Prefabrication of a large pedicled bone graft by engineering the germ for de novo vascularization and osteoinduction

Accepted Manuscript Prefabrication of a large pedicled bone graft by engineering the germ for de novo vascularization and osteoinduction Christian Epple, Alexander Haumer, Tarek Ismail, Alexander Lunger, Arnaud Scherberich, Dirk Johannes Schaefer, Ivan Martin PII:

S0142-9612(18)30785-3

DOI:

https://doi.org/10.1016/j.biomaterials.2018.11.008

Reference:

JBMT 18972

To appear in:

Biomaterials

Received Date: 15 June 2018 Revised Date:

5 November 2018

Accepted Date: 8 November 2018

Please cite this article as: Epple C, Haumer A, Ismail T, Lunger A, Scherberich A, Schaefer DJ, Martin I, Prefabrication of a large pedicled bone graft by engineering the germ for de novo vascularization and osteoinduction, Biomaterials (2018), doi: https://doi.org/10.1016/j.biomaterials.2018.11.008. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Prefabrication of a large pedicled bone graft by engineering the germ for de novo

2

vascularization and osteoinduction.

3

1,a,c

1,a,b

RI PT

4 , Tarek Ismail c, Alexander Lunger c, Arnaud

5

Christian Epple

6

Scherberich a,b,c, Dirk Johannes Schaefer c, Ivan Martin a,b,*

7

1

Christian Epple and Alexander Haumer contributed equally to this study

8

a

Department of Biomedicine, University Hospital Basel, University of Basel, Switzerland

9

b

Department of Biomedical Engineering, University of Basel, Switzerland

10

c

Department of Plastic, Reconstructive, Aesthetic and Hand Surgery, University Hospital

11

Basel, Switzerland

12

* Corresponding author: Prof. Ivan Martin, Department of Biomedicine, University Hospital

13

Basel, Hebelstrasse 20, Basel 4031, Switzerland.

16

17

SC

M AN U

TE D EP

15

AC C

14

& Alexander Haumer

18 19 20 1

ACCEPTED MANUSCRIPT Graphical Abstract

SC

RI PT

21

M AN U

22

Keywords (4-6 words):

24

Large, complex bone defect; graft prefabrication; vascularized bone graft; Off-the-shelf

25

material; extracellular matrix; regenerative medicine

26

TE D

23

Funding sources

28

This work has received funding from the People Programme (Marie Curie Actions) of the

29

European Union's Seventh Framework Programme FP7/2007-2013/under REA grant

30

agreement No. 607868 (iTERM, to A. Scherberich) and from the Swiss National Science

31

Foundation (SNF Grant 310030-156291 to A. Scherberich and I. Martin).

AC C

32

EP

27

33

34

2

ACCEPTED MANUSCRIPT Abstract

36

Large and complex bone defects represent challenging clinical scenarios, typically requiring

37

autologous vascularized bone transplants. In order to bypass the numerous associated

38

limitations, here we aimed at ectopically prefabricating a bone graft surrogate with vascular

39

pedicle.

40

A hollow cylinder of devitalized cancellous bone was used to define the space of a large

41

bone substitute. This space was filled with devitalized pellets of engineered hypertrophic

42

cartilage as bone-inducing material, in combination or not with stromal vascular fraction

43

(SVF) of adipose tissue as source of osteoprogenitors and endothelial cells. Vascularization

44

of the space was targeted through axial insertion of an arterio-venous (AV) bundle.

45

Constructs were subcutaneously implanted in nude rats for 12 weeks and analyzed for bone

46

formation and vascularization by histology and microtomography.

47

Retrieved constructs were extensively vascularized in all conditions, with vessels sprouting

48

from the AV bundle and reaching a higher density in the axially central volume. Bone tissue

49

was formed through remodeling of hypertrophic cartilage, and quantitatively correlated with

50

de novo vascularization.

51

Our study demonstrates feasibility to prefabricate large, pedicled bone grafts in predefined

52

shapes. The combination of an AV bundle with engineered hypertrophic cartilage provided a

53

germ for the coupled processes of vascularization and bone formation. The demonstrated

54

osteoinductivity of devitalized hypertrophic cartilage offers the opportunity of implementing

55

the proposed regenerative surgery strategy through off-the-shelf materials.

56

Introduction

57

The most effective standard of treatment for large and complex bone defects, for example in

58

the maxillo-facial region, relies on the use of autologous vascularized bone grafts (e.g.,

59

pedicled fibula). This is associated with several bottlenecks, such as limited availability and

60

significant donor site morbidity. Furthermore, shaping of complex, three-dimensional

61

structures with available autologous bone can be cosmetically and functionally insufficient,

AC C

EP

TE D

M AN U

SC

RI PT

35

3

ACCEPTED MANUSCRIPT with impeded rehabilitation. An engineered osteogenic material combined with a vascular

63

bundle would represent an attractive alternative to develop a pedicled bone graft surrogate

64

[1,2]. In this regard, arteriovenous (AV) fistulas have proved to be efficient in

65

neovascularization of tissues for the past 40 years [3]. Even though the performance of AV

66

loops as compared to AV bundles is slightly higher in terms of amount of generated tissue,

67

the latter reach superior tissue organization at earlier times [4]. This property, together with

68

the straightforward clinical applicability, guided our selection of the AV bundle as versatile

69

and efficient vascularization tool.

70

In a recently established rat model, we demonstrated that stromal vascular fraction (SVF)

71

cells from human adipose tissue, containing both mesenchymal progenitors and endothelial-

72

lineage cells [5], seeded on a hydroxyapatite scaffold and combined with a vascular AV

73

bundle, efficiently vascularized necrotic bone and formed bone tissue de novo by direct

74

ossification [6]. However, the procedure required culturing of the SVF cells in osteogenic

75

medium for 5 days prior to implantation, and newly formed bone was found in limited

76

amounts and restricted regions.

77

In the past, recapitulating the endochondral route of ossification has proved to be a robust

78

strategy for bone formation in ectopic models as well as in orthotopic non-union models [7–

79

11]. The endochondral route of ossification can be recapitulated by generation of

80

hypertrophic cartilage templates in vitro, which embed the biological signals for the

81

remodeling into a complete bone organ when implanted in vivo. Recently, we demonstrated

82

that these hypertrophic cartilage templates can form bone ectopically and orthotopically after

83

devitalization and ‘reactivation’ by stromal vascular (SVF) fraction cells [12]. Using multiple

84

small modules - engineered & devitalized hypertrophic cartilage chips - would allow for

85

reconstruction of complex, three-dimensional defects.

86

The aim of this study was to prefabricate an engineered, vascularized, pedicled bone graft

87

substitute and to validate it in a rat subcutaneous environment. The design is based on the

88

following three pillars: (i) an outer shell consisting of devitalized cancellous bone

89

(Tutobone©), which defines the shape of the engineered substitute; (ii) an inner core filled

AC C

EP

TE D

M AN U

SC

RI PT

62

4

ACCEPTED MANUSCRIPT with devitalized hypertrophic cartilage, as off-the-shelf material able to induce bone formation

91

in vivo; and (iii) an AV bundle, axially implemented to provide a strong and reliable

92

vascularization of the graft, as well as the pedicle for a later anastomosis to the recipient site.

