Characterization and functional analysis of the adipose tissue-derived stromal vascular fraction of pediatric patients with osteogenesis imperfecta

Pediatric patients with Osteogenesis Imperfecta (OI), a heritable connective tissue disorder, frequently suffer from long bone deformations. Surgical correction often results in bone non-unions, necessitating revision surgery with autogenous bone grafting using bone-marrow-derived stem cells (BM-SC) to regenerate bone. BM-SC harvest is generally invasive and limited in supply; thus, adipose tissue's stromal vascular fraction (SVF) has been introduced as an alternative stem cell reservoir. To elucidate if OI patients' surgical site dissected adipose tissue could be used as autologous bone graft in future, we investigated whether the underlying genetic condition alters SVF's cell populations and in vitro differentiation capacity. After optimizing SVF isolation, we demonstrate successful isolation of SVF of pediatric OI patients and non-OI controls. The number of viable cells was comparable between OI and controls, with about 450,000 per gram tissue. Age, sex, type of OI, disease-causing collagen mutation, or anatomical site of harvest did not affect cell outcome. Further, SVF-containing cell populations were similar between OI and controls, and all isolated SVF's demonstrated chondrogenic, adipogenic, and osteogenic differentiation capacity in vitro. These results indicate that SVF from pediatric OI patients could be used as a source of stem cells for autologous stem cell therapy in OI.

Isolation and cellular outcome of the stromal vascular fraction from pediatric adipose tissue. The overall mean weight of harvested adipose tissue was 6.86 g (SEM: 1.17 g) in the OI group and 7.48 g (SEM: 3.82 g) in controls. Within about 90 min, adipose tissue samples were processed to obtain SVF from HC and OI donors. All 40 SVF isolations were successful. The overall average number of nucleated cells per gram of adipose tissue was 443,000 cells per gram tissue (SEM: 6.66 × 10 4 cells/g) in OI and 462,000 cells per gram tissue (SEM: 17.13 × 10 4 cells/g) in HC (Table 1). Cell outcome did not vary significantly with sex, age, type of OI, disease-causing collagen mutation, or anatomical site of harvest (Supplemental Table A.2).

and A.4).
HC-and OI-ASC reveal similar multi-lineage differentiation capacity. When SVF cells were cultured under osteogenic conditions, expression of Alkaline Phosphatase (AlkP), Runx2, Sox9, and Osteocalcin (OCN) at day 7 and day 21 was higher than in non-treated samples. HC and OI samples showed similar results ( Fig. 2A,B). Alizarin red staining of deposited minerals at day 21 confirmed successful differentiation towards functional osteoblasts (Fig. 2C,D). When SVF cells were cultured under adipogenic conditions, PPARy, Leptin receptor (LEPR), and Leptin were significantly upregulated and revealed similar results for HC and OI samples (Fig. 3A,B). At day 21, oil droplet accumulation confirmed adipogenic differentiation (Fig. 3C). Under chondrogenic conditions, we observed significant upregulation of Collagen 10, Sox7, and Aggrecan (ACAN) with comparable results for HC and OI cells (Fig. 4A,B). Alcian blue staining confirmed chondrogenic differentiation (Fig. 4C). Evaluation of the influence of collagen mutation on multi-lineage differentiation capacity revealed similar differentiation competence towards osteogenic and adipogenic lineage of SVF cells with COL1A1, COL1A2, and WNT1 mutations. Regarding chondrogenic lineage, SVF cells harboring COL1A1 mutations seemed to have a higher differentiation competence than mutations in other genes. But, due to the heterogenous sample distribution (OI-COL1A1, n = 1; OI-COL1A2, n = 6; OI-WNT1, n = 1), a conclusion cannot be drawn yet.

Discussion
In this study, we successfully demonstrated the isolation of SVF from resected adipose tissue of pediatric patients with OI and HC within about 90 min. The number of isolated cells and cell viability was comparable between OI and HC and independent of age, sex, anatomical site of harvest, or genetic OI mutation. Additionally, SVF obtained from OI and HC revealed similar amount of stem cells and tri-lineage differentiation capacity in vitro.
