Osteogenic Potential of 3D Bioprinted Collagen Scaffolds Enriched with Bone Marrow Stromal Cells, BMP-2, and Hydroxyapatite in a Rabbit Calvarial Defect Model

Abstract

This study investigates the effect of three-dimensional (3D) bioprinted collagen (Col) scaffolds (2% w/v collagen) loaded with autologous bone marrow stromal cells (BMSCs) and enriched with bone morphogenetic protein-2 (BMP-2) and hydroxyapatite-based particles (HAPPs) on bone regeneration in calvarial defects in rabbits. Three implant formulations, Col-(BMP-2) (at a concentration of 80 ng/mL), Col-HAPP (1% w/v) and a mixture of the two—Col-(BMP-2)-HAPP (40 ng/mL final concentration and 0.5% HAPP), were compared with a control group C-Per containing only periosteum to assess the influence of material structure, biochemical signals and cell component on osteogenesis. Histological analysis and quantitative computed tomography (CT) imaging parameters (HU values and residual defect diameter) showed significant differences between the groups, highlighting the role of combined strategies for optimal bone repair. The control group demonstrated the weakest regeneration, expressed by minimal lamellar bone and the largest residual defect. Col-(BMP-2) stimulated moderate osteoinduction with active osteoblasts but without a fully organised lamellar structure. Col-HAΡΡ provided more advanced regeneration, with histologically observed thick osteoid lamellae, early calcification, and structured lamellar architecture, emphasising the osteoconductive role of HAΡΡs. The strongest regeneration was reported with Col-(BMP-2)-HAΡΡ, where the synergy between BMP-2, HAΡΡs and BMSCs resulted in formed osteons, well-developed cancellous bone and minimal residual defects. The established negative correlation between bone density and residual calvarial defects emphasises the relationship between mineralisation and the degree of defect filling. The new data presented demonstrate that the combination of the abovementioned structural, biochemical and cellular factors in 3D bioprinted scaffolds offers a promising strategy for osteoregeneration of complex bone defects.

1. Introduction

Bone is one of the most commonly transplanted tissues worldwide, with over two million bone graft procedures performed annually [1,2,3,4]. Despite clinical experience and technological advances, osteoregeneration in critical bone defects remains a challenge [5,6,7]. Autografts are considered the gold standard due to their osteoinductivity and histocompatibility [8], but the limitations imposed by autografting remain a serious obstacle to the effective treatment of critical defects [9,10].

Three-dimensional (3D) bioprinting technology offers the potential to overcome the limitations of traditional methods through customised designs that replicate the architecture of natural bone and combine osteoconductive and osteoinductive components [1]. Despite advances in the use of 3D bioprinting to aid bone regeneration in various in vivo models [11,12,13], these approaches still require further attention to key aspects such as composite design, adapted biomechanics, and biological functionality [1,14]. In biofabrication, 3D bioprinted scaffolds can effectively modulate the cellular response and promote bone regeneration only if they possess structural and biochemical characteristics close to the natural matrix [6,7], which mainly contains collagen and hydroxyapatite. These components provide a structural framework, mineral stability, and signals for osteogenesis [7,15,16]. Collagen (Col) is widely used due to its low immunogenicity and favourable mechanical properties that promote cell adhesion, proliferation and differentiation [17,18,19,20,21], but pure collagen implants suffer from limited osteoinductivity and low strength [22]. To overcome these shortcomings, various collagen composites have been developed by incorporating natural or synthetic polymers as well as bioactive inorganic components [23]. The integration of hydroxyapatite-based particles (HAPPs) improves the mechanical stability and osteoconductivity of the collagen matrix [6,24,25,26,27,28].

Advances in 3D bioprinting technologies for functionalised scaffolds and approaches for incorporating growth factors allow for the optimisation of vascularisation; the attraction of mesenchymal stromal cells; and the regulation of cell behaviour, including adhesion, proliferation, migration, and differentiation through the synergistic action of cytokines [1,15,16]. Recent studies have shown that the combined delivery of osteogenic and angiogenic factors via biodegradable scaffolds significantly improves osteogenesis and vascularisation in models of critical bone defects [29,30,31]. The inclusion of bone morphogenetic protein-2 (BMP-2) provides a biochemical stimulus complementary to the mechanical framework of collagen [28,32], with BMP-2-containing scaffolds demonstrating significantly improved bone regeneration both on the surface and in the depth of the construct. BMP-2 also induces the differentiation of bone marrow stromal cells (BMSCs) into osteoblasts and stimulates their proliferation [5].

