Tissue engineered humanized bone supports humanhematopoiesis in vivo

Boris M. Holzapfel a, b, 1, Dietmar W. Hutmacher a, c, d, *, 1, Bianca Nowlan e,Valerie Barbier e, Laure Thibaudeau a, Christina Theodoropoulos a, John D. Hooper f,Daniela Loessner a, Judith A. Clements f, Pamela J. Russell f, g, Allison R. Pettit h,Ingrid G. Winkler e, Jean-Pierre Levesque e, i, **

a Regenerative Medicine Group, Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Avenue, Kelvin Grove, QLD4049, Brisbane, Australiab Orthopedic Center for Musculoskeletal Research, University of Wuerzburg, Koenig-Ludwig-Haus, Brettreichstr. 11, 97074 Wuerzburg, Germanyc George W Woodruff School of Mechanical Engineering, Georgia Institute of Technology, 801 Ferst Drive Northwest, Atlanta, GA 30332, USAd Institute for Advanced Study, Technical University Munich, Lichtenbergstraße 2a, 85748 Garching, Munich, Germanye Stem Cell Biology Group and Stem Cells and Cancer Group e Blood and Bone Diseases Program, Mater Research Institute e The University of Queensland,Translational Research Institute, 37 Kent Street, Woolloongabba, QLD 4102, Brisbane, Australiaf Australian Prostate Cancer Research Centre Queensland, Translational Research Institute, 37 Kent Street, Woolloongabba, QLD 4102, Brisbane, Australiag Cells and Tissue Domain, Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Avenue, Kelvin Grove, QLD 4049,Brisbane, Australiah Bones and Immunology Group e Blood and Bone Diseases Program, Mater Research Institute e The University of Queensland, Translational ResearchInstitute, 37 Kent Street, Woolloongabba, QLD 4102, Brisbane, Australiai School of Medicine, The University of Queensland, 288 Herston Road, Herston, QLD 4006, Brisbane, Australia

a r t i c l e i n f o

Article history:Received 9 January 2015Received in revised form24 April 2015Accepted 30 April 2015Available online

Keywords:Tissue EngineeringHumanized nicheBone organHematopoietic stem cellsEngraftmentTransplantation

a b s t r a c t

Advances in tissue-engineering have resulted in a versatile tool-box to specifically design a tailoredmicroenvironment for hematopoietic stem cells (HSCs) in order to study diseases that develop withinthis setting. However, most current in vivo models fail to recapitulate the biological processes seen inhumans. Here we describe a highly reproducible method to engineer humanized bone constructs that areable to recapitulate the morphological features and biological functions of the HSC niches. Ectopic im-plantation of biodegradable composite scaffolds cultured for 4 weeks with human mesenchymal pro-genitor cells and loaded with rhBMP-7 resulted in the development of a chimeric bone organ including alarge number of human mesenchymal cells which were shown to be metabolically active and capable ofestablishing a humanized microenvironment supportive of the homing and maintenance of human HSCs.A syngeneic mouse-to-mouse transplantation assay was used to prove the functionality of the tissue-engineered ossicles. We predict that the ability to tissue engineer a morphologically intact and func-tional large-volume bone organ with a humanized bone marrow compartment will help to furtherelucidate physiological or pathological interactions between human HSCs and their native niches.

© 2015 Published by Elsevier Ltd.

1. Introduction

Multiple cellular players within the organ bone tightly controlthe maintenance, differentiation and proliferation of hemato-poietic stem cells (HSCs) [1]. The components of this controlmechanism form a complex and discrete functional unit calledthe HSC niche. According to the original concept of this niche,hematopoiesis is not possible without the regulating influence ofspecific microdomains of the bone marrow structure, eventhough HSCs per se possess the intrinsic capacity of self-renewal,

* Corresponding author. Institute of Health and Biomedical Innovation, Queens-land University of Technology, 60 Musk Avenue, Kelvin Grove, QLD 4059, Australia.** Corresponding author. Stem Cell Biology Group e Blood and Bone DiseasesProgram, Mater Research Institute, University of Queensland, TranslationalResearch Institute, 37 Kent Street, Woolloongabba, QLD 4102, Brisbane, Australia.

E-mail addresses: dietmar.hutmacher@qut.edu.au (D.W. Hutmacher), jp.levesque@mater.uq.edu.au (J.-P. Levesque).

1 B.M.H. and D.W.H. contributed equally to this study.

Contents lists available at ScienceDirect

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

http://dx.doi.org/10.1016/j.biomaterials.2015.04.0570142-9612/© 2015 Published by Elsevier Ltd.

Biomaterials 61 (2015) 103e114

Fig. 1. Humanized tissue engineered bone constructs (hTEBCs) recapitulate the morphological characteristics and cellular bone marrow composition of a physiological organ bone.(A) Experimental outline: Human mesenchymal progenitor cells (MPCs) were isolated from human cancellous bone chips and cultured on scaffolds under osteogenic conditions toinduce differentiation and production of extracellular matrix proteins. After 8 weeks of culture, scaffolds were implanted into NSG mice (strain: NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ)(n ¼ 6, 2 scaffolds per mouse). Ten-weeks post implantation, humanized ossicles and mouse femora were collected to extract bone marrow cells for flow cytometry. (B) Necropsyshowed that the implanted constructs were surrounded by a highly-vascularised layer of dense connective tissue mimicking periosteum. After careful resection of this tissue, theossicle's cortical shell was visible. The constructs were filled with blood as indicated by their dark red colour. Micro-CT analysis demonstrated the physiological morphology of thehumanized TEBCs. [scale bar: 1 mm]. (C) Histological analyses (H&E) showed that the trabecular spaces of the ossicle were filled with bone marrow. Bone-lining osteoblasts werelocated at the surface of the trabeculae and osteocytes were scattered within the bony matrix. The bone marrow included well preserved expanded sinusoids filled with eryth-rocytes and hematopoietic cell clusters of different lineages such as erythrocytes, lymphocytes and granulocytes. Staining for tartrate resistant acid phosphatase (TRAP, red)

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114104

multi-potency and long-term reconstitution [2]. The endosteal[3,4], mesenchymal [5,6] and vascular systems [7] have beenidentified as the main regulating components of the HSC niches,nevertheless the concept of the niche embodies the physicalentity of all its single constituents [8]. In the last years, re-searchers have become more aware of the fact that the nicheitself can be a driver for pathogenesis, particularly in bonemetastatic disease or leukemia [9,10]. This has led to a plethora ofnew investigative approaches to target not only the replicatingcancer cells but also their microenvironment [11]. A fortiori, thecharacterisation of the cells and extracellular cues that regulatehuman HSC function is essential in order to elucidate the path-ophysiology of diseases that develop within thismicroenvironment.

