A Tunable Calcium Phosphate Coating to Drive in vivo Osseointegration of Composite Engineered Tissues

Many orthopedic interventions are predicated on the integration of implant materials with native bone. This is especially true for fracture healing treatments, spinal fusions, mechanical arthroplasties of the intervertebral disc (of which 38.31% of patients suffer complications [Guo et al., 2020]), and joint replacements of the knee and hip. Hydroxyapatite (HA) and other calcium phosphate coatings are some of the most utilized bioactive and osteoconductive surface modifications spanning a variety of materials [Surmenev et al., 2014; Habraken et al., 2016; Eliaz and Metoki, 2017]. Most commonly, HA surface coatings are applied to metallic and nondegradable polymeric implants, such as titanium and PEEK interbody fusion cages [Rao et al., 2014; Lee et al., 2017]. As the field of orthopedics investigates potential therapeutic transitions from well-tested replacement devices toward experimental tissue engineering and regenerative medicine therapies, the challenge of osseointegration will remain a critical priority. To address this need, HA coatings for biodegradable polymeric scaffolds continue to be developed and examined [Alizadeh-Osgouei et al., 2019]. However, after decades of study, exactly how and to what degree apatite molecules drive successful osteogenesis still remains unclear as many HA-coated implants fail to accelerate osseointegration in vivo [Bral and Mammaerts, 2016; Surmenev and Surmeneva, 2019]. Additionally, composite biomaterials have become a recent focus, especially in interventions that seek to regenerate multiple tissue complexes simultaneously, creating a need for robust, tunable, and easily scalable material modifications that can discretely induce osteogenesis in one or more regions of a composite.

Although HA is too brittle to be used on its own [Rao et al., 2014; Mondal and Pal, 2019], it is frequently deployed in conjunction with other materials [Mavrogenis et al., 2009; George et al., 2022]. HA can be incorporated directly through the addition of HA particles into polymer solutions [Guarino et al., 2008; Chuenjitkuntaworn et al., 2010; Bhattacharjee et al., 2016]; or HA coatings can be applied to polymer surfaces through electrodeposition [Bhattacharjee et al., 2017; Nie et al., 2019], sol-gel [Costa et al., 2012a], or simulated body fluid (SBF) techniques [Koju et al., 2017], among other methods [Alizadeh-Osgouei et al., 2019; Jaroszewicz et al., 2019].

Biomimetic coating via immersion in SBF has been a popular approach to HA surface functionalization compared to other techniques due to it being a low cost and easily scalable process with few technical barriers. Furthermore, successful functionalization of scaffold surfaces with HA utilizing this technique has been modeled extensively [Lu and Leng, 2005] and confirmed experimentally [Costa et al., 2012a, b; Miszuk et al., 2018; Raz et al., 2018; Chen et al., 2021]. Previous studies examining the biological effects of these HA coatings established their capacity to upregulate osteogenesis in vitro [Liu et al., 2015; Miszuk et al., 2018]. However, in vivo data have reported varying degrees of success. HA-coated polycaprolactone (PCL) scaffolds experienced significant volumes of de novo bone formation when implanted subcutaneously in mice and supplemented with BMP-2 [Miszuk et al., 2018]. A separate study reinforced this finding in a rat calvarial defect model, documenting that bone formation in HA-coated PLGA scaffolds only occurred with the addition of BMP-2 [Kim et al., 2008]. Conversely, accelerated bone deposition has been documented elsewhere in the absence of growth factors [Chuenjitkuntaworn et al., 2010]. 3D-printed PCL scaffolds functionalized with collagen and subsequently immersed in SBF for 7 days experienced large areas of new bone formation in comparison to uncoated PCL implants after implantation in rabbit radii for 12 weeks [Wang et al., 2015]. However, in a separate study of HA-coated collagen scaffolds, significant areas of new bone formation were not observed in smooth scaffolds [Hwangbo et al., 2021]. Significant levels of de novo bone only occurred after 6 weeks in murine vertebral bodies in scaffolds containing microchannels less than 20 μm in diameter. Similarly, only minor improvements in the rate of bone deposition were observed in HA-coated silk fibroin scaffolds when compared to uncoated scaffolds after implantation in rat cranial defects for 16 weeks [Liu et al., 2015].

It is unclear why such differences in osteogenic outcomes exist between studies, but, notably, many studies employ different methods of SBF immersion as this procedure is highly tunable. The size, shape, and surface coverage of HA crystals can be controlled by modulating the duration of scaffold hydrolysis, the concentration of hydrolytic agent used, the duration of SBF immersion, the concentration of SBF constituent molecules, and the rate of SBF replacement [Siriphannon et al., 2002; Yang et al., 2008; Costa et al., 2012b]. Differences in HA crystal shape and surface nanotopography as a result of these coating variables may account for the variable biologic response of cells to HA-coated surfaces [Bral, 2016; Guarino et al., 2016; Bhattacharjee et al., 2017; Pang et al., 2018]. Ultimately, how differential HA surface functionalization impacts osteogenesis has not been thoroughly scrutinized.