BMSCs play a key role in maintaining bone marrow homeostasis and contribute to bone regeneration by stimulating osteogenic differentiation and modulating the inflammatory environment through the secretion of cytokines, chemokines, and growth factors [33]. The incorporation of BMSCs into 3D bioprinted scaffolds allows for the creation of personalised constructs that can significantly improve bone repair and cartilage regeneration. The differentiation potential of stromal cells depends on their tissue origin. It has been reported that adipogenic stromal cells from Wharton’s jelly exhibit higher proliferative and immunoprivileged activity compared to BMSCs, while dedifferentiated BMSCs retain adipogenic potential comparable to stromal cells derived from Wharton’s jelly [34]. Although the latter exhibit high proliferative activity and immunoprivileged properties, BMSCs provide osteogenic potential directly relevant to bone defect regeneration.

The combination of BMSCs with biologically based scaffolds represents a promising strategy for stimulating osteogenesis and functional bone formation [35,36,37]. The combination of BMSCs with 3D bioprinted collagen (Col)-HAPP scaffolds provides a favourable microenvironment for cell loading, high viability of BMSCs and accelerated proliferation [6,7]. The use of HAPP-based scaffolds as carriers for BMSCs shows limited osteoconductive capacity when the material is applied alone [38].

The inclusion of BMP-2 provides an additional osteoinductive stimulus, induces the differentiation of BMSCs into osteoblasts, and improves the integration of newly formed bone [5,28,32]. In this context, the study of complex platforms incorporating Col, HAPPs, BMP-2 and autologous BMSCs is particularly relevant for optimising osteoregeneration and the structural integration of bone tissue [39].

Of note, with regards to biosafety, all of the abovementioned materials have been successfully applied in the clinic. Collagen I sponges received FDA approval more than 20 years ago [40], and the addition of BMP-2 [41], HAPPs or different types of BM-derived cells [42] has also demonstrated promising results in patients.

The aim of the present study is to evaluate the effect of 3D bioprinted collagen scaffolds loaded with autologous BMSCs and enriched with BMP-2 and HAPPs on bone regeneration in rabbit calvarial defects, with a view to identify the optimal combination of biomaterials and biological stimuli for creating a favourable osteogenic microenvironment.

In addition, we present new data on the use of 3D bioprinted collagen scaffolds enriched with BMSCs, BMP-2, and HAPPs in a rabbit calvarial defect model. They prove that the combination of structural, biochemical, and cellular factors in 3D bioprinted scaffolds ensures optimal osteoregeneration. Our results could facilitate the translation of these strategies into clinical practice for the treatment of complex bone defects.

2. Materials and Methods

2.1. Animals and Ethics

In this study, 6 male New Zealand white rabbits (3 months old; 2.5 ± 0.3 kg live weight) were used for BMSC isolation and calvarial defect testing. The animals were housed and reared at the Biobase of the Faculty of Veterinary Medicine at Trakia University—Stara Zagora, under controlled conditions and with free access to a standard diet and water. All experimental procedures were approved by the national control authority—Bulgarian Food Safety Agency at the Ministry of Agriculture and Food, Sofia, Bulgaria (permit for the use of animals in experiments No. 380/2024).

2.2. Bone Marrow Aspiration

Bone marrow aspirate (~5 mL) was obtained by puncture with a Bone Marrow Aspiration Needle 16ga × 2.688in Max (Argon Medical Devices, Inc., Athens, TX, USA) from the tibia of anaesthetised rabbits. The material from each animal was collected separately in a sterile syringe with citrate buffer (Figure 1). The procedure was performed under strictly aseptic conditions. All samples were processed within 2 h after collection.

2.3. Cell Culture

Bone marrow aspirates were used for the isolation of BMSCs (including MSCs). Aspirates were diluted with PBS to 5 mL and were carefully layered on 5 mL Pnacoll (cat.# P04-60100, Pan Biotech, Aidenbach, Germany) in 15 mL centrifuge tubes. The samples were then centrifuged at room temperature for 30 min at 400× g (with the break switched off) to remove the vast majority of erythrocytes. As a result of the gradient centrifugation, the ring containing white blood cells and stromal cells was transferred with a pipette to a new 15 mL tube, washed twice with PBS and resuspended in DMEM supplemented with 10% FBS and 1% pen/strep (cat.# P04-03590, P40-37500, P06-07100, Pan Biotech, Germany). After 48 h of culture in standard conditions, non-adherent cells were gently washed off, and the remaining stromal cells were expanded for ~4 weeks in 6-well plates and T25 flasks prior to 3D bioprinting.

SCP1 human BMSCs [43] were generously provided by Assoc. Prof. Marian Draganov at the Department of Medical Biology (Medical University of Plovdiv) and cultured in standard conditions in DMEM, supplemented with 10% FBS and 1% Pen/strep (as above). Cells were split when reaching ~80% confluence at a 1:5–1:6 ratio.

2.4. 3D Bioprinting and Hydrogel Development and Characterisation

2.4.1. 3D Bioprinting

Extrusion-based 3D bioprinting was carried out with a BioX bioprinter (Cellink, Gothenburg, Sweden) and custom-made collagen type I-based hydrogels purchased from MatriChem Ltd. (Sofia, Bulgaria). For initial bioprinting optimisation and in vitro bioink biocompatibility tests, human BMSCs (SCP1 cells) were used. Even though human MSCs would not necessarily have the same behaviour as rabbit MSCs, a large number of cells were needed to produce sufficient numbers of bioprints, and that many cells could not be obtained from rabbits without sacrificing them.