93

Methods

94

Culture and devitalization of hypertrophic cartilage micromasses

95

Human bone marrow derived stromal cells (hBMSCs) from three donors (all male, 33.3 ±

96

11.7 years) were expanded as monolayers for two to four passages as previously described

97

[13]. hBMSCs were detached and distributed in 1.5 ml screw cap Eppendorf tubes with

98

0.5x106 cells per tube and centrifuged at 300g for 5 minutes. The resulting micromass pellets

99

were then cultured in serum-free medium (Dulbecco’s modified Eagle’s medium, 1.25 mg/ml

100

human serum albumin, 10 mM HEPES, 1 mM sodium pyruvate, 100 U/ml penicillin, 100

101

mg/ml streptomycin, 0.29 mg/ml glutamate, and ITS-A [10 mg/ml insulin, 5.5 mg/ml

102

transferrin, 5 ng/ml selenium, 0.5 mg/ml bovine serum albumin]; all from Invitrogen),

103

supplemented with 10 ng/ml transforming growth factor-b3 (R&DSystems), 10-7 M

104

dexamethasone, and 0.1 mM ascorbic acid 2-phosphate (Sigma-Aldrich, St. Louis, MO,

105

USA, https://www.sigmaaldrich.com) (chondrogenic medium). After 3 weeks of chondrogenic

106

differentiation the resulting cartilaginous pellets were further cultured in hypertrophic medium

107

(serum-free medium with 50 nM thyroxine, 10 mM b-glycerophosphate, 10-8 M

108

dexamethasone, 0.1 mM ascorbic acid 2-phosphate, (Sigma-Aldrich) for 2 weeks as

109

previously described with medium changes twice weekly [12]. A part of the resulting pellets

110

were devitalized by three cycles of freezing (-196°C for 10 minutes) and thawing (37°C for 10

111

minutes) and a final wash with deionized water. All fluids were removed and pellets stored at

112

-80°C until further use (devitalized pellets condit ions).

113

SVF isolation

114

SVF cells were isolated from two liposuction and one excision fat samples from 3 donors

115

(28.3 ± 3.7 years, two females and one male). The fractionated adipose tissue was incubated

AC C

EP

TE D

M AN U

SC

RI PT

90

5

ACCEPTED MANUSCRIPT with 0.15% collagenase type 2 solution (Worhtington-Biochem.com, Ref: LS004176, 345

117

u/mg dw) for 60 minutes, centrifuged at 300g for 10 minutes and the supernatants discarded.

118

Cells were washed with complete medium to deactivate the collagenase, filtered through a

119

sterilized tea strainer and a 100 µm mesh filter to remove undissolved tissue debris, and

120

finally counted in a Neubauer counting chamber using crystal violet. The isolated SVF cells

121

were then frozen in medium consisting of fetal bovine serum and 10% dimethyl sulfoxide and

122

kept in the gaseous phase of liquid nitrogen until construct manufacturing.

123

Graft manufacturing

124

Figure 1B shows an overview on the process of graft manufacturing and implantation in the

125

rats. In order to generate constructs, fibrinogen (Fibrinkleber SD TIM3, VNK4J002) and

126

thrombin (SD TIM5 500 IE/ml, VNF4H023, both Baxter AG) were first separately dissolved in

127

sterile water, 40 mM of calcium chloride dihydrate (Fluka, now Sigma-Aldrich Ref. 21102)

128

added to the thrombin and both solutions frozen at -20°C for later use. 40 hypertrophic

129

cartilage pellets were resuspended in 600 µL of fibrin gel (1:2 thrombin:fibrinogen). For the

130

devitalized group with SVF cells, 12x106 SVF cells were added to each construct. The pellets

131

were inserted in the cylindrical chamber (r=5mm, h=12mm) generated within a cylinder

132

(r=7mm, h=12 mm) of processed bovine cancellous bone (Tutobone©, Tutogen Medical,

133

Neunkirchen, Germany). The whole construct was then wrapped in a semipermeable

134

anorganic-based silicone membrane (Biobrane©, UDL Laboratories Inc., Rockford, IL, USA)

135

to prevent ingrowth of host cells from the surrounding tissues [14] . An 18G peripheral

136

venous catheter with needle (B. Braun Melsungen AG, Melsungen, Germany) was finally

137

used to pierce through the packed construct to allow for an easy introduction of the vascular

138

bundle into the construct [6].

139

Implantation

140

For in vivo implantation, 18 young athymic nude rats (RH-Foxn1rnu, Envigo, Venray, The

141

Netherlands) were operated with permission of the Federal Veterinary Office (permit BS

142

2598). Animals were held in groups of three with unrestricted access to food and water.

AC C

EP

TE D

M AN U

SC

RI PT

116

6

ACCEPTED MANUSCRIPT Before surgery, 0.05 mg/kg buprenorphine (Temgesic©, Reckitt Benckiser, Wallisellen,

144

Switzerland) was injected subcutaneously for analgesia. Then inhalation anesthesia was

145

induced and maintained with 1.5-3 % Isofluran (Baxter, Volketswil, Switzerland) in 1 l/min

146

oxygen. The rats were placed in a supine position on an electrically heated mat, the

147

operation area disinfected and a sterile operating field established. An incision of

148

approximately 15 mm was made along the left groin and the superficial inferior epigastric

149

artery and vein dissected free with help of an operation microscope, making sure to leave a

150

small amount of connective tissue in place around the bundle. After distal ligation and

151

transection of the bundle, the ligature thread was used to gently pull the vascular bundle

152

through the prepared construct, using the peripheral venous catheter as a guide and

153

removing it in the process. A subcutaneous pouch was then created by blunt dissection and

154

the pedicled construct inserted carefully. The wound was then closed by layered subdermal

155

and intradermal suturing with absorbable material (5-0 Monocryl, Ethicon, US) and the

156

animal returned to the cage, closely monitored and treated with analgesia according to the

157

veterinary study guidelines.

158

Construct harvest

159

After 12 weeks, deep anesthesia of the animals was induced by intraperitoneal injection of

160

mg/kg ketamine (Ketasol© 100, Dr. E. Graeub AG, Switzerland) and 10 mg/kg xylazine

161

(Xylasol©, Dr. E. Graeub AG). Additionally an inhalation anesthesia with 1.5.-3 % Isoflurane

162

in 1 l/min oxygen was used. The animals were again placed in a supine position and through

163

a median laparotomy the abdominal aorta was exposed and dissected free from the inferior

164

vena cava. The aorta was then cannulated with a 23G butterfly cannula (BD, Heidelberg,

165

Germany) and a heparinized saline solution (100 IU/ml, 0.9 % NaCl, B. Braun AG,

166

Switzerland). Based on a previously developed protocol allowing for analysis of native bone

167

and vasculature by microtomography (µCT), as well as histology [15], we then injected a

168

dark-blue contrasting agent (5% w/v Gelatine-Gold, Carl Roth GmbH, Karlsruhe, Germany;

169

50% v/v Indian Ink, Lefranc & Bourgeois, Le Mans, France; 4% w/v D-Mannitol, Carl Roth

170

GmbH; distilled water) until the skin of the lower extremities turned black, followed by 7

AC C

EP

TE D

M AN U

SC

RI PT

143

ACCEPTED MANUSCRIPT immediate euthanasia of the rat by exsanguination. The cadavers were refrigerated at 4°C

172

overnight to allow for polymerization of the contrast agent-containing gelatin. The constructs

173

were then cautiously dissected free from the surrounding tissue, the state of the vascular

174

bundle macroscopically assessed and the constructs including the bundle transferred to a

175

solution of 1% PFA in distilled water for 24 hours.