Over the last decade, a small number of in vivo models havebeen developed that use tissue engineering concepts to mimicthe physiological conditions of a functionally intact organ bone.Using these technologies, it is possible to create a tailored nichefor HSCs in mice [12e17]. However, most previous studies lack acomprehensive analysis of the viability of the transplanted cells,not to mention that they fail to prove whether the cells areindeed functional after transplantation into the host and able tosecrete extracellular matrix components [18,19]. Moreover, someof these studies show none or only small amounts of mineralizedtissue formation e not a fully regenerated organ bone e andoften the newly formed bone and marrow compartment isinterspersed with large volumes of scaffold material, which in-terferes with the development of a coherent physiologicaltissue network [18,19]. Importantly, most current models inves-tigating hematopoietic niche physiology fail to recapitulate theprocesses seen in humans as they analyse the behaviour andfunction of murine e and not human e HSCs within their niches[16,17,20].

To move the field forward, firstly bone engineering modelsare needed that more closely recapitulate both the morphologicaland functional features of a human bone organ and allow totransplant a significant amount of HSCs obtained from asingle engineered bone construct to perform multiple reconstitu-tion assays [16]; and secondly more advanced humanized mousemodels are necessary to study the interactions between humanHSCs and their human niches in vivo including not only the cellularrepertoire of the human niche but also their extracellular matrixproducts [21,22]. We hypothesized that a strategy rooted in theconcept of scaffold guided bone regeneration can be applied toengineer large-volume ectopic bone constructs that incorporatehuman cellular and extracellular niche components and supporthuman hematopoiesis.

2. Materials and methods

2.1. Scaffold preparation and primary cell culture

Tubular medical-grade polycaprolactone (mPCL) scaffolds wereproduced by melt-electrospinning, which has been previouslydescribed in detail by our group [23,24]. mPCL tubes were coatedwith calcium phosphate (CaP) following an in-house protocol topromote cell adhesion, osteogenic differentiation in vitro and boneformation in vivo [25].

2.2. Isolation and culture of human and mouse cells

Approval by the human ethics committee of the QueenslandUniversity of Technology and the Prince Charles Hospital Brisbane,Australia (Approval number: 060/232) was obtained before initia-tion of this study. Human cells were isolated from pelvic cancellousbone fragments and marrow of an otherwise healthy male patientwho underwent total hip replacement. Written informed consentfor sample collection was obtained from the donor in accordancewith the World Medical Association's Declaration of Helsinki.Mesenchymal progenitor cells were isolated and cultured as pre-viously described by our group [26]. To isolate hematopoietic cells,bone fragments were eliminated through a 40 mm cell strainer (BDBioscience, USA). After one wash with 200 mL of PBS containing 2%FCS and 2 mM EDTA, cells were resuspended in 20 mL IMDM (LifeTechnologies, Australia) supplemented with 10% FCS and under-layered with 20 mL Ficoll Paque Plus (GE Healthcare, Australia) in a50 mL tube. After spinning at 400 g, 30 min no break at 18 "C,mononucleated cells were harvested, washed in 50 mL magneticactivated cell sorting buffer (MACS, Miltenyi Biotec, Australia).Pelleted cells were incubated with human CD 34 magnetic selec-tion beads using the Miltenyi CD34þ selection kit (Miltenyi Biotec,Cat# 130-046-702) and separated using theMiltenyi AutoMACS ProSeparator (Miltenyi Biotec, Cat# 130-092-545). CD34þ cells werecounted and resuspended in media (IMDM þ 20% heat inactivatedFCS þ Penicillin-Strepromycin-Glutamine, Life Technologies) con-taining 10 ng/mL recombinant human Stem Cell Factor (rhSCF,Biolegend, Cat# 573902) and incubated overnight at 4 "C on rotor.FACS analysis on an aliquot with PE-conjugated CD34 antibodyshowed that more than 90% of the cells were CD34þ. After 24 h,cells were washed, counted, centrifuged and resuspended in salinesupplemented with 2% FCS.

All animal studies were conducted in conformity with theAustralian Code of Practise for the Care and Use of Animals forScientific Purposes and approved by the ethics committees of theUniversity of Queensland (approval number: 510/09) and theQueensland University of Technology (approval number: 130/025).C57BL/6 B6.SJL-Ptprca Pep3bBoyJ mice were purchased from theAnimal Resources Centre (Perth, Australia). NOD.Cg-Prkdcscid

Il2rgtm1Wjl/SzJ mice and C57BL/6-Tg(UBC-GFP)30Sca transgenicmice (ubcGFP) expressing GFP were derived from an in-housebreeding colony at the Mater Research Institute. Animals usedwere female and 6e8 weeks of age. Plastic adherent mouse MPCswere isolated from the femora of specified mouse strains. Animalswere sacrificed by CO2 asphyxiation. After the hind limb boneswere excised, they were immediately cleaned of all adherent softtissue using a scalpel blade. Bone marrow was flushed out of eachindividual bone using a 21 gauge needlemounted on a 1mL syringeloaded with PBS containing 2% FCS serum and rinsed several times.Empty bones were minced into small pieces and bone fragmentswere washed 3e5 times with PBS containing 2% FCS. Murine MPCswere cultured under the same conditions as human MPCs. Mediachange was performed every 3e4 days. Cellular outgrowthoccurred after 6e9 days. Cells were expanded to passage 3 forfurther experiments.

2.3. Cell seeding and characterization of the implanted constructs

Tubular mPCL scaffolds were sterilized and transferred into a

confirmed the presence of osteoclasts degrading the bone matrix. Megakaryocytes were identified as parasinusoidal cells with a large amount of cytoplasm, multilobulated nucleiand positive staining for von Willebrand Factor (vWF, brown). [scale bars: 100 mm]. (D) Flow cytometry analysis demonstrated the presence of phenotypic HSCs (LKSþ CD48$

CD150þ), multipotent progenitors (LKSþ CD48$ CD150$) and lineage restricted progenitors (LKSþ CD48þ) in the humanized ossicle and mouse femora 10 weeks post-implantation[humanized ossicles are represented by the bright plots, mouse femora by the dark plots]. (For interpretation of the references to colour in this figure legend, the reader is referredto the web version of this article.)

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114 105

Fig. 2. Human bone cells within the TEBCs are metabolically active and contribute to the development of the HSC microenvironment. (A) Cells positive for human nuclear mitoticapparatus 1 (huNuMa; brown stain) were abundantly scattered within the bone matrix (top panel). Cuboidal cells on the surface of bone trabeculae -thus equivalent to osteoblasts-were shown to be largely of human origin (left high power view, white arrows) demonstrating that the human osteoblasts are actively contributing to the production of bone matrixcomponents. Cartilage islands in the centre of the construct also contained human cells (right high power view). All tissues incorporated mouse cells (blue stain, black arrows)indicating that the newly formed organ is a chimera between both species. The bone marrow cells were shown to be exclusively of murine origin. (B) Large parts of the extracellularmatrix proteins were human-derived as shown by positive staining for human-specific collagen type-I (huCol-I, brown) and osteocalcin (huOC, brown). (C) Acid etching of resin-embedded ossicles demonstrated that the human cells were embedded within an intact and functional osteocyte syncytium. This syncytium was closely associated to the vascularsystem (vessel marked with asterisk). [scale bars: 50 mm]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114106