Three different collagen-based osteogenic bioinks were initially tested to determine their biocompatibility properties using an established human MSC cell line—SCP1 [44]—two bioinks containing 5% and 1% HAPP and one with 80 ng/mL BMP-2. This concentration of BMP-2, albeit higher than the physiological one of around 130 pg/mL [45], falls well within what has been used in in vitro osteogenic studies (of 50–100 ng/mL [46,47]) and is considerably lower than what some studies have used for in vivo osteogenesis (50 µg/mL [48]). Each bioink contained 7 million SCP1 cells/mL (+/− 10%) that were resuspended in ~100 µL media and added to ~1 mL of hydrogel by manual mixing with two syringes connected by a luer-lock adapter. A cooling head pre-cooled at 4 °C maintained the printing temperature below 14 °C to avoid thermal crosslinking during the process. Discs with a diameter of 5 mm and a height of 0.8 mm were used as printing models, and infill density was set to 15% in the built-in BioX slicing software, which resulted in the manufacturing of constructs consisting of 2 layers. Thus, a sufficient number of bioprints with SCP1 cells were prepared to allow for an assessment of the biocompatibility of the three bioinks at three time points (day 1, 3 and 7) in triplicate (3 × 3 × 3 = 27 samples in total). The bioprints were maintained under standard culturing conditions (the same as for SCP1 cells) in DMEM supplemented with 10% FBS and 0.5% penicilin/streptomycin (all supplied by PAN Biotech, Germany) at 37 °C and 5% CO₂.

Cell viability analyses were performed by fluorescence microscopy (Nikon Eclipse Ni, Nikon, Tokyo, Japan,) after staining live cells in the whole bioprints with Calcein AM (1 µL per 200 µL of growth medium to a final concentration of 10 µg/mL, Sigma-Aldrich, Saint Louis, MO, USA) incubated for 15 min at 37 °C and subsequent staining of dead cells with Propidium iodide (1 µL per 200 µL of print medium to final concentration of 5 µg/mL) immediately before microscopy. Images were taken of the whole bioprints as sectioning would result in mechanical damage and more cell death at 4× and 10× magnification in both channels of the two dyes—FITC and Texas Red, respectively. Upon merging the two images (green and red channel) from two pictures taken from biological duplicates at 10× magnification, live and dead cells were counted manually using Nikon Eclipse 4.3.0 software in areas of ~2 mm², with the areas selected in such a way that there was a minimum of 30 cells in the selected field.

Primary rabbit BMSCs were expanded for ~5 weeks in DMEM and supplemented with 10% FBS and 1% pen/strep. Due to differences in cell growth between the different primary samples over the course of these 5 weeks, and since we wanted all autologous implants to contain approximately the same concentration of cells, we had to comply with the minimum number of cells that we could obtain from each sample. Therefore, ~1 million cells/mL were 3D bioprinted as described above with bioink 1 (containing 80 ng/mL BMP-2), bioink 2 (containing 1% HAPP) and bioink 3 (a 50:50 mixture of both resulting in 40 ng/mL BMP-2 and 0.5% HAPP final concentrations). Bioprints were then transported in the culturing medium until transplantation in rabbits.

2.4.2. Hydrogel Development and Characterisation

The development of biofunctionalised collagen bioinks, including the stable retention of growth factors and the homogenous dispersion of solid-phase nano- and micro-particles, was based on the previous know-how of Matrichem [49] and unpublished data validated by x-ray tomography. Hydrogel printability and rheology tests showed very similar properties of the hydrogels to what was previously published [49] (Figure 2).

Printability of collagen hydrogel over 1, 2, 4, 8 and 16 mm gaps demonstrated well-maintained structural integrity.

The final bovine collagen type I concentration was 20 mg/mL (2%), BMP-2 (Proteintech, HZ-1128, Rosemont, USA) was 80 ng/mL and single-phase HAPP (2.5 μm, Sigma-Aldrich, 900195, Saint Louis, MO, USA) was 1% w/v (and 5% w/v as well for the initial biocompatibility tests), which was also chosen based on previously published literature.

Characterisation of BMP-2 retention/release from the collagen hydrogel into media over 4 days was carried out by ELISA (Gentaur, AFG-E1049, Potters Bar, UK) (Figure 2B). Based on a standard curve, concentrations of BMP-2 released in the media at days 0, 1, 2, 3 and 4 were calculated and demonstrated sustained release of BMP-2 in media including on day 4 and a retention of more than 30% of the initial BMP-2 concentration, which is still in excess compared to physiological concentration and fall within the relevant range for induction of osteogenesis (Figure 2B).