Because of its ability to induce osteogenesis, calcium phosphate surface coatings have also been combined with other biomaterials in composite scaffolds to mediate bone deposition in a spatially controlled manner. Recent studies of electrospun polymeric ligament and tendon replacements that received HA coating in a discrete zone [Olvera et al., 2020] or in a gradient [Chen et al., 2021] via spatially controlled SBF immersion were both able to drive regionalized upregulation of osteogenic gene markers, significantly improving bone-tendon integration strengths in vivo. However, similar combination approaches have yet to be explored for tissue engineering applications in the spine. Of particular interest to our group is the osseointegration of an endplate-modified disc-like angle-ply structure (eDAPS), a tissue-engineered total disc replacement designed for the treatment of end-stage disc degeneration as an alternative to spinal fusion surgery and mechanical arthroplasty [Martin et al., 2017; Gullbrand et al., 2018a]. Few studies have considered boney integration of a tissue-engineered disc, particularly after in vivo implantation. Work culturing nucleus pulposus (NP) cells on calcium phosphate substrates demonstrated the development of immature NP-like tissue [Séguin, 2004] and extremely weak interfacial shear strengths [Hamilton, 2006] after 6 weeks in vitro. A more recent study observed the development of tissue gradients in a tissue-engineered disc in vitro following migration of annulus fibrosus (AF) cells out of AF replacement materials and into adjacent cartilaginous endplate (EP) analogs [Chong et al., 2020]. Investigation of tissue-engineered discs built on those foundations demonstrated successful implantation into the bovine caudal disc space up to 1 month but did not characterize the effect of the calcium phosphate EP on osseointegration with native bone [Iu et al., 2017]. Previous implantation of eDAPS into the rat caudal disc space discovered that the integration strength of replacement discs with adjacent vertebral bone was only 50% that of native intervertebral discs at 20 weeks [Gullbrand et al., 2018a]. Expediting the osseointegration of the eDAPS would advance the translation of these composites toward clinical practice.

This study sought to establish a well-defined, replicable, and robust protocol for creating tunable calcium phosphate surface modifications of biodegradable polymeric scaffolds using a one-time, short-duration SBF immersion. Our initial goals were to determine the extent to which HA coating could drive osteogenesis of mesenchymal stem cells (MSCs) in the absence of osteogenic signaling factors and to understand whether differential degrees of HA surface functionalization would result in differential degrees of downstream osteogenic protein deposition. To this end, we hypothesized that longer durations of scaffold immersion in an SBF-like solution of calcium phosphate would drive more robust HA crystal formation, resulting in a stronger downstream osteogenic response from infiltrating MSCs. Following in vitro investigation and selection of an effective HA coating, we next wanted to determine if this coating could accelerate osseointegration in vivo, and we hypothesized that HA-coated PCL implants would have significantly more new bone deposition than noncoated implants after 15 weeks in the rat caudal disc space. Finally, in preparation for implementing this HA coating with our composite engineered discs, we evaluated whether this coating impacted adjacent tissue-engineered disc components.

The overall study design is shown in Figure 1. Salt-leached PCL scaffolds were either left unmodified or were HA-coated for 1 (1HA) or 7 (7HA) days in SBF. To investigate the HA coating’s ability to promote osseointegration as part of our established tissue-engineered total intervertebral disc replacements, three experiments were conducted. First, the scaffold’s osteogenic potential was evaluated in vitro. Scaffolds were seeded with bovine MSCs (as an in vitro model of progenitor cell infiltration from the bone marrow) and cultured with either basal media or osteogenic media supplemented with vitamin C, β-glycerophosphate, and dexamethasone for up to 10 weeks. Second, the osseointegration of HA-coated scaffolds with native vertebral bone was assessed in vivo. Hydrolyzed PCL or 7-day HA-coated PCL plugs were implanted into the caudal spines of athymic rats and matured for 15 weeks. Third, the compatibility of these HA-coated constructs as EP components in the eDAPS composite implant was validated in vitro. We combined the salt-leached PCL scaffolds with tissue-engineered AF and NP components to assess if the HA coating would elicit mineralization in neighboring non-HA-coated elements. eDAPS matured for 10 weeks in culture. All data were analyzed using parametric or nonparametric one-way ANOVAs based on normal/non-normal distribution where significance was defined as p < 0.05.