Fig. 3. The marrow contained in ectopic ossicles is host-derived and contains functional HSCs. (A) Experimental outline: Mouse MPCs were isolated from femora of C57BL/6CD45.2þ donors and cultured on the scaffolds. Following osteogenic culture, scaffolds were implanted into congenic B6.SJL CD45.1þ recipients. Tenweeks post-implantation, ossicleswere collected and crushed to extract ectopic bone marrow cells for flow cytometry analyses. Aliquots of ectopic bone marrow cells were subsequently tested in a long-termcompetitive repopulation assay in competition with 3 % 105 marrow cells from untreated C57/BL/6 mice. (B) Proportion of CD11bþ myeloid cells, B220þ B cells and CD3þ Tcells in ossicle bone marrows and proportion of myeloid cells, B cells and T cells expressing the CD45.2 donor and CD45.1 recipient antigens 10 weeks post implantation. (C) Long-term competitive repopulation assay demonstrating that ectopic ossicle marrow contains functional HSCs. Scatter-plots showing expression of CD45.1 and CD45.2 antigens onmyeloid (CD11bþ), B (B220þ) and T (CD3þ) cells in a typical recipient mouse. Diagrams on the bottom row show percentage of CD45.1þ cells amongst myeloid, B and T cells. Each dotrepresents a recipient mouse. Black dots represent values from mice transplanted with ossicle marrow; white dots are values from control recipients that were transplanted withcompeting cells only. The dotted line shows background values derived from control mice.

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114 107

24-well plate. After harvesting with 0.25% trypsin and 1 mMEDTA, 3 % 105 human or mouse MPCs were suspended in 40 mLmedia, added onto each scaffold and incubated for 2 h to allowcell attachment before topping up with 1 mL of culture media perwell. After 4 weeks of static culture, the seeded scaffolds weretransferred to the vessel of a bi-axial rotating bioreactor (Quin-Xell Technologies, Singapore) filled with 500 mL of aMEM sup-plemented with 10% FCS, 10 mM b-glycerophosphate, 0.1 mMdexamethasone and 150 mM phosphoascorbic acid for another 4weeks dynamic cell culture to induce cell differentiation andproduction of extracellular matrix proteins. Cell sheets wereproduced as previously described (Supplementary Fig. 1) [27].The scaffold/cell constructs were extensively characterized priorto implantation into the host animal to assess the quality of thetransplanted biological material in terms of cell viability andamount of extracellular matrix produced (Supplementary Fig. 2)[24]. Then the constructs were subcutaneously implanted intothe right and left flank of the mice as shown in SupplementaryFig. 3.

2.4. Flow cytometry

Mice were sacrificed by CO2 asphyxiation, the tissue engineeredossicles and mouse femora were removed and immediately placedinto ice-cold PBS containing 2% FCS. Samples were either furtherprocessed for flow cytometry analyses or fixed in PBS with 4%paraformaldehyde for 24 h at 4 "C for histology processing orMicro-CT scanning. For flow cytometry, ossicles were gentlycrushed with a mortar and pestle in ice-cold PBS with 2% FCS andthen filtered on a 40 mm cell strainer. Femoral bone marrow washarvested by flushing the bones with PBS containing 2% FCS. Cellsuspensions were then spun at 400 g for 5 min at 4 "C. Supernatantwas removed and cells were resuspended in 1 mL PBS containing2% FCS.

Flow cytometry analyses of the murine hematopoietic systemwas performed as previously described [28]. Aliquots of 1 % 106

bone marrow cells were stained with mouse CD11b-PECY7, B220-APCCY7, mouse CD3ε-Pacific Blue, mouse CD45.1-PE and mouseCD45.2-APC to identify CD45.2 vs CD45.1 chimerism inmyeloid and

Fig. 4. Osteocytes and osteoblasts in ectopic ossicles are mainly donor-derived. (A) Experimental outline: MPCs were isolated from femora of ubcGFP CD45.2þ donors, and culturedon the scaffolds. Following additional culture under osteogenic conditions, scaffolds were implanted into syngeneic C57BL/6 CD45.2þ recipients. 10 weeks post-implantation,ossicles were collected and analysed for GFP expression by immunohistochemistry. Aliquots of ectopic bone marrow cells were subsequently tested in a long-term competitiverepopulation assay in competition with 2 % 105 marrow cells from congenic B6.SJL CD45.1þ mice. (B) Epifluorescence of cell-seeded scaffold before implantation showing that allcells are GFP-positive (left panel). Immunohistochemistry on ossicle sections with GFP antibodies (right panel, GFPþ cells are stained brown). Note the absence of GFPþ cells in themarrow counterstained with haematoxylin. (C) Long-term competitive repopulation assays with the bone marrow from 4 individual ectopic ossicles (os1 e os4). The percentage ofCD45.2þ GFP-negative host derived myeloid, B- and T-cells and the number of repopulating units (RU) per 1 % 106 marrow cells is shown for each individual ossicle. [scale bars:50 mm]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114108

lymphoid lineages. Bone marrow cells were also stained with bio-tinylated mouse CD3ε, CD5, B220, CD11b, Gr1, Ter119, streptavidin-Alexafuor700, anti-Sca1-PECY7, anti-Kit-APC, CD48-PacificBlue andCD150-PE to numerate HSPCs. Dead cells were excluded using 7-amino actinomycin D.

The humanized hematopoietic system was analysed with thefollowing combinations of antibodies mixed in mouse CD16/32hybridoma supernatant and 5 mg/mL purified mouse IgG to blockIgG Fc receptors: A) biotinylated huCD33 (Miltenyi Biotech, Cat#130-098-916) with streptavidin-PECF594 (BD Bioscience,Cat#562284), huCD19-FITC (Biolegend, Cat# 302205), huCD20-FITC (Biolegend, Cat# 302304), huCD38-PECY7 (Biolegend, Cat#303515), huCD3-Pacific Blue (Biolegend, Cat# 300431),huCD34-APC (Biolegend, Cat# 343608), hu/muCD11b-BV605 (Biolegend,Cat# 115539), huCD45-PE (Biolegend, Cat# 304008) and mouseCD45-APCCY7 Biolegend, Cat# 103116200); B) Gr1-FITC (Biolegend,Cat# 108406), hu/muCD11b-BV605 (Biolegend, Cat# 115539), anti-Sca1-PECY7 (Biolegend, Cat# 108114), muCD117-APC (Biolegend,Cat# 105812), muCD48-Pacific Blue (Biolegend, Cat#103418),muCD150-PE (Biolegend, Cat# 115904) and muCD45-APCCY7(Biolegend, Cat#103116). Analysis was gated on viable cellsfollowing exclusion of dead cells with 7-amino actinomycin D (BDBioscience, Cat# 555815).

2.5. Micro-computed tomography (Micro-CT)

Specimens were analysed with a high-resolution Micro-CTscanner (mCT 40, Scanco Medical AG, Switzerland) and scanned at avoxel size of 16 mm. Samples were evaluated at a threshold of 120, afilter width of 1.0 and filter support of 2. Specified samples weredecalcified with 10% EDTA (pH 7.4) for 2 months and the newlyformed vessel network within the TEBC was visualized using aradio-opaque dye (microfil MV-120 blue, MV diluent solution 1:1;Flowtech, MA, USA) as described by Bolland et al. [29]. Sampleswere scanned at a voxel size of 8 mm and evaluated at a threshold of100, a filter width of 0.8 and a filter support of 1.0.