2.5. Implantation of the 3D Bioprinted Scaffolds

2.5.1. Anaesthesia and Surgical Procedure

The animals were premedicated with a combination of Tiletamine/Zolazepam (Zoletil 50 mg/mL, 15 mg/kg, i.m.; Virbac, Carros, France) and xylazine hydrochloride 2% (Xylazin^®, 2 mg/kg, i.m.; Alfasan International B.V., Woerden, Netherlands). Anaesthesia was induced with 3% isoflurane and 1 L/min oxygen and maintained with 1% isoflurane and 1 L/min oxygen via an anaesthetic mask. Dexamethasone (2 mg/mL, 1 mg/kg; Alfasan International B.V., Woerden, Netherlands) was injected subcutaneously.

After removal of the hair in the calvaria area and antiseptic treatment with 2% chlorhexidine and 10% povidone-iodine, a median skin incision was made, followed by dissection of the muco-periosteal flap (Figure 3A).

Using a trephine drill (Acurata GmbH & Co KG, Thurmansbang, Germany) with continuous cooling with saline solution and at slow speed (900 rpm), four standardised calvarial defects (i–iv) were created, each with a critical size of φ = 5 mm, taking care to avoid damage to the dura mater, underlying arterial vessels, and cranial sinus (Figure 3B). The resulting calvarial disc was removed. The defects were filled with implants designed as described in

Table 1. Design of the experimental treatment of calvarial defects in rabbits.

Group	Bioink	Formulation of the Bioink	Cellular Component	Description
i	–	C-Per	–	Empty control with periosteum repositioning
ii	No1	Col-(BMP-2)	Autologous BMSCs	3D bioprinted Col disc with BMP-2
iii	No2	Col-HAPP	Autologous BMSCs	3D bioprinted Col disc with HAPPs
iv	No3	Col-(BMP-2)-HAPP	Autologous BMSCs	3D bioprinted Col disk with a 50:50 mixture of Bioink 1 and Bioink 2

. The periosteum was removed from all openings except for the empty control (i). The skin incision was closed with single interrupted absorbable sutures (3-0 PGA, Kruuse, Denmark) (Figure 3C).

After surgery, all rabbits recovered without complications.

2.5.2. Postoperative Period

The animals were placed in an oxygen chamber with controlled temperature. Daily clinical examinations were performed, with observation of the wound surface. Antibacterial therapy included Baytril 5% (10 mg/kg, i.m., 24 h, 5 days, Bayer Animal Health GmbH, Leverkusen, Germany), and analgesia was provided with Meloxidyl 1.5% (0.5 mg/kg, orally, 5 days, Ceva Santé Animale, Loudéac, France). To prevent dysbacteriosis, the probiotic Pro-digest (40 g/Versele Laga, 3 measuring spoons in 100 mL of drinking water, 10 days, Versele-Laga, Deinze, Belgien) was administered.

2.6. Computed Tomography (CT)

Five weeks after implantation of the 3D discs, the calvarial defects were assessed by computed tomography (CT) for density of the newly formed tissue and repair diameter. A SIEMENS SOMATOM Emotion Syngo 2006 A (Siemens Healthineers, Erlangen, Germany) device was used according to the following protocol: non-contrast examination (NATIV) with a slice thickness of 1.25 mm, reconstruction kernel H70s (bone filter), X-ray current 40 mA and X-ray tube voltage 110 kV. After scanning, three-dimensional (3D) images were reconstructed using RadiAnt DICOM Viewer 2025.2 software (accessed on 20 September 2025). The bone density characteristics (HU) of the calvarial defects (i–iv) were measured at 11 randomly selected points for each of them. The repair diameter was measured in a similar manner. All data are presented as mean ± standard deviation (SD).

The significant differences in bone density and residual diameters of calvarial defects treated with implants with different bioinks were determined by Univariate ANOVA analysis with post hoc Tukey test. A p-value < 0.05 was considered statistically significant. Correlation analysis was used to examine the strength and direction of the relationship between bone density (HU) of the defects and their residual diameters. Data processing was performed with IBM^® SPSS^® Statistics 26.0, Copyright 1989, 2019 (IBM Corp., Armonk, NY, USA).

2.7. Histological Examination

One week after CT scanning, euthanasia was performed with a barbiturate overdose, and the calvaria of each rabbit was carefully removed. The obtained calvarial samples were fixed with a 10% paraformaldehyde solution for 2 weeks and then decalcified in a 15% EDTA solution (cat.# 8.19040, Sigma-Aldrich, USA). The samples were then dehydrated using an ethanol gradient. After clarification with xylene, they were embedded in paraffin and cut into 4 µm thick sections. For histological evaluation, the samples were stained with Hematoxylin and Eosin (HE) and Masson-Trichrome (MT); then, they were observed by light microscopy with a microscope equipped with a digital camera (Leica DMC 2900, Leica Microsystems, Danaher, Washington, USA) and software (Leica Application Suite, version 5.0.0).