Fabricating Poly(ε-Caprolactone) Scaffolds. PCL scaffolds were fabricated according to a previously established salt-leaching protocol [Kim et al., 2020] in which 64 g of sodium chloride was passed through a 106-μm sieve and subsequently dissolved with 16 g of 90 kDa MW PCL in 80 mL of chloroform stirring at 80–160 rpm and 45°C overnight. The mixed solution was poured into a custom polydimethylsiloxane multi-scaffold mold form and maintained in a ventilated fume hood for 3–5 days to allow for solvent evaporation. Following solvent evaporation, PCL-salt slabs were demolded. Scaffolds were punched from the slabs using an appropriately sized biopsy punch and manually sanded to the desired height. Scaffolds were fabricated at two size scales for eDAPS fabrication (small: 5 mm diameter × 1.5 mm height; large: 16 mm diameter × 1.5 mm height) and at a third size scale for implantation in vivo in the rat caudal disc space (small plug: 5 mm diameter × 5 mm height). Finally, sodium chloride crystals were washed out of the scaffolds on an orbital shaker for 48 h using a ratio of 1 mL diH₂O to 3.75 mm³ scaffold volume.

Developing a Replicable HA-Coating Process. HA crystal formation was induced by hydrolyzing the washed salt-leached PCL scaffolds in a basic solution of 2 M sodium hydroxide for 24 h and subsequently submerging them in a calcium phosphate solution similar to 10X SBF for 1 or 7 days. The formula for SBF (Table 1) was modified from Tas and Bhaduri [2004]. All components were dissolved in diH₂O ahead of time, except for sodium bicarbonate which was added immediately before scaffold incubation to adjust pH of the solution to 6.5 and facilitate the reaction. Prior to hydrolysis, scaffolds were rehydrated through 30-min washes in an increasingly aqueous ethanol gradient (100%, 70%, 50%, and 30% ethanol) followed by two 30-min washes in 1X PBS.

Although the ratio of SBF components remained consistent, the process of HA coating was refined and optimized through investigation of the following variables: (1) volume of diH₂O and method of agitation used during the salt-washing step of scaffold fabrication, (2) duration of scaffold hydrolysis, (3) duration of SBF immersion, and (4) concentration of SBF solution. For all variables, the residual salt content and the formation of HA crystals was characterized using scanning electron microscopy imaging and µCT (Scanco µCT50) with a 10-μm voxel size. All hydrolysis and SBF incubation steps occurred on a rocking platform set to 70 rpm for gentle agitation or 120 rpm for aggressive agitation, respectively. Variables were investigated as follows:

• Constructs were washed at a ratio of 3.75, 7.50, 11.25, 18.75, or 26.25 mm³ scaffold to 1 mL diH₂O on either a platform rocker (set to its max 120 rpm) or an orbital shaker (set to speed 8 of 10) to determine the optimal washing ratio for removing salt from scaffolds (3.75 mm³ scaffold to 1 mL water) (online suppl. Fig. S1; for all online suppl. material, see www.karger.com/doi/10.1159/000528965). Washing control improved consistency and replicability of HA coating across scaffold batches (online suppl. Fig. S2).

• Small constructs were hydrolyzed in 2 M NaOH for 16, 20, 24, or 28 h at a ratio of at least 1 mL 2 M NaOH to 3.75 mm³ scaffold volume (online suppl. Fig. S3a, b).

• In order to create materials with varying degrees of HA surface functionalization, small constructs were hydrolyzed for 24 h in 2 M NaOH and immersed in 10X SBF for 0.25, 1, 3, or 7 days at a ratio of at least 1 mL SBF to 3.75 mm³ scaffold volume (online suppl. Fig. S3c, d).

• In order to replicate the coating efficiency when scaling these parameters from small scaffolds to large scaffolds, SBF concentration needed to be adjusted. Large constructs were hydrolyzed for 24 h in 2 M NaOH and immersed in 10X, 20X, or 40X SBF for 1 or 7 days to determine the most effective concentration (online suppl. Fig. S4). 40X SBF coated large scaffold surfaces more homogeneously than did 10X or 20X SBF.

The optimal HA-coating variables selected for experimentation for each PCL scaffold size are outlined in Table 2.

Acellular Scaffold Mechanical Testing. Constructs were compressively tested using an Instron 5948 to characterize their bulk mechanical properties. Scaffolds underwent cyclic compressions of 0.5 N to 3 N for 3 cycles at 1 Hz (n = 5). A bilinear fit of the stress-strain curve was utilized to determine toe and linear modulus and transitional and maximum strain [Martin et al., 2017]. Constructs were also subjected to indentation testing to a depth of 15 μm using the Optics11 Piuma nanoindenter and a 163.8 N/m probe with a radius of 48.5 μm. Five scaffolds from each group were indented at 10 unique points across each surface to obtain 50 unique Young’s Moduli for each group.

Cell Seeding of PCL Scaffolds. Prior to cell seeding, small scaffolds (both uncoated and HA-coated) were hydrated and sterilized through an ethanol gradient and coated overnight in 20 μg/mL fibronectin through immersion. Scaffolds were seeded with 41,500 juvenile bovine MSCs on both surfaces (2,818 cells/mm³), and after each side was seeded, scaffolds were maintained for 1 h at standard culture conditions (21% O₂, 5% CO₂, 37°C). PCL-only and HA-coated PCL scaffolds were then divided into groups given either basal media (Dulbecco’s Modified Eagle’s medium containing 10% fetal bovine serum and 1% penicillin/streptomycin/fungizone) or osteogenic media (basal media supplemented with 50 μg/mL ascorbic acid, 10 mM β-glycerophosphate, and 10 nM dexamethasone). Samples were removed at 3 and 7 days for cell morphology characterization and at 5 and 10 weeks for all other analyses.