2.6. Histology/immunohistochemistry

Hematoxylin/Eosin (H&E) staining was performed on paraffinsections using a standard protocol to visualize tissue morphology[30]. For immunohistochemical assessment, sections were depar-affinized, re-hydrated and incubated with 3% H2O2 (Sigma-eAldrich) for 30 min. To target intracellular proteins, 0.1% Triton X-100 in PBS was applied for 6 min. Sections were incubated with theprimary antibodies in 2% BSA/PBS following the respective targetretrieval method and having been blocked for 1 h in 2% BSA(Supplementary Table 1). Chromagen development was achievedby incubating sections with 3,3-diaminobenzidine (Sigma Aldrich)for 5 min. Sections were then counterstained with Mayer's hema-toxylin (Sigma Aldrich) and mounted using Eukitt mounting media(Kindler, Germany). Human and mouse bone sections were used aspositive and negative controls, respectively. Finally, samples wereexamined using an Olympus BX-51microscope with a DP-70 digitalcamera and DP controller imaging software (Olympus).

2.7. Visualization of the osteo-canalicular network

Specified ossicles were embedded in poly(methyl-methacrylate). The resin fills the lacunar-canalicular system of theosteocytes, the osteoid and marrow but it does not penetrate theinorganic bone phase. With mild acid etching, the mineral phase isremoved leaving the resin cast behind. A minor finish polish of theembedded sample was achieved by sequential wet sanding with400, 600 and 1200 grit sandpapers. The final polish was performed

with 1.0 alpha alumina powder using a soft cloth polishing wheel.According to the protocol described by Feng et al. [31], the polishedfaces were then etched with 37% phosphoric acid for 2e10 s,washed with sodium hypochlorite for 5 min and briefly rinsed withddH2O. The dried blocks were then coated with gold sputter andanalyzed with a FEI Quanta 200 SEM.

2.8. Transplantation of ossicle marrow leukocytes in long-termcompetitive repopulation assays

The hematopoietic potential of leukocytes contained in theossicle marrow was tested in competitive transplantations inlethally irradiatedmice as previously described [28]. Briefly, 8 weekold female recipient mice (either C57BL/6 or B6.SJL CD45.1þ) werelethally irradiated with 2 doses of 5.5 Gy 3 h apart the day prior totransplantation using a Caesium g source. For transplantations, 106

leukocytes were extracted from ectopic ossicles mixed with300,000 femoral bone marrow cells from naïve congenic mice(either C57BL/6 or B6.SJL CD45.1þ) in a volume of 200 mL of sterilesaline for injection, and then injected into the retro-orbital sinus aspreviously described [33]. 16 weeks post-transplantation, bloodsamples were collected from the retro-orbital sinus and GFP,CD45.1 and CD45.2 expression was measured by flow cytometry inthe B-, T- and myeloid lineages. The number of repopulating units(RU) per 1 % 106 ossicle marrow leukocytes was calculated usingthe following formula: RU ¼ D x C

ð100$DÞ, where D is the percentage ofcells derived from the ossicle marrows amongst B- and myeloidcells, and C is the number of competing RU from the bone marrowof congenic mice co-transplanted with the ossicle marrow cells(C ¼ 1 for 100,000 competing marrow cells) [32,33].

2.9. Humanization of the ossicle's bone marrow

Mice carrying humanized ossicles were irradiated with 2.25 Gyusing a Caesium g source. 24 h later, mice were transplanted withhuman hematopoietic cells isolated from pelvic bone marrow ob-tained from a patient undergoing hip arthroplasty as describedabove. To humanize the ossicle's bone marrow, each mousereceived 13 % 105 CD34þ cells and 1 % 106 CD34$ cells in a volumeof 200 mL via retro-orbital intravenous injection.

2.10. Statistical analyses

Data was analysed with SigmaStat 3.5® software for Windows.For descriptive statistics, values were reported as the mean, stan-dard deviation and range. If the test statistics followed a normaldistribution we used the Student's t-test to evaluate differencesbetween two groups. For non-normally distributed data the Man-neWhitney U-test was performed. The level of significance was setat p ( 0.05.

3. Results

3.1. Tissue engineered bone serves as a humanized HSC niche

Histological and micro-CT analyses revealed that humanizedTEBCs had morphological and structural similarities to native bonetissue, with well-differentiated trabeculae forming a networkinterspersed with bone marrow (Fig. 1 AeC, Supplementary Movie1). The bone mineral density of the humanized TEBCs was com-parable to the density of the mouse femora (876.9 ± 52.6 mgHA/cm3, range 821.9e944.5 [TEBCs] vs. 924.4 ± 31.4 mgHA/cm3, range887.6e974.9 [femora]). After filling the arterial system of the ani-mal with a radio-opaque dye and decalcification of the ossicles for 2months, micro-CT analysis and visualisation of the vascular tree

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114 109

Fig. 5. Human HSCs engraft predominantly within the humanized bone microenvironment. (A) Experimental outline: Human MPCs and human hematopoietic cells were isolatedfrom human cancellous bone chips. Humanized bone constructs were engineered within NSG mice (n ¼ 6, 2 scaffolds per mouse) as described before. Ten weeks post implantation,mice carrying humanized ossicles were irradiated with 2.25 Gy. 24 h later, they were transplanted with 13 % 105 CD34þ cells and 1 % 106 CD34$ cells via the retro-orbital venousplexus. After 5 weeks, mice were euthanized to harvest humanized ossicles and mouse femora. (B) Immunohistochemical staining for human CD45 demonstrated that largenumbers of leukocytes within the bone marrow of the humanized ossicle and the mouse femur were of human origin (huCD45; brown stain). Only few cells were found to bepositive for human CD34 (huCD34; brown stain). Hematopoietic progenitors were identified as cells that stained positive for both huCD45 and huCD34 on serial sections (white

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114110

indicated that not only the outer part of the TEBC but also its centrewas well vascularised (Supplementary Movie 2). The cellularcomposition of the marrow included hematopoietic cells of variouslineages and differentiation stages. Hematopoietic cell clustersconsisting of erythrocytes, lymphocytes, granulocytes and vonWillebrand factor-positive megakaryocytes accumulated at themargins of well preserved expanded sinusoids. The newly formedbone was viable as indicated by an abundance of well-preserved,morphologically intact osteocytes within the lacunae. The pres-ence of cuboidal cells consistent with osteoblasts and tartrate-resistant acid phosphatase-positive osteoclasts in resorption pits,both found on the surface of the newly formed bone matrix, indi-cated an active remodelling process (Fig. 1C).