176

Histology

177

All histological samples were embedded in paraffin using a TES Valida embedding station

178

and sectioned using a Microm HM 340E Microtome.

179

To check for successful differentiation into hypertrophic cartilage of the expanded hBMSCs,

180

6 randomly picked pellets were processed histologically and stained with Safranin-O (Sigma

181

Ref. 84120), fast green FCF (Sigma Ref. F-7252) and immunohistochemically stained with

182

collagen II (Abcam ab49945) and collagen X (Abcam ab185430) antibodies. For

183

immunological stainings, a goat anti-mouse IgG/ Biotin antibody (Dako E 0433) was used as

184

a secondary antibody and the staining developed with the Vectastain ABC / AP Kit (Linaris

185

AK-5000) and Vector Red AP Substrate kit 1 + Levamisole (Linaris SK-5100) and then

186

counterstained with hematoxylin.

187

Explanted constructs, after assessment by microtomography (see below), were sectioned

188

into a proximal and distal half, processed, embedded in paraffin and sections cut with 5-10

189

µm slice thickness. To achieve a close histological representation of the constructs, slides

190

were made on three levels per half construct to cover 290 µm per level and 670 µm in total.

191

Hematoxylin-eosin (HE Papanicolaou J.T.BAKER, Eosin 2% + Erythrosin B 1%; from

192

Biosystems: Hematoxylin 38-7025-00/38-73, Eosin 84-0012-00 Erythrosin 84-0013-00) and

193

Masson’s trichrome staining (Réactifs RAL, Martillac, France) were performed to assess

194

tissue morphology, bone formation and maturation. ALU-sequence staining (Zytovision

195

GmbH, Bremerhaven, Germany) for cells of human origin and immunohistochemistry for

196

human CD34+ endothelial cells (Millipore CBL 496 from DAKO) were performed to

197

investigate their presence and contribution in SVF-seeded constructs [6].

AC C

EP

TE D

M AN U

SC

RI PT

171

8

ACCEPTED MANUSCRIPT Image acquisition

199

Single images for illustration purposes were taken with Olympus BX61, BX63 and IX83

200

microscopes. Serial pictures for quantification were taken with a Nikon up right Ni, connected

201

to a Prior robot and with 4x and 10x objectives.

202

Image quantification

203

Images were analyzed with Olympus cellSens Dimension 1.15 (Build 14760) and Nikon NIS-

204

Elements AR 4.51 (build 1143) software. For vessel quantification, structures were counted

205

on pictures taken with 10x magnification. Vessels were defined by endothelial cell lining, a

206

well-defined lumen and the presence of either red blood cells or contrasting agent (Indian

207

ink) inside the lumen. Counting was done by two independent observers in blinded samples.

208

Bony regions were identified by the presence of densely organized, homogenous, cell-poor

209

areas with or without bone marrow. These regions, total pellet area and the inside area of the

210

Tutobone© cylinder were selected manually and the area provided by the imaging software.

211

Percentages of remodeled area to total pellet area and of pellet area to total construct area

212

were calculated to account for the distribution of the pellets in the constructs.

213

Microtomography

214

For microtomography, constructs were placed in a 10-ml syringe barrel containing 1 ml of

215

HistoGel (HG-4000-012, Thermo Fisher Scientific, 168 Third Avenue, Waltham, MA USA

216

02451) to conserve and stabilize the sample during transport and scanning. All samples were

217

scanned twice: first without any contrast agent (for imaging of mineralization/bone) and a

218

second time after decalcification with a 15% w/v EDTA (Sigma-Aldrich) solution in distilled

219

water at 37°C for one week and consecutive radio-de nse contrasting of the vessels. The

220

latter was performed with 5% w/v phosphotungstic acid (PTA, Sigma-Aldrich) in distilled

221

water at room temperature for 24 hours, to image the vascular bed, as previously described

222

[15]. This sequence allowed us to first obtain native scans of the radio-dense pre-bone and

223

bone material without any signal overlap that a radio-dense vessel contrasting agent would

224

produce if already applied at the time of explantation. All samples were scanned at 360°

AC C

EP

TE D

M AN U

SC

RI PT

198

9

ACCEPTED MANUSCRIPT using a tungsten x-ray source with a voltage of 70kV and a current of 260 µA (Nanotom M,

226

GE, USA). To reduce beam hardening artifacts, a 0.5 mm aluminum filter was applied.

227

Volumes were reconstructed with 8.5-12 µm voxel size using the manufacturer’s software

228

and analyzed with VGStudio Max version 2.2 (Volume Graphics, Heidelberg, Germany).

229

Bone/ mineralization assessment

230

Total mineralization of the implanted construct was assessed by creating a cylindrical region

231

of interest (ROI) of the Tutobone©’s inside and using VGStudio Max Volume analysis tool.

232

Gray values thresholds were chosen with help of the histogram and visual control to

233

eliminate all non-dense material and background signal. Voxels of the selected gray values

234

were counted by the software and corresponding volumes automatically calculated, giving

235

the total mineralized volume of the constructs. Absolute bone volume was quantified by

236

matching the µCT data on mineralized tissue with histologically-proven bone tissue. The

237

number and volume of all pellets was counted and measured manually by creating ROIs for

238

each pellet. The morphology of the pellets was assessed for the presence of bone-like

239

structures like vascular ingrowth perforation of the mineralized shell, trabeculae inside the

240

pellets, fused pellets and round shaped osteoid deposits, and compared with corresponding

241

histological sections. Number and volumes of pellets with signs of remodeling into bone were

242

measured.

243

Vascularization assessment

244

Vascularization was assessed in reconstructed volumes of PTA-stained constructs with a

245

focus on characterizing the AV bundle and its branching vessels. The AV bundle and

246

branching vessels were identified by radiodensity, and the diameters and numbers measured

247

or counted on transversal planes. Measurement of the largest vessel was performed on short

248

and long axes on three levels: the proximal entry point of the bundle into the Tutobone©, in

249

the middle and at the distal end of the Tutobone© just over the drill hole. Furthermore, the

250

shortest and longest distance between remodeled pellets and the central vessel were

251

measured. By superimposing µCT-datasets taken with and without phosphotungstic acid, it

AC C

EP

TE D

M AN U

SC

RI PT

225

10

ACCEPTED MANUSCRIPT was possible to subtract the signal of the radio-dense pellets from the 3D-reconstruction to

253

visualize selectively the PTA stained structures.

254

Statistical analysis

255

All statistical analyses were performed using GraphPad Prism software (GraphPad Software,

256

Inc., California, USA). Kruscal-Wallis test, Dunn’s multiple comparisons test and Spearman’s

257

rank correlation coefficient were performed considering a p-value of <0.05 significant.

258

Results are reported as mean ± standard deviation unless otherwise indicated.