Quantifying Cellular Metabolism Using an Alamar Blue Assay. Every other week (with time 0 being the first media change), cellular metabolism was quantified using an Alamar blue assay (n = 6). Stock Alamar blue was diluted 1:10 in basal and osteogenic media to create two working solutions. Scaffolds were incubated in the appropriate working solution (1 mL per scaffold) for 3 h on an orbital shaker with gentle agitation, and 1 mL of each working solution was incubated in its own clean, empty control well. After 3 h, 100 μL of surrounding medium was taken in triplicate from each sample, added to a 96 clear-bottomed black well plate and fluorescently read on a BioTek Synergy H4 hybrid plate reader at excitation/emission wavelengths of 560/590 nm.

Characterizing Cell Morphology. Scaffolds (n = 3) were removed from culture at 3 and 7 days and immediately washed with 1X PBS. Cells were fixed via immersion of constructs in 4% paraformaldehyde for 20 min at 37°C and subsequently washed in 1X PBS three times. Cells were then permeabilized for 5 min at 4°C using a solution of 0.5% v/v Triton X-100, 10.75% w/v sucrose, and 0.06% w/v magnesium chloride in 1X PBS followed by three more washes in 1X PBS. Actin was then labeled by incubating samples for 1 h at 37°C in a solution of 0.001% v/v Alexa Fluor 546 Phalloidin (ThermoFisher, A22283) and 1% w/v bovine serum albumin (Sigma, A7906) in 1X PBS, followed by three washes in 1X PBS. Finally, scaffolds were incubated in 5 mM DRAQ5 (ThermoFisher, 62251) for 15 min at ambient temperature to stain cell nuclei, washed three times in 1X PBS, and imaged using a multichannel confocal microscope (Nikon A1R).

qRT-PCR Analyses for Osteogenic Markers. A quantitative real-time polymerase chain reaction was utilized to quantify the presence of specific gene markers indicative of osteogenesis. The genes of interest and their primer sequences are outlined in Table 3. At 5 weeks of culture, samples (n = 3) were removed from media and frozen at −80°C. Total RNA extraction was completed using a TRIzol-based assay. Each scaffold was dissolved in 300 μL TRIzol and centrifuged at 13,000 rpm for 30 s. TRIzol solutions were then transferred to new RNA-free microcentrifuge tubes and mixed with a half volume of isopropanol through manual pipette agitation before proceeding with the Zymo Research Microprep Kit instructions (Zymo Research, Direct-zol RNA Microprep Kit, R2062). Isolated RNA concentrations were then quantified (NanoDrop 3300) and reverse-transcribed into cDNA using Superscript IV VILO Mastermix Reverse Transcriptase (Life Technologies, 11756500).

Quantitative real-time polymerase chain reaction was performed using a final concentration of 0.4 ng/μL cDNA, 1 μM forward and reverse primers, and 1X PowerTrack SYBR Green PCR Master Mix (Life Technologies, A46109). Melt curves were assessed on four samples to verify the suitability of each primer set. Relative gene expression was determined using the ΔΔCT method with glyceraldehyde 3-phosphate dehydrogenase serving as the housekeeping gene and PCL scaffolds incubated in basal media as the reference control.

Histological Analyses. At 5 and 10 weeks, each scaffold was embedded in optimal cutting temperature liquid (OCT) and cryosectioned at 10 μm using the Kawamoto tape method [Kawamoto and Shimizu, 1986]. Slices were then fixed in 10% buffered formalin, glued to glass slides using Krazy Glue Super Glue, and allowed to dry completely. Calcium deposition and cell density in each of the sections was visualized through Von Kossa staining (n = 3–6). For immunohistochemical detection of osteocalcin (OCN) (Santa Cruz Biotechnology, ABOC-5021) and osteopontin (OPN) (Novus Biologicals, NBP1-59190), sections (n = 3–6) were incubated in proteinase K for 5 min followed by 3% hydrogen peroxide for 10 min, both at room temperature. Sections were then incubated in a blocking solution of 2.5% normal horse serum (Vector Laboratories, S-2012-50) for 30 min at room temperature. OCN (2 μg/mL) and OPN (10 μg/mL) antibodies were applied in a humidity chamber at 4°C overnight followed by a 1 h room temperature incubation with R.T.U. biotinylated secondary antibody (Vector Laboratories, BP-1400-40). Antibody binding was visualized using a DAB chromagen reagent (Vector Laboratories, SK-4100) according to the manufacturer’s protocol. Brightfield images were acquired using a Leica Biosystems Aperio CS2 slide scanner.