Marrow was extracted from specified humanized TEBCs (n ¼ 6)and corresponding mouse femora and further analysed by flowcytometry (Fig. 1D) which showed that the average leukocytenumber in the humanized TEBCs was comparable to the one foundin the mouse marrow (10.4 ± 5.8 % 106, range 3.9e19.5 [TEBCs] vs.15.7 ± 2.5 % 106, range 10.7e17.4 [femora]). The ossicles containedphenotypically definedmouse CD11bþ granulocytes andmonocytesand all populations of mouse hematopoietic stem and progenitorcells (HSPCs). Indeed, lineage-negative Kitþ Sca1þ (LKSþ) mouseHSPCs were abundant and further subdivided into LKSþ CD48þ

lineage-restrictedprogenitors and LKSþCD48$ CD150$multipotentprogenitors in proportion comparable to that observed in femoralbonemarrowof the samemice. Interestingly, the proportion of LKSþ

CD48$ CD150þ mouse HSCs was significantly higher in the mousefemora than in the humanized ossicles. No staining for CD3þ orB220þ was performed as NSG mice lack mature B and T cells [34].

To determine the species-origin of the cells assimilated into thebone tissue, immuno-staining with human-specific antibodies wasperformed. A large proportion of the cells within the bone matrix,the cartilage and the fibrous tissue was found to be of human origin(Fig. 2A). These cells were shown to be metabolically active andcapable of developing a humanized bone microenvironment.Extracellular matrix proteins within the newly formed bone werepositive for human-specific collagen type-I and osteocalcin (Fig. 2B).Cell bodies and filopodiae of osteocytes within the humanized bonewere visualized by acid etching of hard-tissue embedded samples.Scanning electronmicroscopy (SEM) demonstrated a high degree ofinterconnecting cytoplasmatic extensions, suggesting a functionalsyncytial humanized osteocyte network (Fig. 2C).

3.2. The marrow of tissue engineered ossicles is functional

In the next experiment we tested whether the hematopoieticsystemwithin the TEBCs is functional. To be able to track the originof transplanted hematopoietic cells in long-term reconstitutionassays, we used a congenic transplantation system using the C57BL/6 strain that expresses the CD45.2 antigen and the B6.SJL congenicstrain that expresses the CD45.1 antigen on leukocytes. For thisexperiment we didn't use NSG mice as a congenic transplantationsystem that allows differentiation between donor and host cells isnot available for this strain. On the other hand, we have demon-strated that immuno-competent strains such as the C57BL/6 don'tallow the development of a humanized organ bone (data notshown). Consequently, we seeded scaffolds with MPCs isolatedfrom femora of CD57BL/6 (CD45.2þ) mice as described for humanMPCs before. Then they were implanted into the flanks of B6.SJL

recipients expressing the CD45.1 antigen (Fig. 3A). Flow cytometryof the ossicle bone marrow revealed 28.8 ± 4.7% B cells, 7.6 ± 1.7% Tcells and 33.0 ± 7.0% myeloid cells (Fig. 3B) which is much closer toskeletal bone marrow (20e30%, 60e70%, 5e10% respectively) thanblood (60%, 20%, 20%) values. All leukocytes contained in TEBCmarrows expressed the CD45.1 antigen with no CD45.2þ leuko-cytes, demonstrating that all hematopoietic cells contained in theossicle marrows were derived from the B6.SJL recipients (Fig. 3B).As observed in the xenogenic assay, the ossicle marrow againcontained all populations of HSPCs in composition and proportionsimilar to that found in femora of mice [35] (SupplementaryFig. 4A). There were no histo-morphological differences betweenmurine TEBCs and the humanized TEBCs (Supplementary Fig. 4B).

To further prove that HSCs contained in the marrow of theseossicles were functional, marrow cells were transplanted in acompetitive repopulation assay into lethally irradiated congenicrecipients. To this end, TEBC marrow cells from 5 different ossicleswere pooled and aliquots of 1 % 106 ossicle marrow cells (CD45.1þ)were mixed with 300,000 bone marrow cells from naïve untreatedC57BL/6 (CD45.2þ) adult congenic mice and transplanted intolethally irradiated C57BL/6 recipients. Sixteen weeks post-transplantation, blood leukocytes were stained for CD45.1 andCD45.2 antigens. CD45.1 chimerismwas detected in myeloid, B andT cells of all recipients (Fig. 3C) demonstrating that the marrowcontained in ectopic ossicles is functional with self-renewing HSCscapable of long-termmulti-lineage hematopoietic reconstitution ina competitive setting.

In order to determine the origin of mesenchymal cells that hadbeen integrated into the mineralized matrix of the ossicles and inits marrow, MPCs were isolated from the femora of ubcGFP trans-genic mice that overexpress GFP in all tissues under the control ofthe human ubiquitin C gene promoter (Fig. 4A). Scaffolds wereseeded with MPCs expressing GFP as detected by fluorescencemicroscopy (Fig. 4B) and cultured as described above. Ten weekspost-implantation into syngeneic C57BL/6 mice, TEBCs wereremoved. Immunohistochemistry for GFP was performed todetermine the origin of the bone forming cells within the ossicles.Immunostaining of ectopic implants clearly demonstrated thatlarge areas of the mineralized ossicles contained donor-derivedGFPþ osteocytes. In particular, cuboidal cells at the interface be-tween the bone matrix and its marrow, thus with the location andmorphology of osteoblasts, were GFPþ. As expected, GFP expressionwas not detected in endothelial cells forming blood vessels or in theossicle bone marrow (Fig. 4B). This experiment confirms the resultsobtained in the xenogenic transplantation assay.

To quantify the HSC content in the marrow of each individualossicle, we performed long-term competitive repopulation assayswith three recipient mice per ossicle following the experimentalstrategy shown in Fig. 4A. Triplicates of 106 leukocytes from fourindividual ossicles were mixed with 200,000 femoral bone marrowcells from congenic B6.SJL CD45.1þ mice and transplanted into 3lethally irradiated B6.SJL mice per ossicle. Blood chimerism 16weeks post-transplantation demonstrated robust competitiveCD45.2þ repopulation in myeloid, B- and T-lineages demonstratingthat the marrow from all ossicles contained long-term competitiverepopulating HSCs. Based on the percentage of CD45.2þ cells in theB- and myeloid lineages, we calculated that ossicles contained13.5 ± 9.5 repopulating units per 1% 106 cells, which is very close tothe frequency observed in the femoral marrow (by definition, one

arrows). Endothelial cells were negative for huCD34 indicating that they are of murine origin. [scale bars: 20 mm; staining for huCD45 and CD34 was performed on serial Section 4mm apart from each other] (C) Flow cytometry analysis demonstrated the presence of phenotypic human hematopoietic progenitors (huCD45þ CD34þ), HSCs (huCD45þ CD34þ

CD38$), B (huCD45þ CD19þ CD20þ) and T cells (huCD45þ CD3$) in the ossicle and femur bone marrow 5 weeks post-transplantation [humanized ossicles are represented by thebright plots, mouse femora by the dark plots]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114 111

RU per 1 % 105 cells) (Fig. 4C). Therefore, quantification by long-term competitive repopulation assays demonstrated that theossicle bone marrow contained host-derived HSCs in equivalentfrequencies to that of femoral marrow in adult mice.