To isolate SVF containing ASCs, traditionally harvested lipoaspirate is exposed to enzymatic dissociation followed by several centrifugation steps 10,18-22 . This is a relative time-consuming procedure and could not be www.nature.com/scientificreports/ performed in OI if isolated SVF is immediately be used for induction of bone regeneration within the same bone-corrective surgery. Therefore, we performed SVF isolation according to the technique described by Tevlin et al. 17 with minor modifications in order to use this method for resected adipose tissue and optimized processing time. With this protocol, we were able to isolate SVF with about 90% viable cells from OI and HC. This yield is comparable to established non-intraoperative isolation protocols and intraoperative isolation procedures applied to adipose tissue or lipoaspirates 23 . Further, isolated ASCs presented the typical stem cell markers and quantity of ASCs isolated from dissected adipose tissue comparable to previously published yields achieved from pediatric and adult patients 24 and by liposuction and enzymatic ASC isolation (about 25%-30%) 21 . Additionally, we found that the yield of ASCs was irrespective of anatomical harvest site as described before 25,26 . Yet, the stromal cell population including pericytes, ASCs, and supra-adventitial stromal cells, are the most important cell types in regenerative therapies because of their multi-lineage differentiation capacity 15,27 . Supraadventitial stromal cells and pericytes are both identified as precursor cells of ASCs, although there still remains some discussion [28][29][30][31] . Composition of isolated SVF from OI and HC was similar to each other and comparable to published data 23 . Still, SVF from OI patients revealed a significantly lower percentage of endothelial progenitor cells, which play a role in angiogenesis 32 . Angiogenesis is a key factor in bone repair as new blood vessels bring oxygen and nutrients to the highly metabolically active regenerating callus and serve as a route for inflammatory cells, cartilage, and bone precursor cells to reach the injury site 33 . The lower percentage of endothelial progenitor cells in OI suggests a diminished capacity of angiogenesis in OI and remains a question for future studies.
Regarding multi-lineage differentiation capacity, ASCs from OI and HC had similar osteogenic, adipogenesis, and chondrogenic differentiation capacity and are in line with published studies of human ASCs 10,15,[34][35][36][37] . Concerning the underlying collagen mutation, we found similar differentiation capacity towards osteogenic and adipogenic lineage. ASCs derived from OI patients with COL1A1 mutation seemed to have higher chondrogenic www.nature.com/scientificreports/ differentiation capacity suggesting that they are more susceptible to TGF-β induced chondrogenesis. TGF-β is known to stimulate chondrogenic differentiation 38 . Further, TGF-β seems to play a role in OI pathology as mouse models of recessive (mutation in the Crtap gene; Crtap −/− ) and dominant (collagen type I mutations; Col1a2 tm1.1Mcbr & Col1a1 Jrt/+ ) OI showed excessive TGFβ-signaling in the skeleton 39,40 . Interestingly, anti-TGFβ treatment using a neutralizing antibody corrected bone fragility in Crtap −/− and Col1a2 tm1.1Mcbr mouse model but not in the Col1a1 Jrt/+ mouse model, suggesting a link between collagen mutation and TGF-β signaling 39,40 . However, our sample size was too small to draw definite conclusions, and further studies are needed. A limitation of our study is that we investigated pediatric patients only as it has been shown before that ASC isolation and bone regeneration/wound healing of autologous transplanted ASCs in patients between 6 and 72 years of age were similar 24,41,42 . Further, we did not compare multi-lineage differentiation of isolated ASCs to BM-MSCs as it also has been demonstrated before that compared to BM-MSCs, ASCs have a better resistance to cell senescence 43,44 and are more effective in multi-lineage differentiation [44][45][46][47][48] . Additionally, we did not investigate cell senescence as it was demonstrated before that ASCs cell characteristics are stable for up to 10 passages 49 .