Scaffold Implantation. Following IACUC approval, 6 male athymic rats (Foxn1^mu retired breeders, Envigo) underwent total discectomy of a caudal intervertebral disc space and the placement of a ring-type external fixator according to our previously described surgical methods [Martin et al., 2017]. Following discectomy, 3 rats received a hydrolyzed PCL small plug, and 3 rats received a 7-day HA-coated PCL small plug. Rats were euthanized 15 weeks post-implantation.

Histological Analyses. Implants and their adjacent vertebral bodies were fixed in 10% buffered formalin and soaked in a solution of 2% polyvinylpyrrolidone (Sigma, P5288) and 20% sucrose (Sigma, S7903) in 1X PBS for 7 days. Implants were then embedded in OCT and cryosectioned in the sagittal plane at 10 μm using the Kawamoto tape method [Kawamoto and Shimizu, 1986]. Cryosections underwent immunohistochemical detection of OCN and OPN. The remainder of each implant was subsequently decalcified in 40 mL Formical-2000 (Fisher, NC9974148) for 4 days with gentle mechanical agitation and processed for paraffin histology. Samples were sectioned in the sagittal plane at 10 μm and stained with hematoxylin/eosin, RGB trichrome [Gaytan et al., 2020], and Mallory-Heidenhain trichrome.

Fabricating a Tissue-Engineered Total Disc Replacement. Tissue-engineered total disc replacements – eDAPSs (endplate-modified disc-like angle-ply structures) – were fabricated at a size physiologically relevant for disc replacement in goat cervical spines [Gullbrand et al., 2018a] by combining an AF replacement (layers of electrospun 90 kDa PCL seeded with goat bone marrow-derived MSCs), a NP replacement (core of 4% agarose hydrogel seeded with goat bone marrow-derived MSCs), and an acellular endplate analog (large salt-leached PCL scaffolds) according to our established methods [Gullbrand et al., 2018a, b]. For this study, the PCL EP analogs were either uncoated (n = 4 eDAPS) or HA-coated (n = 4 eDAPS) for 7 days. eDAPS were cultured in chondrogenic media containing 10 ng/mL TGF-β3 (Dulbecco’s Modified Eagle’s Medium containing 1% penicillin/streptomycin/fungizone [Antibiotic-Antimycotic; Gibco], 0.1 mM dexamethasone, 50 μg/mL ascorbate-2-phosphate, 40 μg/ml L-proline, 100 μg/mL sodium pyruvate, 1% ITS + Premix Universal Culture Supplement [Corning, 254352], 1.25 mg/mL bovine serum albumin, and 5.3 μg/mL linoleic acid albumin) and left to grow at standard culture conditions with constant mechanical agitation until their removal from culture at 10 weeks.

Histological Analyses. eDAPS were cut in half immediately following their removal from culture. The first half was embedded in OCT and cryosectioned in the sagittal plane at 15 μm using the Kawamoto tape method [Kawamoto and Shimizu, 1986]. Cryosections underwent immunohistochemical detection of OCN and OPN as described above. The second half of each eDAPS was fixed in 10% buffered formalin, decalcified, and processed for paraffin histology. Samples were sectioned in the sagittal plane at 15 μm and stained with hematoxylin/eosin, Alcian blue/Picrosirius red, and Mallory-Heidenhain trichrome.

Following PCL scaffold fabrication, constructs were HA-coated for 1 (1HA) and 7 (7HA) days (Fig. 2a). Small scaffolds were coated in 10X SBF, and large scaffolds were coated in 40X SBF. With respect to bulk mechanical properties, scaffolds experienced a reduction in linear modulus after hydrolysis followed by a significant increase in linear modulus after both 1 day and 7 days of HA coating with no difference with increasing HA-coating duration (Fig. 2b). At the microscale, 7HA scaffolds were significantly stiffer locally than 1HA scaffolds, and both 1-day and 7-day HA-coated scaffolds were significantly stiffer locally than both PCL and hydrolyzed PCL scaffolds (Fig. 2c). Longer periods of SBF immersion correlated with greater density of HA crystal nucleation throughout the interior of both small and large scaffolds (Fig. 2e) while maintaining surface crystal size (Fig. 2d). HA surface coverage also increased on large constructs from 1 to 7 days of immersion, becoming more homogeneous with time (online suppl. Fig. S4).

In vitro, bovine MSCs attached more readily to HA-coated scaffolds than uncoated scaffolds. By 3 days, cells cultured in osteogenic media spread more across the surface of HA-coated scaffolds than cells cultured in basal media (Fig. 3a). Cell spreading was lower in all basal media scaffolds as well as in PCL-only scaffolds in osteogenic media. At 7 days, an increase in actin staining that followed the morphology of the scaffolds’ crystalline surface features was apparent in both HA osteogenic groups (Fig. 3b). Actin staining was minimal in both PCL groups. Overall, the greatest amount of cell surface attachment and actin staining was observed in 1HA osteo scaffolds. Regardless of scaffold group, MSCs cultured in basal media proliferated between 0 and 6 weeks of culture, while cells cultured in osteogenic media did not proliferate (online suppl. Fig. S5).