Bone regeneration was assessed by histological analysis using a semi-quantitative graded scoring system (0–3) based on predefined morphological criteria. The scoring was defined as follows: 0—absent, 1—minimal/poorly developed, 2—moderate/partially developed, and 3—extensive/well-developed bone regeneration.

The following histological parameters were evaluated: osteoid formation, organisation of lamellar bone, degree of mineralisation/calcification, formation of osteons or osteon-like structures, and the amount of fibrous connective tissue. The assessment of fibrous connective tissue was performed using an inverse scoring system, whereby a higher score corresponded to a lower amount of fibrous tissue.

3. Results

3.1. Cell Culture

Primary rabbit BMSCs were successfully isolated. The cultured BMSCs exhibited spindle-shaped, fibroblast-like morphology (Figure 4). Cells at Passage 2 were used for subsequent experiments.

3.2. Testing of Biocompatibility of Osteogenic Bioinks

After analysis of live/dead cells (Appendix A,

Table A1. Numbers of viable and dead cells used to determine cell viability and total cell numbers over time.

A	Live Cells
	5% HAPP	1% HAPP	BMP
Day 1	29	51	67
	32	53	94
Day 4	124	158	111
	72	170	166
Day 7	39	139	168
	68	144	155
B	Dead Cells
	5% HAPP	1% HAPP	BMP
Day 1	20	9	24
	14	13	25
Day 4	7	14	20
	15	22	35
Day 7	9	41	50
	6	47	38

(A and B), the percentage of live cells at days 1, 3 and 7 (Figure 5A,B) demonstrated that 5% HAPP resulted in considerable cell death on day 1 after bioprinting (~40% dead cells) and did not support sufficient cell proliferation as judged by the significantly lower cell numbers on day 7 after bioprinting. Even though studies report cell viability of above 60% with 5% HAPP, which is in line with our results [50] and previous work with patients and bioceramics for bur-holes in craniotomy using up to 50% HAPP with good regeneration without signs of toxicity [51], bioinks containing 1% HAPP or BMP provided an excellent biocompatible environment and support for cell growth (especially noticeable from day 1 to day 4).

3.3. Computed Tomography (CT)

CT showed distinct intergroup differences in the degree of bone regeneration. The most significant defect closure and highest density (HU) values were observed in group (iv) Col-(BMP-2)-HAPP, followed by groups (ii) Col-(BMP-2) and (iii) Col-HAPP, which showed an intermediate (moderate) level of regeneration. The control defect (i) C-Per demonstrated minimal recovery (Figure 6).

Quantitative statistical analysis (HU and residual diameter, mm) confirmed the results of computed tomography regarding the degree of bone regeneration. The composition of the bioink has a moderate, statistically significant effect (p = 0.001) on the bone density of calvarial defects (Figure 7). Approximately 44.2% of the variations in bone density are due to the different composition of the bioink (R^{^}2 = 0.442). The highest HU values (992.05 ± 235.22) are observed in (iv) Col-(BMP-2)-HAΡΡ. They differ significantly from both the control group (i) C-Per and group (ii) Col-(BMP-2) (742.70 ± 421.42). The lowest HU values are reported in the control group (i) C-Pr (580.32 ± 501.63), which, apart from group (iv) Col-(BMP-2)-HAΡΡ, also differed significantly from group (iii) Col-HAΡΡ (880.97 ± 313.63).

The average diameter of the defect varies significantly between the different bioinks (p < 0.01). The largest residual diameter is observed in C-Pr (4.419 ± 0.073 mm), which shows the weakest regeneration (

Table 2. Results from univariate ANOVA showing the differences in the residual diameters of calvarial defects treated with implants containing different bioinks.

	N	$\overline{\mathit{x}}\pm \mathit{S}\mathit{D}$	min	max	R^2
C-Pr	66	4.419 ± 0.073 a	4.30	4.50	0.978
Col-(BMP-2)	66	1.434 ± 0.135 ab	1.17	1.66
Col-HAPP	66	2.267 ± 0.428 abc	1.68	3.29
Col-(BMP-2)-HAPP	66	0.237 ± 0.099 abc	0.10	0.38

^a,b,c Same superscripts within the columns represent significant differences at p < 0.05; post hoc test: Tukey; SD—standard deviation; R^{^}2—coefficient of determination; N—number of observations.

). The groups treated with Col-HAΡΡ and Col-(BMP-2) showed intermediate residual diameter values (2.267 ± 0.428 mm and 1.434 ± 0.135 mm, respectively). The smallest residual diameter is reported for Col-(BMP-2)-HAΡΡ (0.237 ± 0.099 mm), reflecting the strongest regeneration of the defect. Post hoc analysis with the Tukey test shows that the average residual defect diameter of the control group C-Pr differs significantly from all other groups. The average diameter of the group of defects treated with Col-(BMP-2) differs significantly from both the control group and the other two experimental groups. This also applies to the average diameter of the defect group treated with Col-(BMP-2)-HAΡΡ, which differs significantly from the diameters of the control group and the diameters of the groups treated with Col-(BMP-2) and Col-HAΡΡ. According to the coefficient of determination (R^{^}2 = 0.978), it can be concluded that approximately 98% of the variations in the residual diameter of the defects are due to the different composition of the bioink.