259

Results

260

Successful in vitro cartilaginous differentiation and hypertrophy of the hBMSC micromass

261

pellets was demonstrated by high glycosaminoglycan content (red in Saf-O staining) and

262

positive immunohistochemical stainings for collagen II and X (Fig. 1A). Devitalized pellets,

263

embedded in fibrin gel with or without SVF cells, were then inserted in the cancellous bone

264

cylinder together with an AV bundle and implanted in nude rats (Fig. 1B). Vital pellets were

265

also used as positive control, to validate the biological functionality of the engineered

266

hypertrophic cartilage through its capacity to undergo endochondral ossification [13]. During

267

and after graft insertion, no bleeding problems or hematoma formation were observed in our

268

case series, likely since the bundle was first ligated and then pulled through into the

269

construct. All 18 implanted constructs (6 constructs containing vital pellets without SVF cells,

270

6 constructs containing devitalized pellets without SVF cells and 6 constructs containing

271

devitalized pellets with SVF cells) were explanted after 12 weeks in vivo. A visual inspection

272

revealed signs of thrombosis in AV bundles of four constructs (2 vital, 1 devitalized and 1

273

devitalized + SVF) leading to limited engraftment with fibrotic tissue and marked

274

inflammatory response. Due to the surgery-related technical failure, these four constructs

275

were excluded from the study and not further analyzed.

AC C

EP

TE D

M AN U

SC

RI PT

252

11

ACCEPTED MANUSCRIPT Bone assessment and quantification

277

Bone tissue was formed in 3 out of the 4 constructs containing vital pellets and in 3 out of the

278

5 constructs containing devitalized pellets, both in the presence and in absence of SVF cells

279

(Fig. 2). When bone was formed, H&E-stained sections demonstrated a dense collagen

280

matrix and fused pellets, indicating maturation into trabecular bone, including the formation of

281

bone marrow elements (Fig. 2A,B,C). Quantification of the percentage of the average pellet

282

area remodeled into bone tissue by histological morphometry revealed a higher amount of

283

bone formation in constructs containing vital hypertrophic pellets (17.2 ± 9.7 %; Fig. 2D) than

284

in constructs containing devitalized pellets (2.8 ± 4.3 %) or devitalized pellets with SVF cells

285

(1.5 ± 2.4 %). However, given the high variability of bone formation within the experimental

286

groups (e.g. varying from 0 - 43 % in the vital pellets’ group) and the limited number of

287

available replicates, the difference in terms of bone formation was not statistically significant

288

among the groups. All three conditions exhibited the same density of infiltrating CD68+

289

macrophages (Supp. Fig. 1). M2-like macrophages, defined as CD68+/CD163+, were

290

significantly more frequent in the vital constructs, than in constructs with devitalized pellets

291

(respectively 40% vs. 6% of all macrophages). The addition of SVF cells to devitalized pellets

292

was able to restore the percentage of invading macrophages to the levels of living pellets

293

(29%). Staining for human ALU-sequences evidenced the presence of human cells inside

294

and surrounding the pellets (Supp. Fig. 2) after 12 weeks in vivo, documenting the survival of

295

at least a fraction of the implanted SVF cells, with a possible contribution in cartilage matrix

296

remodeling and bone formation.

297

Total mineralized volume was assessed by µCT. Thresholding for densities corresponding to

298

mineralization, eliminating the signal due to soft tissue and background, was possible on the

299

gray value histogram by visually controlling the selected voxels (Supp. Fig. 3). Qualitative

300

imaging confirmed fusion of some pellets through mineralized structures (Fig. 3A).

301

Quantitative analysis indicated that mineralization was not different among the three groups

302

(12.8 ± 3.2 mm3, 11.8 ± 2.4 mm3 and 14.5 ± 3.8 mm3; Fig. 3B). Bone volume was then

303

calculated by adjusting the imaging threshold such that the regions of bone in µCT cross-

AC C

EP

TE D

M AN U

SC

RI PT

276

12

ACCEPTED MANUSCRIPT sections would correspond to those detected histologically in corresponding specimens (Fig.

305

3C,D; dotted yellow line). Volumetric reconstruction indicated that bone tissue formation was

306

not homogenous throughout the implanted constructs, but only took place at a radial distance

307

of 0.2-2.2 mm from the central vessel (on a maximal potential radius of 5 mm, Fig. 3E). In all

308

experimental groups, bone formation was rather variable along the longitudinal axis, with a

309

higher tendency to form bone in the proximal half of the construct (64% proximal half vs. 21%

310

in the distal half). The average bone volume was higher in the vital pellets group (5.1 ± 2.2

311

mm3), as compared to the devitalized pellets without (1.2 ± 0.7 mm3) or with SVF cells (1.4 ±

312

0.7 mm3) (Fig. 3F). However, differences were not statistically significant. The average pellet

313

volume transformed into bone was also evaluated by µCT and was consistent with

314

histomorphometry data, though again with no significant difference among the groups (Fig.

315

3G).

316

Vascularization analysis

317

We then assessed the vascularization inside the bone grafts after 12 weeks in vivo

318

implantation time. Phosphotungstic acid-contrast enhanced µCT showed that in all constructs

319

included for analysis the graft was fully vascularized, with small and intermediate sized

320

vessels arising from the central AV bundle, spreading towards the surrounding devitalized

321

bone (Fig. 4A). Histological analysis confirmed the high level of vascularization in every

322

experimental condition, as documented by the presence of vascular lumina filled by Indian

323

ink (black arrows in Fig. 4B), around the vascular bundle and in vicinity of the pellets.

324

Although a large part of the vessels was found in the fibrin compartment, ingrowth of

325

vasculature was also observed in the regions of de novo bone formation, corresponding to

326

the implanted hypertrophic cartilage pellets (Supp. Fig. 4). The radial distribution of vessels

327

showed a statistically significant difference in all groups between the central (0-2.5 mm) and

328

the peripheral (2.5-5 mm) half of the construct. In particular, the central part close to the main

329

bundle was more extensively vascularized, with an up to 5-fold higher vessel density than the

330

peripheral part (Fig. 4C).

AC C

EP

TE D

M AN U

SC

RI PT

304

13

ACCEPTED MANUSCRIPT Quantification of vessels’ diameter showed no significant difference between the

332

experimental groups, although the mean diameter of central vessels in the constructs

333

containing living pellets was in average twice larger (468 ± 111 µm) than in the devitalized

334

pellet groups (190 ± 70 µm and 243 ± 68 µm respectively, with and without SVF cells) (Fig.

335

4D). The number of vessels branching from the central bundle also did not statistically differ

336

among groups (Fig. 4E).

337

Correlation of vascularization and bone formation

338

No statistically significant difference could be evidenced among the groups for the diameter

339

of central vessel, density of vascular network, or amount of bone formation, likely due to the

340

rather limited number of implanted specimens per experimental group (i.e., 4 or 5). Despite

341

this limitation, a correlation could be established between vascularization and bone

342

formation. The size of the central vessel diameter was positively linear correlated to the

343

volume of formed bone (r2=0.74; p=0.0002) (Fig. 5A).

344

correlation (r2=0.77; p<0.0001) could be demonstrated between the vessel density and the

345

newly formed bone volume (Fig. 5B).

M AN U

SC

RI PT

331

TE D

346

An even stronger positive linear

Discussion

348

This study demonstrates that by combining engineered hypertrophic cartilage with an AV

349

bundle within a form-stable outer shell, it is possible to prefabricate a vascularized bone graft

350

with an inherent germ of osteoinduction and vascularization. The vessels sprouting from the

351

AV bundle were strongly correlated to the bone forming capacity, underlined by the fact that

352

in the peripheral areas where vascularization decreased, bone formation did not occur.