By 5 weeks, OCN (Fig. 4a) expression increased in PCL osteo and 7HA basal scaffolds, corresponding with a robust deposition of OCN protein at 10 weeks (Fig. 4g). In contrast, no upregulation of OCN expression in PCL basal and 1HA basal groups occurred at 5 weeks, and a corresponding lack of OCN staining was observed at 10 weeks (Fig. 4g). Both 1HA osteo and 7HA osteo constructs did not show OCN upregulation at 5 weeks. However, they did show increased staining for OCN at the same time point (Fig. 4d), suggesting that OCN had been upregulated earlier than in other groups, leading to deposition of protein and calcium (Fig. 4f) at 5 weeks. A similar trend was observed in the expression of runt-related transcription factor 2 (RUNX2) (Fig. 4b) and COLX (Fig. 4c) at 5 weeks, in which PCL osteo and 7HA basal groups were upregulated, respectively. OPN staining at 10 weeks followed a similar trend as OCN staining, with 7HA basal scaffolds showing more robust osteogenic protein deposition than 7HA osteo scaffolds (Fig. 4h; online suppl. S6). Von Kossa staining at 10 weeks demonstrated robust calcium deposition homogeneously throughout the HA osteo constructs of both coating durations and concentrated around the very exterior of PCL osteo and 7HA basal constructs (Fig. 4i). Von Kossa staining was minimal in PCL basal scaffolds and heterogeneous in 1HA basal constructs (Fig. 4i; online suppl. S7). Altogether, these data indicate that the increase in osteogenic gene expression and resulting osteogenic protein deposition is accelerated by osteogenic media as well as in 7HA basal scaffolds independent of osteogenic signaling molecules (Table 4). We therefore selected the 7HA coating for further evaluation in vivo in a rat caudal disc replacement model.

Small plugs HA-coated for 7 days in 40X SBF achieved similar levels of HA crystal size and infiltration as small scaffolds HA-coated for 7 days in 10X SBF (online suppl. Fig. S8). These 7HA small plugs were successfully implanted into rat tail caudal disc spaces and could be visualized intra-operatively via fluoroscopy (Fig. 5a). After 15 weeks in vivo, native cells had fully infiltrated all scaffolds (Fig. 5b; online suppl. S9). This cellularization led to increased deposition of OCN and OPN (Fig. 5c, d; online suppl. S10) throughout HA-coated scaffolds as well as increased staining for collagen (Fig. 5e; online suppl. S11) and unmineralized bone (Fig. 5f; online suppl. S12), compared to uncoated scaffolds.

When utilized as the EP component of the eDAPS composite, these salt-leached PCL scaffolds were combined with cell-seeded AF and NP analogs during eDAPS in vitro chondrogenic maturation. HA coating of acellular EPs did not induce mineralization in, or affect the chondrogenic maturation of, the adjacent cell-seeded NP and AF analogs during in vitro culture of the tissue-engineered disc composite (Fig. 6; online suppl. S13, S14).

Calcium phosphate coatings are often paired with the delivery of osteogenic growth factors that accelerate the rate of de novo bone tissue formation [Bose and Tarafder, 2012; Dadestan et al., 2015; Gronowicz et al., 2017]. Unfortunately, growth factors are expensive and require controlled delivery systems to prolong their short half-lives – factors that add additional complications to biomaterial systems. To eliminate these complications, this study sought to enhance osteogenesis in MSCs without the use of growth factors by employing a tunable biomimetic HA-coating process for functionalization of biodegradable polymer surfaces.

Our in vitro studies suggested that osteogenesis was accelerated in HA-coated scaffolds cultured in osteogenic media compared to all other groups. MSCs on osteo HA scaffolds began depositing OCN and OPN as early as 5 weeks, particularly in the 1HA osteo group, whereas MSCs on 7HA basal and PCL osteo scaffolds did not deposit these proteins until 10 weeks. OCN and OPN are key regulators of continued osteoblast differentiation, without which mineralization is severely attenuated [Carvalho et al., 2020]. These interdependent genes are most prominently expressed and deposited at the onset of mineralization [Stein et al., 1990], and the presence of bone apatite-like mineral on the surface of HA scaffolds mimics this environment. Downstream calcium deposition at 10 weeks in response to prior OCN and OPN staining was subsequently observed robustly in 1HA and 7HA osteo scaffolds and peripherally in 7HA basal and PCL osteo scaffolds. Prior to protein deposition, OCN upregulation was observed in 7HA basal and PCL osteo groups at 5 weeks. Upregulation of OCN in MSCs was observed simultaneously with the expression of collagen X in 7HA basal scaffolds and with RUNX2 in PCL osteo scaffolds, as has been individually documented in other 10X SBF-mediated HA-coated constructs [Olvera et al., 2020; Chen et al., 2021]. Upregulation of osteogenic genes in 1HA and 7HA osteo groups likely occurred at an earlier time point as osteogenic protein deposition was observed at 5 weeks. Overall, these data suggest that 7HA scaffolds can independently induce osteogenesis to the same extent as uncoated scaffolds given osteogenic signaling molecules.