3.3. Tissue engineered humanized bone supports humanhematopoiesis in vivo

Finally, we tested whether humanized ossicles support theengraftment and maintenance of transplanted human HSCs(Fig. 5A). Scaffolds were seeded with human MPCs and subcuta-neously implanted into NSG mice as described above. After 10weeks of bone formation, mice were irradiated with 2.25 Gy. 24 hlater, they received human CD34þ cells by retro-orbital intravenousinjection. After 5 weeks, the experiment was terminated and thefrequency of human hematopoietic stem and progenitor cellswithin the humanized bone and the mouse femur was analysed byflow cytometry. Immunohistochemical analyses demonstrated thepresence of human CD45þ and CD 34þ cells within the ossicle'sbone marrow and mouse femur. CD34þ cells were mainly locatedadjacent to sinusoids, whereas CD45þ cells were scatteredthroughout the bone marrow (Fig. 5B). Flow cytometry revealedstable engraftment for human hematopoietic progenitors(huCD45þ CD34þ) and hematopoietic stem cells (huCD45þ CD34þ

CD38$) in both the mouse femur and the humanized ossicle,respectively (Fig. 5C). Interestingly, we observed a significantlyhigher proportion of huCD45þ CD34þ cells and huCD45þ CD34þ

CD38$ HSCs in the humanized ossicle than in the mouse femur.Human T (huCD45þ CD3þ) and B cells (huCD45þ CD19þ CD20þ)were abundant in both the humanized ossicle and the mousefemur.

4. Discussion

Recent advancements in biomaterials science, tissue engineeringand stem cell biology provided the regenerative medicine commu-nity with the opportunity to establish a toolbox necessary to createin vivo niches for HSCs [16e20]. Scotti et al. [16] used collagen type-Imeshes seeded with human MSCs and cultured them under chon-drogenic conditions before subcutaneous implantation into nudeC57BL/6 mice. The engineered hypertrophic cartilage constructswere remodelled into a bone organ, which was shown to host asimilar proportion of hematopoietic progenitors as the mouse fe-mur. After congenic transplantation of the ossicle's bone marrowcells into lethally irradiated recipients, stable long-term reconsti-tution was achieved [16]. Torisawa et al. [20] placed a mixture ofmurine demineralised bone powder, collagen-I, BMP-2 and BMP-4into a polymer carrier and implanted this construct subcutane-ously into CD-1 mice. The cellular composition of the engineeredbonemarrowwas comparable to the one found in the femora of themice. The ossicles were positioned within a microfluidic porousdevice (“lab-on-chip”) and perfusedwith culturemedia in vitro for 4days. Reconstitution assays demonstrated that the engineered os-sicles contained functional and self-renewing mouse HSCs [20].Groen et al. [18] tried to reconstruct the human hematopoetic nicheby subcutaneous transplantation of calcium phosphate particlesseeded with human MSCs into RAG$/$gc

$/$ mice. Though theyobserved new bone formation around the calcium phosphate par-ticles after 8 weeks, they were not able to prove whether the newlyformed osseous constructs were of human origin [18].

Prima facie, these platforms might provide new insights into thecomplex mutual interactions between HSCs and their niches.However, as they do not account for the obvious differences be-tween murine and human cells [15e17,20] or don't mimic themorphological and functional features of the human niche [18,19],

these model systems - strictly speaking - do not allow a translationof preclinical findings from bench to bedside [36,37]. Relying solelyon such models makes it difficult to predict results in humanmedicine [37].

The application of in regenerative medicine rooted strategies tobioengineer a humanized organ bone with a functional humanizedhematopoietic systemwithin a mouse makes it possible to explorethe full potential of current model systems and will open up newvistas for the treatment of hematopoietic disorders or malig-nancies. In our study, we demonstrated that it is feasible to engi-neer humanized bone constructs in vivo that are capable ofrecapitulating both the morphological features and biologicalfunctions of the human HSC niche. Humanized tissue engineeredbone forms a functional unit, which is able to self-organize and tosupport human hematopoiesis. Furthermore, our observationssuggest that the mutual interactions between HSCs and theirmicroenvironment might be species-specific.

In our experiments, we used a standardized and reproduciblein vitro bone engineering protocol, which allows an in-processcontrol of the seeded cells and their extracellular matrix productsbefore implantation into the animal (Supplementary Figs. 1 and 2).Most other studies lack this validation step [15,17,18,20]. However,due to the utilization of human primary cells, the development of ahighly reproducible in vitro protocol before implantation of theseconstructs into animals is a conditio sine qua non to specificallyanswer questions about any disease development within the HSCniche in the future.

Comprehensive immunostaining for human cells and matrixproteins demonstrated that the human cells incorporated into theconstruct were indeed metabolically active and contributed signif-icantly to the niche formation. Moreover, osteocytes were charac-terised by a high degree of interconnecting osteocytic processessuggestinga functional syncytium.This differs fromprevious ectopicbonemodels that did not provewhether the implanted human cellsare functional and able to generate humanmatrix proteins [15e19].It is well known that the bone microenvironment controls thebehaviour of cells via multiple pathways that include not onlycellecell interactions but also the interplay between cells and theextracellular matrix. Within this context, it has been shown thatbone extracellular matrix proteins mediate adhesion of cancer cells,therebyactivating signallingpathways that regulate cell survival andproliferation [38e41]. Therefore, humanization of the extracellularmatrix is essential to recapitulate species-specific interactions be-tween human tumours and human bone and may also be useful tostudy local factors that influence human stem cell behaviour.

In order to prove that the tissue engineered chimeric organserves as a humanized niche for murine HSCs, the bone marrow ofthe ossicles and mouse femora was analysed using flow cytometry.Indeed, the proportion of murine lineage-negative Kitþ Sca1þ

(LKSþ) progenitors, LKSþ CD48þ lineage-restricted and LKSþ CD48$

CD150$multipotent progenitors in the ossicles were comparable tothat observed in femora of the same mice. However, the proportionof LKSþ CD48$ CD150þ HSCs was significantly higher in the mousebones than in the humanized ossicles.

For NSG mice, a congenic system that allows tracing of cells inlong-term reconstitution assays is not available yet. Therefore, acongenic transplantation assay was used to demonstrate that theossicle bone marrow is fully functional. As seen in the humanizedtransplantation system, the proportion of leukocytes and HSPCs inthe murine tissue engineered ossicles closely resembled thecellular composition found in the marrow of the mouse femora. Viacompetitive long-term reconstitution assays we demonstrated thatthe marrow of ectopic ossicles was functional and contained self-renewing HSCs capable of long-term multi-lineage hematopoieticreconstitution. We calculated that ossicles contained 13.5 ± 9.5

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114112

repopulating units per 106 cells, which is very close to the fre-quency observed in the femora of mice as one RU is defined as thenumber of long-term competitive HSCs contained in 105 bonemarrow leukocytes. In summary, these findings suggest that TEBCscan establish a physiological bone organ capable of maintaining afunctional hematopoietic system.

