count of stained cells for OPG and RANKL. ED1 counts were performed analogously.

5T33-mouse model—bone healing and control of myeloma cell proliferation5T33-mouse model

Animal housing and experimental procedures were realized according to the Institutional Animal Care and Use Committee of the Vrije Universiteit Brussels (license number. LA1230281) and all procedures were approved by the Ethics Committee for Animal Experiments (#15–281-6). sicXer (n = 8) and boXer-500/−2500 (n = 10 each), with particle sizes < 125 µm, were implanted into a drill hole defect of 1 mm diameter and 1 mm depth (see Fig. 4D) in the left femur of the 5T33-myeloma mouse model [29, 30] after anesthesia with Ketamine 100 mg/kg and Xylazine 10 mg/kg (both intraperitoneally). Meloxicam 1 mg/kg was administered subcutaneously for pain control. Mice were sacrificed three weeks after surgery and femora were fixed in 4% paraformaldehyde; one femur of each group was fixed in yellow fix, another one in dry ice for further analyses.

Histology and immunohistochemistry

The femoral condyles were fixed in 4% paraformaldehyd (Carl Roth) and embedded in Technovit 9100 (HeraeusKulzer) for histological and immunohistochemical examinations according to the manufacturer protocol.

All condyles were processed sagittally in sections of 4 µm thickness by using a Rotary Microtome (RM 2265, Leica Microsystems).

Masson Goldner staining was performed according to the manufacturer protocol (Carl Roth).

For CD138 immunohistochemistry, sections were deacrylated for 2 × 15 min in 2-methoxyethyl acetate (Merck), hydrated with decreasing alcohol concentration, heat treated at 90 °C with 0.1 M citrate buffer for 20 min, blocked 5 min in 3% H2O2, and incubated over night at 4 °C with the primary antibody against human, rat, mouse and hamster CD138 (Acris, PAB 9567) diluted 1:1000 in antibody diluent (Dako, DK, S3022). Then, sections were treated 2 × 20 min with SuperVison 2 Single Species HRP-Polymere Rabbit (DCS, PD 000POL-K), subsequently incubated with DAB (DCS, DC 137 C 100) for 10 min, and finally covered with DePeX (Serva).

Histomorphometry

All sections were photographed with a 5 × objective by using a Leica DFC 320 mounted on a Zeiss Axiophot. The sections stained by means of Masson Goldner were used to mark and measure the defect area and to calculate the amount of osteoid localized in close vicinity to the defect area. For this purpose, an area of interest was defined by using the Lasso of Adobe Photoshop CS6 Extended including the defect area itself and an area of 100 µm around the defect.

By using the Magic Wand Tool in Adobe Photoshop CS6 Extended, red stained areas of osteoid were marked and measured in mm2.

For detection of CD138 positive cells within the bone marrow, a ROI was defined by using the Lasso of Adobe Photoshop CS6 Extended. This area covered the bone marrow within the distal epiphysis around the defects and additionally, the subsequent area of the bone marrow extending 2 mm into the distal diaphysis. By using the Magic Wand Tool in Adobe Photoshop CS6 Extended, the respective area of membrane associated CD138 immunostaining was collected, and the proportional coverage of CD138 positive areas was calculated.

UPLC-MS/MS analysis

Bortezomib was determined with UPLC/MS/MS after liquid–liquid extraction as previously described [31]. For ex vivo analysis, explanted tissues (rat model) were homogenized in hydrochloric acid (500 µL, 0.1 M) using an ULTRA-TURRAX (IKA) and subsequently centrifuged (400 × g, 5 min, 4 °C). For liquid–liquid extraction, internal standard solution (25 µL) was added to each sample followed by vortexing (10 s) and sonication (3 min). After addition of 5 mL methyl tert-butyl ether, samples were automatically shaken (overhead, 15 min) and centrifuged (3000 × g, 10 min). The obtained supernatant was evaporated under a stream of nitrogen and reconstituted in 250 µL ACN/H2O (30/70, v/v + 0.01% formic acid).

Aliquots of 20 µL were injected into the UPLC-MS/MS system which consisted of an Acquity UPLC System (Waters Sample Manager and Binary Solvent Manager) and a triple stage quadrupole mass spectrometer (Waters Xevo TQ-S). A Waters Acquity BEH C18 column (1.7 μm, 2.1 × 50 mm) with an integrated filter disc was used for chromatographic separation. The acidified eluent (0.01% formic acid) consisted of acetonitrile and H2O at a flow rate of 0.8 mL/min. Using positive electrospray ionization, the mass spectrometer detected [M + H]+ ions at m/z 367.1 (bortezomib) and m/z 375.2 (D8-bortezomib) via the first quadrupole filter (Q1). After collision-induced fragmentation in argon gas (Q2), product ions were monitored via Q3 at m/z 225.9 (bortezomib) and m/z 233.8 (D8-bortezomib).