The capacity of the 7HA coating to drive osteogenesis independent of growth factors may be due to altered surface morphology and stiffness. Cells on HA-coated scaffolds demonstrated increased cell actin staining at 3 and 7 days, particularly in osteogenic media. Similar increases in cell spreading and actin staining have been observed on other HA-coated constructs [Miszuk et al., 2018] and have also been found to be more strongly activated in the presence of osteogenic media [Vaquette et al., 2013]. Mechanistic studies show that MSCs spread across a larger area enter more readily into osteogenesis [Wang and Chen, 2013; Jiao et al., 2020] and that substrate stiffness plays a primary role in mediating this cell morphology [Olivares-Navarrete et al., 2017; Xie et al., 2018; Benayahu et al., 2019]. Stiffening of polymer scaffolds following SBF immersion or HA particle dispersion across both macroscopic [Chuenjitkuntaworn et al., 2010; Costa et al., 2012b; Liu et al., 2015] and microscopic [Yang et al., 2016; Miszuk et al., 2018] length scales has been previously documented, and progressive stiffening concomitant with progressive increases in HA crystal surface coverage was observed over the course of this study as well. Osteogenic differentiation of MSCs was highest on substrates with a compressive stiffness of 4.7 ± 1.0 MPa [Olivares-Navarrete et al., 2017], a value similar to those measured locally on 7HA scaffolds. This increase in surface stiffness and subsequent MSC cell area activates the Wnt/β-catenin signaling pathway that ultimately initiates RUNX2 expression [Xie et al., 2018; Benayahu et al., 2019]. Higher levels of cell spreading have also been observed on scaffolds with submicron-sized surface features, visually similar to those observed on both 1HA and 7HA scaffolds, than on scaffolds with micron-sized surface features [Zhang et al., 2017; Pang et al., 2018; Wang et al., 2019], resulting in higher rates of bone formation in vivo [Zhang et al., 2017; Wang et al., 2019]. These two factors – surface morphology and surface stiffness – working in concert appear to be responsible for differentially driving MSC osteogenesis on 7HA scaffolds.

Critically, 1HA basal scaffolds showed promising osteogenic cell morphologies at 7 days, exhibiting high levels of actin staining and cell spreading, but never reached a significant level of protein deposition indicative of osteogenesis. This lack of osteogenesis in the absence of growth factors with short-term HA coating has also been observed in HA-coated PCL films, where MSCs cultured on 7HA films expressed significantly higher levels of BMP-2 and -7 than MSCs cultured on 2HA films, in a coculture system of bone marrow MSCs, HUVECs, and RAW264.7 cells [Jiang et al., 2022]. Since 1HA and 7HA constructs have comparable nanotopographies, this difference is most likely driven by the change in local scaffold stiffness. In addition to scaffold stiffness, coating biochemistry may have also played a role in activating MSC osteogenesis on 7HA scaffolds but not on 1HA scaffolds. Although the deposition of HA has been verified after 6 h of immersion in the 10X SBF solution used here [Tas and Bhadrui, 2004], similar solutions of 10X SBF deposited a coating more similar to dicalcium phosphate after 1–2 days and a coating more similar to natural bone apatite after 7 days [Yang et al., 2008; Wang et al., 2015].

The fabrication of 7HA scaffolds may still be variable, even after instituting strict washing control measures on salt-leached PCL scaffolds. Our results suggest that the increased micromechanical surface stiffness of 7HA scaffolds is responsible for driving MSC osteogenesis. In order to ensure these local stiffnesses are met, researchers should screen samples from each batch using indentation testing. Although μCT can be an effective way of measuring the density of HA crystal nucleation and could potentially be used to institute quality control cutoffs, indentation testing ensures that scaffolds have the appropriate surface stiffnesses required to activate the MSC osteogenesis observed here while also being more time efficient. Whereas μCT measurements seem to be more variable between batches, indentation measurements on samples across batches are more consistent. These authors recommend excluding samples with an average surface compressive stiffness less than 1.5 MPa and greater than 6 MPa to replicate the most effective scaffolds fabricated over the course of this study. This range of stiffness activates MSC osteoblastogenesis most effectively, whereas anything softer (0.8 MPa) activates MSC chondrogenesis more prominently and anything substantially larger (200–300 MPa) fails to activate either osteogenesis or chondrogenesis [Olivares-Navarrete et al., 2017].