Nevertheless, an important question remains if ASCs from OI show bone regeneration capacity in vivo. In general, ASCs bone regeneration potential in combination with bioengineered scaffolds has been proven in various animal models 50 with calvarial like-defect 51-54 , femoral head osteonecrosis 55 , femur defect 56 , distraction osteogenesis 57 , and spine fusion 58 . Additionally, ASCs bone regeneration potential has also been evaluated in case studies and small-size clinical trials in humans with cranial defects, cranio-maxillofacial skeleton defect, or osteoarthritis [59][60][61][62] . But it still needs to be evaluated if SVF from OI patients have the capability of bone regeneration in e.g., non-union fracture animal models or mouse models of OI. Mechanistically, we hypothesize that transplanted SVF will promote bone regeneration at the surgical site by "SVF-cells"-produced paracrine factors 63 and by ASCs osteogenic differentiation ability itself. Furthermore, studies are also needed to evaluate the optimal delivery system of SVF to the desired site in OI. Encouraging results were recently published showing successful bone regeneration of undifferentiated temporomandibular joint synovial-fluid-MSCs from patients with temporomandibular dysfunctions on 3D polyetherketoneketone scaffolds in a rabbit calvarial critical-sized defect 64 .
Still, our study aimed to shed light on whether adipose tissue-derived SVF, taken from a pediatric OI-patient, can serve as autologous tissue/cell/bone graft to promote bone regeneration at the surgical site resulting in the prevention of a bone non-union in the same patient. It should be kept in mind that stem cells from OI patients contain mutations that cause OI; thus, the newly generated bone matrix will still be fragile. If the final therapy goal is to repair OI bones or produce a healthy bone matrix, stem cells from healthy individuals would be the www.nature.com/scientificreports/ better approach. Although this strategy has its drawbacks of (a) finding a donor (best: age-and sex-matched), (b) painful bone marrow harvest from the donor, and (c) graft-versus-host reaction in the OI patient leading to rejection of the transplant and other yet unknown consequences. Additionally, obtaining SVF from normal individuals would entail going through numerous regulatory hurdles imposed by e.g., FDA and Health Canada.
In conclusion, our study demonstrated the feasibility of isolating SVF-containing ASC from adipose tissue of pediatric OI patients. We demonstrated that yields of isolated ASC from OI patients are comparable to ASCs from healthy controls. And we verified that isolated ASCs from OI patients express the same stem cell markers and possess multi-lineage differentiation capacity as controls. Most importantly, osteogenic differentiation potential was irrespective of OI mutation. Thus, as a platform for future therapeutic use, SVF-containing ASC can be isolated within the same surgery and immediately be used for bone regeneration in OI patients.

Material and methods
If not indicated otherwise, Supplemental Table A SVF isolation. Isolation of SVF was performed according to the optimized method by Tevlin et al. 17 with minor modifications. Briefly, harvested adipose tissue samples from the site of the surgical incision, were weighed and incubated in fetal bovine serum (FBS)-free ice cold culture medium (DMEM:F12 supplemented with 1% antibiotic-antimycotic and 1% penicillin-streptomycin) for 10 min at 4 °C. Then, adipose tissue samples were minced manually into small 1 × 1 mm pieces using sterile surgical scissors and homogenized using a 25 ml Sarstedt serological pipette. Subsequently, homogenized samples were mixed in a ratio of 1:2 (weight per volume, W/V; gram of adipose tissue/buffer) with freshly prepared digestion-collagenase buffer by dissolving 2.2 mg/ml collagenase (collagenase NB 6, GMP grade) in Hank's balanced salt solution supplemented with 10% bovine serum albumin (BSA) and incubated in a water bath at 37 °C for 60 min with vigorous shaking every 10 min for 1 min. Afterward, collagenase was neutralized by adding culture medium, followed by filtration using 100 µm membrane filters, and centrifugation at 700×g for 10 min. The pellet was collected as SVF (Fig. 5) and the number of nucleated cells counted using trypan blue. Finally, SVF was prepared for flow cytometry or cultured for multi-lineage differentiation.  www.nature.com/scientificreports/ we analyzed the presence of "activated" mesenchymal stem cells (CD45−/CD34+/CD73+/CD90+/CD105+). Applied gating strategy is depicted in Fig. 6.