353

The implementation of an axial AV bundle led to a thorough vascularization of the entire

354

critically sized construct, which is clinically relevant for large bone grafts [16–18], and

355

enabled colonization of the chamber by host cells, which could otherwise not reach it due to

356

the silicon wrap. The prefabricated bone graft offers the feature of a vascular pedicle, which

357

upon orthotopic transfer can be anastomosed to the recipient vessels, and thus could 14

AC C

EP

347

ACCEPTED MANUSCRIPT guarantee immediate vascularization. A two-stage procedure involving a prefabrication step

359

is typical for large size reconstructions. From a surgical standpoint, translation of the

360

developed approach in a clinical setting, e.g. for the replacement of the scaphoid bone,

361

would require to identify a suitable region for ectopic prefabrication, which would not

362

aesthetically compromise the patient and where donor vessels could be sacrificed without

363

functional impairment. In this regard, microsurgical generation of an AV bundle from the

364

superficial epigastric vessels of the rat emulates a procedure which is feasible and is actually

365

typically applied in clinical settings [19,20]. The generation of an AV bundle, despite being a

366

long a complex surgical procedure, would allow to reduce donor site morbidity for the

367

otherwise required free tissue transfer (e.g., fibula flap). Moreover, the occurrence of

368

complications could be diminished by adequate patient selection and thorough perioperative

369

care [21].

370

In the clinical implementation of large tissue engineered bone grafts, the inability to rapidly

371

perfuse tissues has been a major limiting factor [22]. In our model, the high vessel density in

372

the area surrounding the central AV bundle, together with the high number of branches

373

arising from the main trunk, indicate a robust vascularization process. Not only did

374

vascularization occur in the loose fibrin areas, which could be interpreted as inflammatory

375

granulation tissue, but also in the regions of the implanted pellets, likely through release of

376

angiogenic factors embedded within hypertrophic cartilage and known to critically regulate its

377

bone remodeling [23]. In mature individuals, luminal vessel sprouting is believed to be the

378

major process during angiogenesis and in AV bundles it typically occurs from the central vein

379

[4,24]. The process underlying angiogenic luminal sprouting begins with promoting factors

380

urging detachment of pericytes and endothelial cells [25]. The initial hypoxic and

381

inflammatory environment within our constructs, as well as VEGF present in the hypertrophic

382

cartilage, are strong promoters for the detached, reactive endothelial cells. Proteases further

383

liberate proangiogenic factors from the ECM such as VEGF and FGF and endothelial

384

sprouting is initiated by migrating and proliferating endothelial cells [26]. In our study, we

385

could observe a limit in extent of vascularization at a radial distance of 2.5 mm from the AV

AC C

EP

TE D

M AN U

SC

RI PT

358

15

ACCEPTED MANUSCRIPT bundle. The vessel density in the peripheral area beyond this point was significantly lower

387

than in the central part. A possible explanation for this decrease in vascularization could lie in

388

the nature of tip cells. These cells are equipped with filopodia which sense environmental

389

cues to guide them, such as ephrins and semaphorins [27]. In the absence of these factors in

390

the peripheral zone of our construct, tip cells may have become quiescent and stabilized.

391

Once stabilized, endothelial cells are covered by pericytes and a basal membrane is

392

deposited, marking the end of the sprouting process.

393

Besides a thorough vascularization, we generated a germ of bone formation within our

394

constructs by using engineered hypertrophic cartilage. It has been shown that cartilaginous

395

tissues engineered from hBMSC can act as a template for bone formation derived from the

396

synergy of donor and host cells in both ectopic and orthotopic animal models [7,10].

397

Moreover, freeze and thaw (F&T) devitalized hypertrophic cartilage pellets were so far shown

398

to undergo endochondral ossification upon in vivo implantation if reactivated by addition of

399

SVF cells [12]. In order for devitalized cartilaginous tissue to undergo ossification in vivo

400

without the addition of external cells, it was reported that a devitalization method, able to

401

maintain important biological cues in the ECM, is required [28]. In our constructs,

402

hypertrophic cartilage pellets were devitalized by F&T and thus resulting in a compromised

403

composition of the ECM [28]. It is noteworthy that despite that, they were still able to form

404

bone, in a quantity, which was not statistically significant inferior as compared to the vital

405

ones. A possible explanation for this finding lies in the privileged environment created by the

406

AV bundle. The surrounding tissue transplanted together with vessels, where capillaries and

407

fibroblasts proliferate, may provide osteoclasts and osteoblasts which resorb avital bone and

408

at the same time stimulate bone formation and remodeling.

409

Tissue remodeling is often associated with specifically increased immunological cell

410

subtypes, such as alternatively activated (pro-regenerative/M2-like) macrophages [29].

411

However, we could not observe any correlation between M2 polarization of invading

412

macrophages and new bone formation. Transplanted SVF cells were able to induce the

413

presence of M2 macrophages, as previously observed in a similar model [6], but the

AC C

EP

TE D

M AN U

SC

RI PT

386

16

ACCEPTED MANUSCRIPT phenotype change did not lead to increased bone formation. This emphasizes the fact that to

415

enhance a regenerative process in bone tissue engineering a precise understanding of the

416

role and dynamics of immunological cell populations is necessary [23,30] and the specific

417

role of macrophagic populations, with phenotypes hardly categorized in artificial classes, is

418

still elusive.

419

In our generated bone graft, the functional AV bundle was sufficient to lead to bony

420

remodeling of the implanted pellets and addition of SVF cells could not enhance the bone

421

quantity. Bone forming capacity tightly correlated with vascularization of the construct (i.e.

422

central vessel diameter and vessel density). This finding is consistent with the notion that

423

increased vascularization of a tissue can increase its regenerative capacity, since

424

vascularization is a “critical obstacle in engineering metabolically active tissues” [31]. In our

425

model, vascularization could have enhanced regenerative capacity by contributing in bony

426

transformation of the cartilaginous templates. Two mechanisms are likely responsible for this;

427

first, attracted by VEGF within the ECM, the invading vessels could provide chondroclasts for

428

the degradation of cartilage as well as osteoblasts and osteoclasts, responsible for the

429

deposition and remodeling of bone matrix [32]. Thus, reduced bone formation in less

430

vascularized areas could be due to a limited infiltration of host cells. Secondly, since oxygen

431

tension plays an important role in bone fracture healing and bone remodeling [33],

432

vascularization could efficiently supply oxygen required for the activity of many enzymes

433

enhancing bone healing, such as cycloxygenase-2, and reduce hypoxia-driven osteoclast

434

formation and bone resorption [34,35].

435

The here gained knowledge could inspire possible improvements to extend bone formation

436

beyond the central zone. The observed decrease of vascularization at the periphery, and

437

associated negligible bone formation, could be explained by the type and proportion of

438

materials used in the model [36]. For example, the high porosity and pore structure of the

439

hydroxyapatite scaffold used in Ismail et al.’s work was shown to be permissive for guided

440

vascularization [6]. In fact, with the same axial AV bundle technique, bone formation was

441

observed up to 4.5 mm. Thus, to increase vessel density beyond 2.5 mm from the central AV

AC C

EP

TE D

M AN U

SC

RI PT

414

17

ACCEPTED MANUSCRIPT bundle, one could either increase the ratio of scaffolding material, i.e. pellet volume

443

(decreasing the amount of fibrin), or stimulate vessel formation by a growth factor-enhanced

444

environment. Increasing the cartilage pellet number, which in our study was limited to 40 per

445

construct, could enhance both vascular guidance and bone forming foci. It was demonstrated

446

that hBMSC can undergo endochondral ossification when seeded on type I collagen sponges

447

of variable sizes [13,37]. Seeding hBMSC on collagen sponges would allow generating larger

448

pellets and thus facilitating production of large quantities of matrix, which would be then

449

devitalized and possibly used as off-the-shelf filler material.