In the final step of our experiments we aimed to develop ahumanized bone marrow compartment within the humanizedossicles. After irradiation of NSG mice carrying humanized TEBCs,stable engraftment of transplanted HSPCs was observed in themouse organs and humanized ossicles. Interestingly, the propor-tion of human leukocytes was not significantly different betweenthe mouse skeletal marrow and the humanized ossicle marrow.However, the proportion of human HSPCs was significantly higherin the humanized TEBCs suggesting that we have successfullyreconstituted an ectopic humanized HSC niche. On the other hand,we have shown that the proportion of murine HSCs was signifi-cantly higher in the mouse femoral marrow than in the humanizedTEBC marrow, whereas the proportion of murine leukocytes wascomparable between both niches. These observations might beexplained by the lack of species cross-reactivity of many importantfactors that mediate stem cell migration, survival and proliferation[38]. It has been shown that murine IL-3 and granulocyte monocytecolony-stimulating factor (GM-CSF) do not bind to human re-ceptors and that mouse stem cell factor (SCF, kit ligand) binds onlypoorly to human stem cell factor receptor (c-kit) [37,42,43]. Ourunique bone tissue engineering concept is a first step towards thedevelopment of a tailored humanized niche that could potentiallyinclude different human cell types such as mesenchymal or endo-thelial cells and thereby provide the extracellular matrix in com-bination with essential soluble factors necessary for themaintenance and survival of transplanted human HSCs.

5. Conclusion

We have recently shown that it is feasible to engineer a vital andmorphologically intact humanized bone organ within an immuno-compromised murine host by transplantation of bio-degradabletubular composite scaffolds seeded with human mesenchymalprogenitor cells (MPCs) and loaded with recombinant human bonemorphogenetic protein 7 (rhBMP-7) [24]. In the present study wedemonstrated that these humanized tissue engineered bone con-structs (TEBCs) host a fully functional hematopoietic system andrecapitulate not only the morphology but also the function of aphysiological bone organ. We showed that a morphologically intactand large-volume humanized bone organ is supportive of thelodgement and maintenance of human HSCs. After transplantationinto mice via the retro-orbital plexus, human HSCs preferentiallyhomed to the humanized ossicles, resulting in a higher level of bonemarrow humanization than in the mouse femur. This humanizedtissue engineered platform makes it possible to unravel the in-teractions between human HSCs and their native cellular andextracellular niche constituents.

Acknowledgments

This work was supported by the German Research Foundation(DFG HO 5068/1-1 to B.M.H.), the Australian Research Council(Future Fellowship Program to D.W.H.), the National Health andMedical Research Council of Australia (Project Grant 604303 to J.P.L.and I.G.W.; Senior Research Fellowship 1044091 to J.P.L.; ProjectGrants 2643688 and 1082313 to D.W.H. and Career DevelopmentFellowship 1033736 to I.G.W.) and the Mater Foundation (to A.R.P.).Fibrin glue (TISSEEL™ kit) was kindly provided by Baxter Health-care (Australia/New Zealand).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.biomaterials.2015.04.057.

References

[1] J.P. Levesque, F.M. Helwani, I.G. Winkler, The endosteal 'osteoblastic' nicheand its role in hematopoietic stem cell homing and mobilization, Leukemia 24(2010) 1979e1992.

[2] R. Schofield, The relationship between the spleen colony-forming cell and thehaemopoietic stem cell, Blood Cells 4 (1978) 7e25.

[3] J.K. Gong, Endosteal marrow: a rich source of hematopoietic stem cells, Sci-ence 199 (1978) 1443e1445.

[4] J. Zhang, C. Niu, L. Ye, H. Huang, X. He, W.G. Tong, et al., Identification of thehaematopoietic stem cell niche and control of the niche size, Nature 425(2003) 836e841.

[5] L. Ding, S.J. Morrison, Haematopoietic stem cells and early lymphoid pro-genitors occupy distinct bone marrow niches, Nature 495 (2013) 231e235.

[6] A. Greenbaum, Y.M. Hsu, R.B. Day, L.G. Schuettpelz, M.J. Christopher,J.N. Borgerding, et al., CXCL12 in early mesenchymal progenitors is requiredfor haematopoietic stem-cell maintenance, Nature 495 (2013) 227e230.

[7] M.J. Kiel, O.H. Yilmaz, T. Iwashita, C. Terhorst, S.J. Morrison, SLAM family re-ceptors distinguish hematopoietic stem and progenitor cells and revealendothelial niches for stem cells, Cell. 121 (2005) 1109e1121.

[8] A. Nakamura-Ishizu, T. Suda, Hematopoietic stem cell niche: an interplayamong a repertoire of multiple functional niches, Biochim. Biophys. Acta 1830(2013) 2404e2409.

[9] A.E. Karnoub, A.B. Dash, A.P. Vo, A. Sullivan, M.W. Brooks, G.W. Bell, et al.,Mesenchymal stem cells within tumour stroma promote breast cancermetastasis, Nature 449 (2007) 557e563.

[10] S.W. Lane, D.T. Scadden, D.G. Gilliland, The leukemic stem cell niche: currentconcepts and therapeutic opportunities, Blood 114 (2009) 1150e1157.

[11] D.F. Camacho, K.J. Pienta, A multi-targeted approach to treating bone me-tastases, Cancer Metastasis Rev. 33 (2014) 545e553.

[12] P.H. Krebsbach, S.A. Kuznetsov, K. Satomura, R.V. Emmons, D.W. Rowe,P.G. Robey, Bone formation in vivo: comparison of osteogenesis by trans-planted mouse and human marrow stromal fibroblasts, Transplantation 63(1997) 1059e1069.

[13] S.A. Kuznetsov, P.H. Krebsbach, K. Satomura, J. Kerr, M. Riminucci,D. Benayahu, et al., Single-colony derived strains of human marrow stromalfibroblasts form bone after transplantation in vivo, J. Bone Min. Res. 12 (1997)1335e1347.

[14] Y. Miura, Z. Gao, M. Miura, B.M. Seo, W. Sonoyama, W. Chen, et al., Mesen-chymal stem cell-organized bone marrow elements: an alternative hemato-poietic progenitor resource, Stem Cells 24 (2006) 2428e2436.

[15] B. Sacchetti, A. Funari, S. Michienzi, S. Di Cesare, S. Piersanti, I. Saggio, et al.,Self-renewing osteoprogenitors in bone marrow sinusoids can organize ahematopoietic microenvironment, Cell. 131 (2007) 324e336.

[16] C. Scotti, E. Piccinini, H. Takizawa, A. Todorov, P. Bourgine,A. Papadimitropoulos, et al., Engineering of a functional bone organ throughendochondral ossification, Proc. Natl. Acad. Sci. U.S.A. 110 (2013) 3997e4002.

[17] J. Song, M.J. Kiel, Z. Wang, J. Wang, R.S. Taichman, S.J. Morrison, et al., Anin vivo model to study and manipulate the hematopoietic stem cell niche,Blood 115 (2010) 2592e2600.