A statistically significant negative correlation is observed between bone density (HU) of defects and their diameter (mm) (r = −0.512, p < 0.01), i.e., higher bone density is associated with smaller diameter. This relationship is clearly shown in Figure 8, which illustrates the relationship between the mean residual diameter of calvarial defects and bone density when treated with implants containing different bioinks.

3.4. Histological Evaluation of Bone Regeneration

Histological evaluation of calvarial defects at 5 weeks post-implantation was performed using a semi-quantitative, comparative approach based on predefined morphological criteria (Figure 9,

Table 3. Semi-quantitative histological evaluation of bone regeneration in rabbit calvarial defects at 5 weeks post-implantation.

Group/ Implant	Osteoid Formation (0–3)	Lamellar Organisation (0–3)	Mineralisation (0–3)	Osteon Formation (0–3)	Fibrous Tissue * (0–3)	Overall Regeneration
C-Per (empty control)	1	0–1	0	0	0	Lowest
Col-(BMP-2)	2	1	1	0–1	1	Intermediate
Col-HAPP	2–3	2	2	1–2	2	Advanced
Col-(BMP-2)-HAPP	3	3	3	2–3	3	Most advanced

* Fibrous tissue score reflects relative reduction of fibrous connective tissue (0 = abundant fibrous tissue; 3 = minimal fibrous tissue).

). These included the presence and organisation of osteoid trabeculae, the degree of lamellar bone formation, evidence of mineralisation, the occurrence of osteon-like structures, and the relative proportion of fibrous connective tissue. This methodology was applied to enable systematic comparison of regenerative patterns between experimental groups rather than to provide absolute quantitative measurements.

In the control group (C-Per), regenerative activity was limited. The defect area was predominantly occupied by fibrous connective tissue, with only sparse areas of woven bone and minimal lamellar organisation, reflecting the restricted spontaneous regenerative capacity of the periosteum alone.

The Col-(BMP-2) group demonstrated moderately developed osteoid trabeculae accompanied by activated osteoblasts and partial lamellar organisation. However, a considerable amount of residual fibrous tissue remained, and the newly formed bone lacked a fully organised lamellar architecture, indicating enhanced osteoinductive stimulation without advanced structural maturation.

In contrast, the Col-HAPP group exhibited a more advanced regenerative pattern, characterised by thicker osteoid lamellae, early mineralisation, and a reduced proportion of fibrous connective tissue. The presence of hydroxyapatite-based particles likely provided osteoconductive support, promoting more organised lamellar bone formation despite the absence of an additional osteoinductive growth factor.

The most advanced histological features were observed in the Col-(BMP-2)-HAPP group. This group showed abundant lamellar osteoid trabeculae, well-developed cancellous bone, and frequent osteon-like structures, together with minimal residual fibrous tissue, indicative of a more mature and structurally organised regenerative outcome compared with the other experimental groups.

Overall, the degree of bone regeneration at 5 weeks followed the order C-Per < Col-(BMP-2) < Col-HAPP < Col-(BMP-2)-HAPP. This hierarchy is consistent with the semi-quantitative histological assessment and is further supported by corresponding computed-tomography-derived parameters. Specifically, groups exhibiting more advanced lamellar organisation and mineralisation also demonstrated higher Hounsfield unit values and smaller residual defect diameters, supporting the validity of the comparative histological evaluation despite the absence of formal histomorphometric analysis.

4. Discussion

This study presents new data on the importance of the composition of 3D bioprinted implants on osteogenesis in critical calvarial defects in a rabbit model and clearly demonstrates that different biomaterial formulations lead to different regenerative outcomes. This is particularly important given the well-documented limited ability of the periosteum to achieve spontaneous osteogenesis in such defects [52,53,54]. Our observations confirm that although the periosteum has a certain natural regenerative potential, it is insufficient to repair critical calvarial defects, especially when additional cellular and biomaterial support is lacking. In confirmation, the control C-Per group in the present study, in which the defect was filled only with repositioned periosteum, demonstrated a minimal regenerative response, as reflected by the lowest HU values (580 ± 501), the largest residual defect diameter (4.419 ± 0.073 mm) and limited woven bone formation with minimal lamellar structuring. These results are consistent with previous reports showing that the absence of BMSCs and a biomaterial carrier leads to poor osteogenesis [52,53]. They are also consistent with the observations of Ratiu et al. [45], who reported that the periosteal contribution to bone repair is minimal and delayed, especially in the early stages of implant osseointegration. Our results correspond to some extent with Caplan’s [55] classic view that periosteal cells are critical for fracture healing, but unlike fractures, bone defects require additional structural, biochemical, and cellular support for effective regeneration. In this context, the periosteal control group in our study represents a critical reference point that clearly demonstrates the limitations of the natural regenerative capacity of the periosteum and highlights the need for integrated bioengineering strategies, including the biofabrication of multicomponent 3D implants.