450

A yet alternative strategy to increase efficiency and uniformity of bone formation in the model

451

could rely on the delivery of hBMSC instead of SVF cells, given the known superior

452

osteogenic capacity [38]. Indeed, hBMSC intraoperatively gained after concentration of bone

453

marrow aspirates have already been tested for the treatment of bone defects [39]. Although

454

such comparison requires a dedicated study, it could be argued that cell populations from

455

concentrated marrow contain a rather limited fraction of vasculogenic cells and a largely

456

variable amount of osteoprogenitors, with unpredictable engraftment capacity, especially in

457

hostile environments [39].

458

TE D

M AN U

SC

RI PT

442

Conclusion

460

In this study we showed that an axially implemented central AV bundle plays a key role in the

461

bone formation capacity of engineered hypertrophic cartilage. In challenging clinical

462

scenarios, e.g. in the maxillo-facial region, the complexity of the defects often limit

463

conventional reconstruction with autologous tissue. For these cases, our proposed approach

464

could provide an alternative bone graft surrogate as an ectopically prefabricated, pedicled,

465

large bone graft with a form-stable, outer shell and an inner core containing a germ for bone

466

and vascularization. While this inner core represents the biological backbone inducing

467

functionality of the graft, the outer shell could be adapted to the defect by 3D CAD-CAM

468

techniques and thus fit the defect site in size and shape.

AC C

EP

459

469 18

ACCEPTED MANUSCRIPT Data availability

471

The raw data required to reproduce these findings are available to download from [INSERT

472

PERMANENT WEB LINK(s)]. The processed data required to reproduce these findings are

473

available to download from [INSERT PERMANENT WEB LINK(s)].

474

References

475 476 477 478

[1]

S. Akita, N. Tamai, A. Myoui, M. Nishikawa, T. Kaito, K. Takaoka, H. Yoshikawa, Capillary Vessel Network Integration by Inserting a Vascular Pedicle Enhances Bone Formation in Tissue-Engineered Bone Using Interconnected Porous Hydroxyapatite Ceramics, Tissue Eng. 10 (2004) 789–795. doi:10.1089/1076327041348338.

479 480 481 482 483

[2]

A. Haumer, T. Ismail, A. Lunger, R. Osinga, A. Scherberich, D.J. Schaefer, I. Martin, From Autologous Flaps to Engineered Vascularized Grafts for Bone Regeneration, in: W. Holnthoner, A. Banfi, J. Kirkpatrick, H. Redl (Eds.), Vasc. Tissue Eng. Regen. Med., Springer International Publishing, Cham, 2017: pp. 1–34. doi:10.1007/978-3319-21056-8_16-1.

484 485 486

[3]

O.O. Erol, M. Sira, New capillary bed formation with a surgically constructed arteriovenous fistula., Plast. Reconstr. Surg. 66 (1980) 109–15. doi:10.1097/00006534-198208000-00093.

487 488 489 490 491

[4]

Y. Tanaka, K.-C. Sung, A. Tsutsumi, S. Ohba, K. Ueda, W. a Morrison, Tissue engineering skin flaps: which vascular carrier, arteriovenous shunt loop or arteriovenous bundle, has more potential for angiogenesis and tissue generation?, Plast. Reconstr. Surg. 112 (2003) 1636–44. doi:10.1097/01.PRS.0000086140.49022.AB.

492 493 494 495 496

[5]

V. Planat-Benard, J.-S. Silvestre, B. Cousin, M. André, M. Nibbelink, R. Tamarat, M. Clergue, C. Manneville, C. Saillan-Barreau, M. Duriez, A. Tedgui, B. Levy, L. Pénicaud, L. Casteilla, Plasticity of human adipose lineage cells toward endothelial cells: physiological and therapeutic perspectives., Circulation. 109 (2004) 656–663. doi:10.1161/01.CIR.0000114522.38265.61.

497 498 499 500 501

[6]

T. Ismail, R. Osinga, A.J. Todorov, A. Haumer, L.A. Tchang, C. Epple, N. Allafi, N. Menzi, R.D. Largo, A. Kaempfen, I. Martin, D.J. Schaefer, A. Scherberich, Engineered, axially-vascularized osteogenic grafts from human adipose-derived cells to treat avascular necrosis of bone in a rat model., Acta Biomater. (2017). doi:10.1016/j.actbio.2017.09.003.

502 503 504 505

[7]

506 507 508 509

[8]

N. Harada, Y. Watanabe, K. Sato, S. Abe, K. Yamanaka, Y. Sakai, T. Kaneko, T. Matsushita, Bone regeneration in a massive rat femur defect through endochondral ossification achieved with chondrogenically differentiated MSCs in a degradable scaffold, Biomaterials. 35 (2014) 7800–7810. doi:10.1016/j.biomaterials.2014.05.052.

510 511 512

[9]

L.T. Kuhn, Y. Liu, N.L. Boyd, J.E. Dennis, X. Jiang, X. Xin, L.F. Charles, L. Wang, H.L. Aguila, D.W. Rowe, A.C. Lichtler, A.J. Goldberg, Developmental-like bone regeneration by human embryonic stem cell-derived mesenchymal cells., Tissue Eng.

AC C

EP

TE D

M AN U

SC

RI PT

470

C.S. Bahney, D.P. Hu, A.J. Taylor, F. Ferro, H.M. Britz, B. Hallgrimsson, B. Johnstone, T. Miclau, R.S. Marcucio, Stem cell-derived endochondral cartilage stimulates bone healing by tissue transformation, J. Bone Miner. Res. 29 (2014) 1269–1282. doi:10.1002/jbmr.2148.

19

ACCEPTED MANUSCRIPT Part A. 20 (2014) 365–77. doi:10.1089/ten.TEA.2013.0321.

513 [10]

J. van der Stok, M.K.E. Koolen, H. Jahr, N. Kops, J.H. Waarsing, H. Weinans, O.P. van der Jagt, Chondrogenically differentiated mesenchymal stromal cell pellets stimulate endochondral bone regeneration in critical-sized bone defects, Eur. Cells Mater. 27 (2014) 137–148.

518 519 520

[11]

W. Yang, F. Yang, Y. Wang, S.K. Both, J.A. Jansen, In vivo bone generation via the endochondral pathway on three-dimensional electrospun fibers, Acta Biomater. 9 (2013) 4505–4512. doi:10.1016/j.actbio.2012.10.003.

521 522 523 524

[12]

A. Todorov, M. Kreutz, A. Haumer, C. Scotti, A. Barbero, P.E. Bourgine, A. Scherberich, C. Jaquiery, I. Martin, Fat-Derived Stromal Vascular Fraction Cells Enhance the Bone-Forming Capacity of Devitalized Engineered Hypertrophic Cartilage Matrix., Stem Cells Transl. Med. (2016). doi:10.5966/sctm.2016-0006.