For calibration and quality control (QC), blank tissue was spiked with 25 µL of the respective calibration or QC solution resulting in ten calibration standards (0.5–2500 pg per sample) and three QC samples (1.65, 472, and 708 pg per sample) in duplicates per analytical run. Peak area ratios obtained from monitored ions were utilized for construction of calibration curves using weighted (1/x) linear least squares regression. Resulting calibration curves always revealed correlation coefficients greater than 0.998 and allowed a lower limit of quantification of 0.5 pg per sample. In accordance with FDA and EMA guidelines [32, 33], within-batch and batch-to-batch accuracies of the QC samples were within the accepted range (± 15%). Data collection, peak integration, and calculations were performed using Waters TargetLynx V4.1 software (Waters).

Micro-CT analysis of xerogel-bearing mice femora

Technovit embedded samples were imaged using the μCT system SkyScan 1173 (Bruker MicroCT). Raw data acquisition, image reconstruction and post processing steps were done following the guidelines for assessment of bone microstructure [34]. Scanning parameters were set as follows: tube current: 200 μA, tube voltage: 40 kVp, rotation step width 0.24°, frame averaging for noise reduction: n = 4. For beam filtration, a 0.5 mm aluminum filter was used. NRecon-Software (Bruker microCT) with a gaussian filter was utilized for Image reconstruction of cross sections with an isotropic voxel size of 5.7 μm.

To quantify new bone formation, a layer package of 0.5 mm thickness was defined in the mid-sagittal plane in perpendicular alignment to the drill hole. Within this volume, the volume (i.e., degradation) of the implant, the new bone formation within the drill hole, and the bone mass around the drill hole were determined based on threshold values. The method was adapted to previously reported techniques [35, 36].

For morphometric analysis, binarization of the gray-valued μCT data was carried out using a locally adaptive thresholding technique (CTAn-Software, Bruker microCT).

Following the suggestions made by Bouxsein et al., treatment induced changes in bone morphology were quantified by relative bone volume, trabecular thickness and trabecular number.

ToF–SIMS analysis

Time-of-flight secondary ion mass spectrometry (ToF–SIMS) enables the simultaneous analysis of organic and inorganic compounds of a sample surface with high mass and high lateral resolution (down to 100 nm). To obtain the chemical information, the sample surface is bombarded with a primary ion beam which impact leads to the emission of secondary ions. These secondary ions are analyzed by a time-of-flight analyzer and separated by their mass to charge ratio. By scanning the primary ion beam over the sample surface, the lateral distribution of the chemical compounds is obtained, and mass landscapes can be created. ToF–SIMS was used here for analysis of bone cross-sections regarding bone quality, degradation of implanted material, and bortezomib release. For detailed methodology and its application in bone research, see Kern et al. [37]. For SIMS-measurements, a ToF-SIMS 5–100 machine (ION-TOF Company) equipped with a 25 keV Bi-cluster ion gun for surface analysis, and an Ar gas cluster ion source as sputter gun was used. Bi3+ was used as primary ion species. The primary ion gun was operated in spectrometry mode with highest mass resolution and a lateral resolution of about 10 µm. Data evaluation was done with the Surface Lab 6.3 software of ION-TOF Company. During sample preparation, the surface was covered with a layer of resin. This was cleaned with Ar-clusters prior analysis. Therefore, a 10 keV Ar1500+ cluster beam scanned over the bone surface with a cleaning speed of 0.010–0.025 mm/s for 2–3 times depending on the thickness of the resin layer.

For imaging of rat femora, single images of the size 500 × 500 µm2 were stitched together to obtain areas of several square millimeters (so called “stage scans”). For bortezomib detection in boXer group, stage scans were taken with pixel density of 100/mm, cycle time of 50 µs, 500 shots/pixel and 5 patches with three scans resulting in 7500 shots/pixel in total. Empty defects and sicXer group were analysed applying the same conditions using 1 scan to reduce measurement time. Here, the obtained count rates are multiplied with 3 for comparison.

Statistical analysis

Computations were performed using R 3.1.1 (http://www.r-project.org/) and Bioconductor 2.14 [38]. Effects were considered statistically significant if the P value of corresponding statistical test was < 5%. If not otherwise stated, results were expressed as means ± standard deviation. For experiments related to cell culture and bortezomib release, statistical significance was evaluated by analysis of variance (two-way ANOVA, Bonferroni correction for multiple testing, GraphPad Prism). Statistical analysis of immunohistochemistry data was conducted using IBM SPSS Statistics (Version 28.0). Due to the non-normal distribution of the data, non-parametric methods were applied. Differences between treatment groups were assessed using the Kruskal–Wallis H test, followed by pairwise comparisons with the Mann–Whitney U test for significant results.