Bone regeneration is a complex multifactorial biological process that integrates a wide range of local cell populations [56,57,58]. All of these play a role, together with a set of hormonal growth factors, in the mechanisms of bone repair. The classic studies by Friedenstein [57] emphasise that bone induction requires a suitable microenvironment that allows for the coordinated action of mechanical, biochemical and humoral signals. Contemporary data confirm that the microenvironment can be remodelled and that the body’s own resident stem cells can form new tissue after stimulation by bioactive factors secreted by exogenously delivered mesenchymal stromal cells [59]. This highlights the key role of the bioprinted scaffold as an integrating element between cell signals and structural support in critical bone defects. It is on this conceptual basis that the design of our 3D bioprinted collagen scaffolds is built, allowing simultaneous delivery of an osteoconductive matrix, osteoinductive biochemical signals, and BMSCs. In this regard, all bioprinted implants in the present study contained autologous BMSCs, while the control relied solely on the periosteum. The BMSCs were not pre-treated with osteogenic differentiation, allowing for the evaluation of the ability of different bioinks and combined factors to stimulate bone regeneration through natural mechanisms in vivo. Since BMSCs do not express MHC-II and RLA-DRA and have a reduced ability to activate CD4+ T cells [34], they are suitable for autologous implantation, and their in vivo effect is mainly due to secreted cytokines and growth factors [60]. The presence of BMSCs likely facilitated the early osteogenic response in all implants, with differences between groups primarily reflecting the influence of material formulation.

This is the first study, to our knowledge, to combine this specific set of technology and components in an animal model—3D bioprinting of collagen type I-based bioinks functionalised with BMP-2, HAPPs, or both, also containing autologous BMSCs to evaluate the potential of such implants for in vivo bone regeneration. Several recent review articles on 3D bioprinting for bone regeneration discuss strategies involving collagen bioinks modified with HAPPs, combined with GelMA or hyaluronic acid, or enhanced with PEGDA [61,62]. The collagen-based bioink from the present study is an optimal choice for bone tissue applications due to its biocompatibility, bioactivity, and ability to promote adhesion and osteogenic differentiation of BMSCs [32,61,62,63].

The integration of HAΡΡs into the collagen matrix further improves mineral stability and osteoconductivity [28,64,65], mechanical strength [66], and rheological properties required for 3D extrusion printing [6,67]. Our in vivo results are consistent with the well-documented role of HAPPs as a matrix for bone invasion and remodelling, which facilitates the early structuring of newly formed bone. In addition, the interaction between Col and HAΡΡs forms a stable composite that supports the bioprinting process through suitable rheological characteristics and increased mechanical strength of the final scaffold [6,67].

Very few studies have used BMP-2 for in vivo application in combination with collagen-based hydrogels [68]. For example, BMP-2 has been combined with collagen sponges as well as more complex 3D printed scaffolds involving PCL and gelatin-alginate hydrogels in dogs, but without autologous BMSCs [69]. These approaches, despite the lack of HAPPs, or a cellular component, result in significant mineralisation at week 4, which is consistent with our observations. Despite the stimulating effect of BMP-2, Col-(BMP-2) without HAΡΡs shows lower density and larger residual defect diameter, highlighting the key role of structural support for effective bone regeneration.

In the course of this study, the addition of BMP-2 to the collagen scaffold Col-(BMP-2) has led to a moderate improvement of regeneration confirmed by the interim HU values and the decreased residual diameter of the calvarial defect. Histologically, more pronounced osteoid trabeculae and active osteoblasts were observed, indicative of stimulated osteoinduction. However, fully organised lamellar formation was not achieved, indicating that BMP-2 alone is not sufficient for accelerated structural remodelling without a suitable osteoconductive platform. Although it contains only an osteoconductive component, Col-HAΡΡ demonstrated more advanced regeneration compared to Col-(BMP-2) at 80 ng/mL (that is, a concentration higher than the physiological one), histologically showing thick osteoid lamellae and early calcification. These results highlight the presence of the HAPP-mineral frame as perhaps more decisive in early bone organisation than isolated BMP-2 (albeit this study was limited to a single concentration) and reveal a unique relationship between structural support and biochemical stimuli. This reinforces the observation that the optimal effect of BMP-2 on bone regeneration is achieved by combination with mineralised materials, such as HAPPs [70]. It should be noted that, within the 5-week observation period, complete resorption of HAPP cannot be expected, which may have influenced CT measurements of bone density, especially at the higher concentration of 1% used compared to the 0,5% in the Col-(BMP-2)-HAPP bioink. Therefore, the CT results are interpreted cautiously and primarily in a comparative context between groups, with histological analysis supporting the observations. A longer observation period would be necessary to more precisely assess long-term mineralisation and bone remodelling. A statistically significant negative correlation between bone density (HU) and defect diameter was observed exclusively in the Col-(BMP-2)-HAPP group, while no such trend was seen in the Col-(BMP-2) or Col-HAPP groups. This likely reflects a synergistic effect of the combined scaffold, as BMP-2 and HAPPs individually stimulate bone formation via distinct mechanisms that do not necessarily produce a linear relationship between mineralisation and defect closure. In contrast, the combined scaffold, albeit at lower concentrations of both BMP-2 and HAPPs, promotes more coordinated and accelerated bone maturation, facilitating the transition from osteoid to lamellar bone. These findings suggest that integrating biochemical and structural cues can achieve a more predictable and uniform regeneration process, where increased mineral density is closely associated with effective defect reduction.