525 526 527 528 529

[13]

C. Scotti, B. Tonnarelli, A. Papadimitropoulos, A. Scherberich, S. Schaeren, A. Schauerte, J. Lopez-Rios, R. Zeller, A. Barbero, I. Martin, Recapitulation of endochondral bone formation using human adult mesenchymal stem cells as a paradigm for developmental engineering., Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 7251–7256. doi:10.1073/pnas.1000302107.

530 531 532

[14]

M. Wiedmann-Al-Ahmad, R. Gutwald, N.C. Gellrich, U. Hübner, R. Schmelzeisen, Search for ideal biomaterials to cultivate human osteoblast-like cells for reconstructive surgery, J. Mater. Sci. Mater. Med. 16 (2005) 57–66. doi:10.1007/s10856-005-6447-z.

533 534 535 536 537

[15]

S. Sutter, A. Todorov, T. Ismail, A. Haumer, I. Fulco, G. Schulz, A. Scherberich, A. Kaempfen, I. Martin, D.J. Schaefer, Contrast-enhanced microtomographic characterisation of vessels in native bone and engineered vascularised grafts using ink-gelatin perfusion and phosphotungstic acid, Contrast Media Mol. Imaging. 2017 (2017). doi:10.1155/2017/4035160.

538 539 540 541

[16]

M. Deschepper, M. Manassero, K. Oudina, J. Paquet, L.E. Monfoulet, M. Bensidhoum, D. Logeart-Avramoglou, H. Petite, Proangiogenic and prosurvival functions of glucose in human mesenchymal stem cells upon transplantation, Stem Cells. 31 (2013) 526– 535. doi:10.1002/stem.1299.

542 543 544

[17]

a W. Orr, C. a Elzie, D.F. Kucik, J.E. Murphy-Ullrich, Thrombospondin signaling through the calreticulin/LDL receptor-related protein co-complex stimulates random and directed cell migration., J. Cell Sci. 116 (2003) 2917–27. doi:10.1242/jcs.00600.

545 546 547

[18]

M.I. Santos, R.L. Reis, Vascularization in bone tissue engineering: Physiology, current strategies, major hurdles and future challenges, Macromol. Biosci. 10 (2010) 12–27. doi:10.1002/mabi.200900107.

548 549 550

[19]

551 552 553

[20]

H.D. Wang, J.C. Alonso-Escalante, B.H. Cho, R.A. DeJesus, Versatility of Free Cutaneous Flaps for Upper Extremity Soft Tissue Reconstruction., J. Hand Microsurg. 9 (2017) 58–66. doi:10.1055/s-0037-1603918.

554 555 556 557

[21]

H.J. Klein, N. Fuchs, T. Mehra, R. Schweizer, T. Giesen, M. Calcagni, G.F. Huber, P. Giovanoli, J.A. Plock, Extending the limits of reconstructive microsurgery in elderly patients, J. Plast. Reconstr. Aesthetic Surg. 69 (2016) 1017–1023. doi:10.1016/j.bjps.2016.01.020.

AC C

EP

TE D

M AN U

SC

RI PT

514 515 516 517

W.M. Rozen, D. Chubb, D. Grinsell, M.W. Ashton, The variability of the Superficial Inferior Epigastric Artery (SIEA) and its angiosome: A clinical anatomical study, Microsurgery. 30 (2010) 386–391. doi:10.1002/micr.20750.

20

ACCEPTED MANUSCRIPT [22]

J.M. Kanczler, R.O.C. Oreffo, Osteogenesis and angiogenesis: The potential for engineering bone, Eur. Cells Mater. 15 (2008) 100–114. doi:10.22203/eCM.v015a08.

560 561 562

[23]

A. Haumer, P.E. Bourgine, P. Occhetta, G. Born, R. Tasso, I. Martin, Delivery of cellular factors to regulate bone healing., Adv. Drug Deliv. Rev. (2018). doi:10.1016/j.addr.2018.01.010.

563 564

[24]

L. Beck, P. a D’Amore, Vascular development: cellular and molecular regulation., FASEB J. (1997) 365–373. doi:10.1002/term.394.

565 566

[25]

P. Carmeliet, R.K. Jain, Molecular mechanisms and clinical applications of angiogenesis, Nature. 473 (2011) 298–307. doi:10.1038/nature10144.

567 568

[26]

N. Ferrara, VEGF-A: A critical regulator of blood vessel growth, Eur. Cytokine Netw. 20 (2009) 158–163. doi:10.1684/ecn.2009.0170.

569 570

[27]

G. Neufeld, O. Kessler, The semaphorins: Versatile regulators of tumour progression and tumour angiogenesis, Nat. Rev. Cancer. 8 (2008) 632–645. doi:10.1038/nrc2404.

571 572 573

[28]

P.E. Bourgine, C. Scotti, S. Pigeot, L. a Tchang, A. Todorov, I. Martin, Osteoinductivity of engineered cartilaginous templates devitalized by inducible apoptosis., Proc. Natl. Acad. Sci. U. S. A. (2014) 1–6. doi:10.1073/pnas.1411975111.

574 575 576

[29]

C. Schlundt, T. El Khassawna, A. Serra, A. Dienelt, S. Wendler, H. Schell, N. van Rooijen, A. Radbruch, R. Lucius, S. Hartmann, Macrophages in bone fracture healing: their essential role in endochondral ossification, Bone. (2015).

577 578

[30]

K.L. Spiller, T.J. Koh, Macrophage-based therapeutic strategies in regenerative medicine, Adv. Drug Deliv. Rev. (2017).

579 580

[31]

R.K. Jain, P. Au, J. Tam, D.G. Duda, D. Fukumura, Engineering vascularized tissue, Nat. Biotechnol. 23 (2005) 821–823. doi:10.1038/nbt0705-821.

581 582

[32]

H.M. Kronenberg, Developmental regulation of the growth plate, Nature. 423 (2003) 332–336. doi:10.1038/nature01657.

583 584 585

[33]

C. Lu, N. Saless, X. Wang, A. Sinha, S. Decker, G. Kazakia, H. Hou, B. Williams, H.M. Swartz, T.K. Hunt, T. Miclau, R.S. Marcucio, The role of oxygen during fracture healing, Bone. 52 (2013) 220–229. doi:10.1016/j.bone.2012.09.037.

586 587 588 589

[34]

X. Zhang, E.M. Schwarz, D.A. Young, J. Edward Puzas, R.N. Rosier, R.J. O’Keefe, Cyclooxygenase-2 regulates mesenchymal cell differentiation into the osteoblast lineage and is critically involved in bone repair, J. Clin. Invest. 109 (2002) 1405–1415. doi:10.1172/JCI200215681.

590 591 592

[35]

593 594 595 596

[36]

B. Feng, Z. Jinkang, W. Zhen, L. Jianxi, C. Jiang, L. Jian, M. Guolin, D. Xin, The effect of pore size on tissue ingrowth and neovascularization in porous bioceramics of controlled architecture in vivo., Biomed. Mater. 6 (2011) 015007. doi:10.1088/17486041/6/1/015007.

597 598 599 600

[37]

C. Scotti, E. Piccinini, H. Takizawa, A. Todorov, P. Bourgine, A. Papadimitropoulos, A. Barbero, M.G. Manz, I. Martin, Engineering of a functional bone organ through endochondral ossification., Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 3997–4002. doi:10.1073/pnas.1220108110.