[18] R.W. Groen, W.A. Noort, R.A. Raymakers, H.J. Prins, L. Aalders, F.M. Hofhuis, etal., Reconstructing the human hematopoietic niche in immunodeficient mice:opportunities for studying primary multiple myeloma, Blood 120 (2012)e9ee16.

[19] J. Lee, M. Li, J. Milwid, J. Dunham, C. Vinegoni, R. Gorbatov, et al., Implantablemicroenvironments to attract hematopoietic stem/cancer cells, Proc. Natl.Acad. Sci. U.S.A. 109 (2012) 19638e19643.

[20] Y.S. Torisawa, C.S. Spina, T. Mammoto, A. Mammoto, J.C. Weaver, T. Tat, et al.,Bone marrow-on-a-chip replicates hematopoietic niche physiology in vitro,Nat. Methods 11 (2014) 663e669.

[21] B.M. Holzapfel, L. Thibaudeau, P. Hesami, A. Taubenberger, N.P. Holzapfel,S. Mayer-Wagner, et al., Humanised xenograft models of bone metastasisrevisited: novel insights into species-specific mechanisms of cancer cellosteotropism, Cancer Metastasis Rev. 32 (2013) 129e145.

[22] L. Thibaudeau, V.M. Quent, B.M. Holzapfel, A.V. Taubenberger, M. Straub,D.W. Hutmacher, Mimicking breast cancer-induced bone metastasis in vivo:current transplantation models and advanced humanized strategies, CancerMetastasis Rev. 33 (2014) 721e735.

[23] T.D. Brown, A. Slotosch, L. Thibaudeau, A. Taubenberger, D. Loessner,C. Vaquette, et al., Design and fabrication of tubular scaffolds via direct writingin a melt electrospinning mode, Biointerphases 7 (2012) 13.

[24] B.M. Holzapfel, F. Wagner, D. Loessner, N.P. Holzapfel, L. Thibaudeau,R. Crawford, et al., Species-specific homing mechanisms of human prostatecancer metastasis in tissue engineered bone, Biomaterials 35 (2014)4108e4115.

[25] C. Vaquette, S. Ivanovski, S.M. Hamlet, D.W. Hutmacher, Effect of cultureconditions and calcium phosphate coating on ectopic bone formation, Bio-materials 34 (2013) 5538e5551.

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114 113

[26] J.C. Reichert, V.M. Quent, L.J. Burke, S.H. Stansfield, J.A. Clements,D.W. Hutmacher, Mineralized human primary osteoblast matrices as a modelsystem to analyse interactions of prostate cancer cells with the bone micro-environment, Biomaterials 31 (2010) 7928e7936.

[27] P. Hesami, B. Holzapfel, A. Taubenberger, M. Roudier, L. Fazli, S. Sieh, et al.,A humanized tissue-engineered in vivo model to dissect interactions betweenhuman prostate cancer cells and human bone, Clin. Exp. Metastasis 31 (2014)1e12.

[28] V. Barbier, I.G. Winkler, J.P. Levesque, Mobilization of hematopoietic stem cellsby depleting bone marrow macrophages, Methods Mol. Biol. 904 (2012)117e138.

[29] B.J. Bolland, J.M. Kanczler, D.G. Dunlop, R.O. Oreffo, Development of in vivomuCT evaluation of neovascularisation in tissue engineered bone constructs,Bone 43 (2008) 195e202.

[30] A.R. Haas, R.S. Tuan, Chondrogenic differentiation of murine C3H10T1/2multipotential mesenchymal cells: II. Stimulation by bone morphogeneticprotein-2 requires modulation of N-cadherin expression and function, Dif-ferentiation 64 (1999) 77e89.

[31] J.Q. Feng, L.M. Ward, S. Liu, Y. Lu, Y. Xie, B. Yuan, et al., Loss of DMP1 causesrickets and osteomalacia and identifies a role for osteocytes in mineralmetabolism, Nat. Genet. 38 (2006) 1310e1315.

[32] L.E. Purton, D.T. Scadden, Limiting factors in murine hematopoietic stem cellassays, Cell. Stem Cell. 1 (2007) 263e270.

[33] I.G. Winkler, A.R. Pettit, L.J. Raggatt, R.N. Jacobsen, C.E. Forristal, V. Barbier, etal., Hematopoietic stem cell mobilizing agents G-CSF, cyclophosphamide orAMD3100 have distinct mechanisms of action on bone marrow HSC nichesand bone formation, Leukemia 26 (2012) 1594e1601.

[34] L.D. Shultz, B.L. Lyons, L.M. Burzenski, B. Gott, X. Chen, S. Chaleff, et al., Humanlymphoid and myeloid cell development in NOD/LtSz-scid IL2R gamma null

mice engrafted with mobilized human hemopoietic stem cells, J. Immunol.174 (2005) 6477e6489.

[35] I.G. Winkler, V. Barbier, R. Wadley, A.C. Zannettino, S. Williams, J.P. Levesque,Positioning of bone marrow hematopoietic and stromal cells relative to bloodflow in vivo: serially reconstituting hematopoietic stem cells reside in distinctnonperfused niches, Blood 116 (2010) 375e385.

[36] D. Loessner, B.M. Holzapfel, J.A. Clements, Engineered microenvironmentsprovide new insights into ovarian and prostate cancer progression and drugresponses, Adv. Drug Deliv. Rev. 79e80 (2014) 193e213.

[37] A. Rongvaux, H. Takizawa, T. Strowig, T. Willinger, E.E. Eynon, R.A. Flavell, etal., Human hemato-lymphoid system mice: current use and future potentialfor medicine, Annu Rev. Immunol. 31 (2013) 635e674.

[38] L.L. Ooi, Y. Zheng, K. Stalgis-Bilinski, C.R. Dunstan, The bone remodelingenvironment is a factor in breast cancer bone metastasis, Bone 48 (2011)66e70.

[39] A.M. Mastro, C.V. Gay, D.R. Welch, The skeleton as a unique environment forbreast cancer cells, Clin. Exp. Metastasis 20 (2003) 275e284.

[40] J.G. Schneider, S.R. Amend, K.N. Weilbaecher, Integrins and bone metastasis:integrating tumor cell and stromal cell interactions, Bone 48 (2011) 54e65.

[41] L. Thibaudeau, A.V. Taubenberger, C. Theodoropoulos, B.M. Holzapfel,O. Ramuz, M. Straub, et al., New mechanistic insights of integrin beta1 inbreast cancer bone colonization, Oncotarget 6 (2015) 332e344.

[42] I. Auffray, A. Dubart, B. Izac, W. Vainchenker, L. Coulombel, A murine stromalcell line promotes the proliferation of the human factor-dependent leukemiccell line UT-7, Exp. Hematol. 22 (1994) 417e424.

[43] X. Jiang, O. Gurel, E.A. Mendiaz, G.W. Stearns, C.L. Clogston, H.S. Lu, et al.,Structure of the active core of human stem cell factor and analysis of bindingto its receptor kit, Embo J. 19 (2000) 3192e3203.

B.M. Holzapfel et al. / Biomaterials 61 (2015) 103e114114