Moreover, our data are further supported by Moeinzadeh et al. [71], who used injectable collagen–alginate hydrogels with BMP-2 in rats. Although the authors only perform an in vitro assessment of the effect of this hydrogel on MSCs (rather than using autologous BMSCs in rat experiments), they conclude that this strategy has excellent osteogenic potential [71]. In the context of the collagen–HAΡΡ combination, the study by Koo et al. [72] is closest to our design, using posterolateral lumbar spinal fusion in mice with a Col-HAΡΡ hydrogel enriched with human adipocyte-derived MSCs. Similarly to our results, the authors conclude that the combination provides improved bone regeneration in the context of osteoporosis. Likewise, Wang et al. [73] demonstrate that 3D-printed scaffolds based on a titanium alloy (not collagen-based, as in our study) loaded with BMP-2 improve osseointegration in osteoporosis by supporting the proliferation, viability, and differentiation of BMSCs in vitro, leading to a significant improvement in bone regeneration.

Despite the positive results, however, the present study has several limitations. First, the experiment was conducted on a rabbit model at a single time point of observation of the effects, which requires careful interpretation when translating to larger animal species or humans [37,74,75]. Second, despite the demonstrated early mineralisation and structural organisation, the follow-up was limited to 5 weeks, which does not allow for an assessment of long-term bone remodelling and the stability of the newly formed tissue. Third, although Col-(BMP-2)-HAΡΡ implants show a synergistic effect at two times lower concentrations than in the single BMP-2 or HAPP bioinks, the optimal dose of BMP-2 and scaffold parameters require further preclinical safety and efficacy evaluations [70]. Similarly, the functional state and the number of BMSCs in the scaffold would also have a critical impact on the outcome. Unfortunately, neither the initial characterisation with biomarkers like CD90, for example, nor the in vivo tracking of the fate of the cells after implantation were within the scope of this study. However, previously published literation on these issues suggests that the in vitro culturing methodology utilised herein provides >90% enrichment of BMSCs [76,77] and that their effect after implantation is most likely double-edged—both through osteogenic differentiation and through paracrine signalling to recruit other resident cells that can enhanced regeneration [78]. Moreover, it is important to note that biomechanical evaluations of the regenerated bone were also beyond the scope of this study, and further investigations would be required to investigate if the therapeutic approach proposed herein can be applicable to weight-bearing bones too. In addition, the histological assessment was conducted using a qualitative and semi-quantitative comparative approach. This method allows for reliable detection of differences in regenerative patterns between groups but does not provide absolute measurements of newly formed bone volume or the degree of mineralisation. Therefore, conclusions regarding the effectiveness of the different implant formulations should be interpreted as comparative rather than definitive quantitative results. Lastly, standardisation of bone regeneration assessment methods is needed to facilitate comparability between studies.

These limitations outline the prospects for future research, including long-term in vivo investigations, optimisation of multicomponent implants, and translation to clinical applications in the treatment of critical calvarial defects or traumatic brain injuries.

5. Conclusions

The present study demonstrates that the composition of collagen-based bioink for 3D bioprinted implants is crucial for the effectiveness of bone regeneration in calvarial defects in rabbits. The control group, treated with periosteum alone, demonstrated limited natural regenerative potential, while the addition of BMP-2 stimulated moderate osteoinduction. The inclusion of HAPPs improves the mineralisation and structural formation of the newly formed bone, but the true maximum effect is achieved by combining BMP-2, HAPPs and autologous BMSCs. This combination demonstrates a synergistic effect, resulting in the highest bone density, smallest residual defect, and well-organised lamellar and cancellous bone. The results emphasise that optimal bone repair requires an integrated approach involving structural, biochemical and cellular support.

These findings provide novel compelling evidence for the importance of multicomponent, cell-enriched 3D bioprinted scaffolds as a potential therapeutic strategy for the treatment of critical bone defects and may serve as a basis for future research and translational applications in clinical practice.