AC C

EP

TE D

M AN U

SC

RI PT

558 559

T.R. Arnett, D.C. Gibbons, J.C. Utting, I.R. Orriss, A. Hoebertz, M. Rosendaal, S. Meghji, Hypoxia is a major stimulator of osteoclast formation and bone resorption, J. Cell. Physiol. 196 (2003) 2–8. doi:10.1002/jcp.10321.

21

ACCEPTED MANUSCRIPT [38]

P. Niemeyer, K. Fechner, S. Milz, W. Richter, N.P. Suedkamp, A.T. Mehlhorn, S. Pearce, P. Kasten, Comparison of mesenchymal stem cells from bone marrow and adipose tissue for bone regeneration in a critical size defect of the sheep tibia and the influence of platelet-rich plasma, Biomaterials. 31 (2010) 3572–3579. doi:10.1016/j.biomaterials.2010.01.085.

606 607 608 609

[39]

P. Hernigou, G. Guerin, Y. Homma, A. Dubory, N. Chevallier, H. Rouard, C.H. Flouzat Lachaniette, History of concentrated or expanded mesenchymal stem cells for hip osteonecrosis: is there a target number for osteonecrosis repair?, Int. Orthop. 42 (2018) 1739–1745. doi:10.1007/s00264-018-4000-1.

RI PT

601 602 603 604 605

610 611

SC

612

Figure Legend

614

Figure 1: Pellet characterization and construct engineering. (A) successful in vitro

615

cartilaginous differentiation and hypertrophy of the hBMSC micromass pellets, demonstrated

616

by positive staining for glycosaminoglycans (Safranin-O, Saf-O) and for collagen types II and

617

X. (B) hBMSC micromass pellets were cultured following a specific, two-staged differentiation

618

protocol, Devitalized pellets (± addition of SVF cells) were inserted in a hollow (Tutobone©)

619

cylinder together with an axially implemented AV bundle. The constructs were then implanted

620

subcutaneously in nude rats for 12 weeks.

621

Figure 2: Bone formation after 12 weeks in vivo. (A) Masson Trichrome and (B&C) H&E

622

stained sections demonstrate a dense collagen matrix, fused pellets and formation of bone

623

marrow elements, indicating the maturation into trabecular bone. (C) Magnification shows

624

dense osteoid matrix containing osteocytes with lacunae (empty arrowheads) and an

625

osteoblast cell lining (black arrows). The asterisk (*) marks the presence of fatty bone

626

marrow. (D) Percentage of pellet area transformed into bone after 12 weeks in vivo. Vital

627

pellets did not generate a significantly higher percentage of bone compared to devitalized

628

and devitalized pellets with SVF cells. Scale bars A & B: 500 µm, C: 100 µm

629

Figure 3: Assessment of mineralization and bone formation after 12 weeks in vivo with µCT.

630

(A) Qualitative analysis shows fusion of pellets and voids in the mineralized matrix. (B)

AC C

EP

TE D

M AN U

613

22

ACCEPTED MANUSCRIPT Quantitative measurements of mineralization shows no difference among the experimental

632

groups. (C, D) Bone volume was quantified by matching the µCT data with histologically

633

proven bone tissue (dotted yellow line). (E) There was no statistical significant difference

634

among the experimental groups in terms of formed bone volume and (F) percentage of the

635

pellet area transformed into bone.

636

Figure 4: Assessment of vascularization after 12 weeks in vivo. (A) Contrast enhanced µCT

637

shows vessels spreading from the central AV bundle to the periphery of the constructs. (B)

638

Histological analysis confirmed the distribution of vascular lumens filled by Indian ink

639

throughout the construct (black arrows). (C) Histological vessel density analysis by

640

calculation of vessel length/per area showed a statistically higher vessel density in the

641

central part of the construct (p=0.022). (D) Mean diameter of the central vessels (calculated

642

by µCT) in the constructs containing living pellets was in average twice as large as in the

643

devitalized or devitalized pellets + SVF groups, however without statistical significant

644

difference among the groups. (E) Also, number of vessels branching from the central bundle

645

did not statistically differ among groups.

646

Figure 5: Correlation between bone formation and vascularization. (A) Direct positive

647

correlation (P=0.0002; r2=0.742) between the bone volume and the proximal vessel

648

diameter. (B) Positive correlation (p< 0.0001; r2=0.77) between the vessel density and bone

649

formation. (C) A higher vessel density could also be evidenced in the radially central half of

650

the construct compared to the periphery (p=0.0016). (D) This is tightly correlated with the

651

pellet area transformed into bone, which was strictly in the central part (p=0.0002)

652

Supplementary Figure Legend

653

Supplementary figure 1: Polarization of invading macrophages. All three conditions

654

exhibited the same density of infiltrating CD68+ macrophages. CD68+/CD163+, M2-like

655

macrophages were significantly more frequent in the vital constructs (Vital vs Devitalized:

656

p=0.0049) representing 40% of all macrophages. In constructs with devitalized pellets, M2-

AC C

EP

TE D

M AN U

SC

RI PT

631

23

ACCEPTED MANUSCRIPT like macrophages were more abundant when SVF cells were added (29%; Devitalized + SVF

658

vs. Devitalized: p=0.04), while only 6% of the invading macrophages were polarized towards

659

an M2 phenotype when only devitalized pellets were implanted.

660

Supplementary figure 2: Staining for human ALU-sequences evidenced the presence of

661

human cells in the condition where SVF cells were implanted together with devitalized pellets

662

after 12 weeks in vivo. Human cells (dark blue-stained cells, black arrows) could be detected

663

both inside the pellets (2A) as well as in the tissue surrounding them (2B). Scale bars A & B:

664

200 µm, C: 100 µm

665

Supplementary figure 3: Thresholding for densities corresponding to mineralization,

666

eliminating the signal due to soft tissue and background, was possible on the gray value

667

histogram by visually controlling the selected voxels.

668

Supplementary figure 4: Ingrowth of vasculature was detected in the loose fibrin tissue

669

(white arrows), as well as in the regions of de novo bone formation, corresponding to the

670

implanted hypertrophic cartilage pellets (black arrows; scale bar 300 µm).

673

674

675

SC

M AN U

TE D EP

672

AC C

671

RI PT

657

676

677

678

679 24

ACCEPTED MANUSCRIPT Figures

681

Figure 1:

683 684 685 686

AC C

682

EP

TE D

M AN U

SC

RI PT

680

687 688 689 690 25

ACCEPTED MANUSCRIPT Figure 2:

AC C

EP

TE D

M AN U

SC

RI PT

691

692 693 694 695

26

ACCEPTED MANUSCRIPT Figure 3:

AC C

EP

TE D

M AN U

SC

RI PT

696

697 27

ACCEPTED MANUSCRIPT Figure 4:

AC C

EP

TE D

M AN U

SC

RI PT

698

699 700 28

ACCEPTED MANUSCRIPT Figure 5:

SC

RI PT

701

702

704

705

Supplementary Figures

707

Supplementary Figure 1:

AC C

EP

TE D

706

M AN U

703

708 709 710 711 712 29

ACCEPTED MANUSCRIPT Supplementary Figure 2:

M AN U

SC

RI PT

713

714 715

717 718

Supplementary Figure 3:

AC C

EP

719

TE D

716

720 721 30

ACCEPTED MANUSCRIPT Supplementary Figure 4:

M AN U

SC

RI PT

722

723

AC C

EP

TE D

724

31