The presence of 7-day HA-coated EPs did not affect the maturation of neighboring AF and NP analogs in a total tissue-engineered composite disc replacement. At the end of the 10-week culture period, no differences in eDAPS tissue development were observed histologically. Although MSCs did not deposit OCN, MSCs deposited varying levels of OPN in the NP region for both PCL-only and HA-coated EPs at 10 weeks, suggesting that the HA coating is not the source of OPN production. Mineralization and cell hypertrophy have been well documented in MSC-seeded hydrogels in long-term culture with TGF-β3 [Bian et al., 2007; Farrell et al., 2014]. MSC hypertrophy is also contingent on continued administration of TGF-β3 [Bian et al., 2011], and mineralization of the NP region of the eDAPS has not been observed following in vivo implantation [Gullbrand et al., 2018a].

After 15 weeks in the rat caudal disc space, HA-coated implants showed early signs of bone tissue deposition. However, this did not lead to full osseointegration in any of the samples. Although the implants were fully cellularized after 15 weeks, the high polymer density and small 106 μm pore size of the PCL scaffolds likely prevented short-term remodeling, as has been reported in a similar in vivo study [Shim et al., 2017]. The presence of concentrated areas of unmineralized bone tissue within HA-coated implants suggests that bone may continue to form over longer periods of time. However, pursuing angiogenic-coupled osteogenesis has been suggested as a more successful strategy for accelerating bone deposition. This could be achieved by introducing a combination of pore sizes (around 100 μm and 300 μm) [Karageorgiou and Kaplan, 2005; Liu et al., 2020] or microchannels [Chen et al., 2018; Hwangbo et al., 2021] that allow for the successful patterning of new vasculature required for the subsequent formation of mature bone [Albrektsson and Johansson, 2001; Mikos et al., 2006]. Optimization of scaffold morphologies that promote angiogenesis will likely increase the success of discrete bone formation in conjunction with tunable biomimetic HA coating for applications that require multi-tissue regeneration.

Ensuring sufficient surface contact between the implant and the native bone remains one of the key limitations of this in vivo model. In comparison to other disc spaces, the rat caudal spine is substantially more mobile than the lumbar, thoracic, and cervical spines due to the lack of posterior elements. In addition, the caudal spine lacks substantial soft tissue coverage and is well within reach of the rat’s mouth, forepaws, and obstructions that may interfere with implant stability. In the future, a new model assessing bone formation in vivo, which involves implanting a scaffold surrounded by a ring of bovine bone subcutaneously in mice [Sastre et al., 2021], may be a more stable, less invasive, and higher throughput way of assessing these biomaterials than traditional craniofacial, femoral fracture, or caudal disc space models.

Differential degrees of HA surface functionalization have been proposed as primary drivers of varied osteogenic response in infiltrating cells. In this study, we successfully fabricated PCL salt-leached scaffolds with two levels of biomimetic calcium phosphate coating to test this hypothesis. Longer duration SBF coating led to increased HA crystal nucleation within scaffold interiors as well as more robust HA crystal formation on scaffold surfaces. Ultimately, the increased surface stiffness of 7HA scaffolds in comparison to 1HA scaffolds led to more robust osteogenesis of MSCs without the assistance of osteogenic signaling molecules. Our results also suggest that the use of SBF-based HA coatings can promote higher levels of osteogenesis in vivo. Finally, HA-coated components of a larger tissue-engineered disc replacement showed no signs of inducing mineralization in or promoting cell migration out of neighboring materials. In the future, the 7HA coating will be implemented in large animal studies assessing eDAPS functional integration at clinically relevant length scales. Overall, our results suggest tunable biomimetic HA coatings are a promising approach for achieving discrete mineralization within composite biomaterials.

The protocol for the in vivo rat study was reviewed and approved by the Corporal Michael J. Crescenz VA Medical Center IACUC, approval number 01252.

The authors have no conflicts of interest to declare.

This work was supported by the Department of Veterans Affairs Rehabilitation Research and Development Service, award numbers I21 RX003289, I01 RX002274, and IK6 RX003416. The contents do not represent the views of the US Department of Veterans Affairs or the US Government. This work was carried out in part at the Singh Center for Nanotechnology, part of the National Nanotechnology Coordinated Infrastructure Program, which is supported by the National Science Foundation grant NNCI-2025608. This work was also carried out in part with support from the Penn Center for Musculoskeletal Disorders (NIH/NIAMS P30AR069619).

Study conception and design: Sarah E. Gullbrand, Matthew Fainor, Robert L. Mauck, and Harvey E. Smith. Acquisition of data: Matthew Fainor, Sonal Mahindroo, Kerri R. Betz, and Janai Augustin. Obtaining of funding: Sarah E. Gullbrand, Robert L. Mauck, and Harvey E. Smith. Drafting of the article: Matthew Fainor and Sarah E. Gullbrand. Revising the article for critically important intellectual content: all authors. Final approval of the version to be submitted: all authors.

All data generated or analyzed during this study are included in this article and its supplementary material files. Further inquiries can be directed to the corresponding author.