Multi-lineage differentiation. Directly after isolation, SVF was seeded and cultured in culture medium in T25-culture flasks and non-adherent cells were removed the next day. Cells were harvested by TrypLE Express Enzyme at 80% confluence and then seeded at a density of 3000 cells/cm 2 in T75-culture flasks. The medium was replaced every 3 days. Cells of passages 2 to 4 were used for multi-lineage differentiation. For multi-lineage differentiation, cells were seeded at a density of 3000 cells/cm 2 in 6-well plates for histochemical analysis and 12-well plates for RNA extraction. At 80% confluence, culture medium was replaced by osteogenic, adipogenic, or chondrogenic maintenance medium (referred as 'treated') or grown again in culture medium as control cells (referred as 'non-treated'). For osteogenic and adipogenic differentiation, cells were exposed to induction media for 3 days, followed by maintenance media until the end of the experiment (Table 3). For chondrogenic differentiation, cells were exposed to chondrogenic induction medium during the entire experiment (Table 3). Media were replaced every 3 days for 21 days.
Assessment of multi-lineage differentiation by histochemistry. On day 21 of culture, cells were fixed with either 70% ethanol for 60 min at − 20 °C for Alizarin Red staining or with 4% formalin for 60 min at room temperature for Oil Red or Alcian blue staining. Alizarin Red, Oil Red, and Alcian blue staining were performed at room temperature according to manufacturers' instructions. Briefly, Alizarin Red staining was performed for 10 min, followed by washes with distilled water and PBS; Oil Red staining was performed for 15 min; and Alcian blue staining for 120 min. Pictures were taken with LEICA DMRB microscope equipped with an Olympus DP70 digital camera, 10 ×/0.30 PL FLUOTAR objective or 40 ×/0.70 PL FLUOTAR objective, and the DP controller software.  Figure 32.3%). (E) Hematopoietic population was further characterized by using CD14 and CD3 markers to separate monocytes/macrophage population (in Figure 28.9%), which was further characterized by using (F) CD14 and CD206 markers to individualize monocytes (M1, CD45 + /CD14 + /CD206 -, in Figure 36.5%) and macrophages (M2, CD45 + /CD14 + /CD206 + , in Figure 62.7%). Non-hematopoietic cell population was characterized by using (G) CD146 marker to identify pericytes (CD45 -/CD34 -/CD146 + ; 2.87%) and (H) CD31 marker to separate cells from endothelial origin (CD45 -/CD34 + /CD31 + ; 7.26%), and supra-adventitial stromal cells (CD45 -/CD34 + /CD31 -; 77.8%). (I) Supra-adventitial stromal cells and non-hematopoietic cell population were pooled and further analyzed for the presence of early mesenchymal stem cells (CD45 -/CD34 + /CD73 + / CD90 + ; in Figure 97.3%) using CD73 and CD90 markers. (J) Applying the marker CD105, early mesenchymal stem cells were further separated for the presence of activated mesenchymal stem cells (CD45 -/CD34 + /CD73 + / CD90 + /CD105 + ; in Figure 0.529%).  www.nature.com/scientificreports/ either using Mann -Whitney-Wilcoxon U-Test for 2 independent parameters such as HC vs OI, sex, and anatomical site, following established critical values of the smallest rank sum test 65 . Kruskal-Wallis-H test was used for evaluation of statistical difference for > 2 independent parameters like age-dependency, type of OI, and disease-causing collagen mutation. Normally distributed data were evaluated using independent T-test for 2 independent samples or one-way ANOVA for > 2 independent samples. Calculations were performed using SPSS software (v 24.0; SPSS Inc). Statistical differences in gene expression between treated and non-treated samples were analyzed by paired T-test. Significant differences in genetic fold-change of OI versus HC were assessed by unpaired T-test. Calculations were performed using GraphPad Prism (v 8.1.1; GraphPad Software, www. graph pad. com) and p < 0.05 was considered significant. www.nature.com/scientificreports/