Developmentally inspired programming of adult human mesenchymal stromal cells towards stable chondrogenesis

Developmentally inspired programming of adult

human mesenchymal stromal cells toward

stable chondrogenesis

Paola Occhettaa_{, Sebastien Pigeot}a_{, Marco Rasponi}b_{, Boris Dasen}a_{, Arne Mehrkens}a_{, Thomas Ullrich}c_{, Ina Kramer}d_, Sabine Guth-Gundeld_{, Andrea Barbero}a_{, and Ivan Martin}a,1

a_{Department of Biomedicine, University Hospital Basel, University of Basel, 4031 Basel, Switzerland;}b_{Department of Electronics, Information, and}

Bioengineering, Politecnico di Milano, 20133 Milano, Italy;c_{Global Discovery Chemistry, Novartis Institutes for BioMedical Research, CH-4056 Basel,}

Switzerland; andd_{Musculoskeletal Disease Area, Novartis Institutes for BioMedical Research, CH-4056 Basel, Switzerland}

Edited by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved March 26, 2018 (received for review December 19, 2017) It is generally accepted that adult human bone marrow-derived

mesenchymal stromal cells (hMSCs) are default committed toward osteogenesis. Even when induced to chondrogenesis, hMSCs typically form hypertrophic cartilage that undergoes endochon-dral ossification. Because embryonic mesenchyme is obviously competent to generate phenotypically stable cartilage, it is questioned whether there is a correspondence between mesen-chymal progenitor compartments during development and in adulthood. Here we tested whether forcing specific early events of articular cartilage development can program hMSC fate toward stable chondrogenesis. Inspired by recent findings that spatial restriction of bone morphogenetic protein (BMP) signaling guides embryonic progenitors toward articular cartilage formation, we hypothesized that selective inhibition of BMP drives the pheno-typic stability of hMSC-derived chondrocytes. Two BMP type I receptor-biased kinase inhibitors were screened in a microfluidic platform for their time- and dose-dependent effect on hMSC chondrogenesis. The different receptor selectivity profile of tested compounds allowed demonstration that transient blockade of both ALK2 and ALK3 receptors, while permissive to hMSC cartilage formation, is necessary and sufficient to maintain a stable chondro-cyte phenotype. Remarkably, even upon compound removal, hMSCs were no longer competent to undergo hypertrophy in vitro and endochondral ossification in vivo, indicating the onset of a constitu-tive change. Our findings demonstrate that adult hMSCs effecconstitu-tively share properties of embryonic mesenchyme in the formation of transient but also of stable cartilage. This opens potential pharma-cological strategies to articular cartilage regeneration and more broadly indicates the relevance of developmentally inspired proto-cols to control the fate of adult progenitor

Q:10 cell systems.

mesenchymal stromal cells

|

articular cartilage

|

developmental engineering

|

microfluidics

|

BMPs

R

ecapitulation of embryonic developmental events is receiving increasing recognition as a strategy to induce robust tissue regeneration. The paradigm of “developmental engineering” (1) indeed postulates the importance of instructing developmental processes rather than engineering replacing grafts to achieve func-tional tissue regeneration (2). However, it is still uncertain whether and to what extent specific adult human stem/progenitor cells conserve the capacity to follow developmental pathways. We and others previously demonstrated that the fate of human mesenchy-mal stromesenchy-mal cells (hMSCs) can be sequentially guided toward the recapitulation of the embryonic endochondral ossification program, as well as in the molecular regulation of associated processes, eventually leading to bone formation in vivo (3–7). However, the design of a reliable strategy to generate stable cartilage by adult hMSCs, as occurring in articulating surfaces, still remains an open challenge. Adult MSCs have indeed been proposed to be in-trinsically committed toward terminal hypertrophic differentiation upon induction of chondrogenesis. As a consequence,

hMSC-derived cartilaginous templates typically undergo remodeling in-to bone ossicles when transplanted in vivo (8, 9). “Re-pro-gramming” hMSCs fate by exposure to specific signals has been postulated as a potential strategy to reverse this tendency and maintain the stability of cartilage formed by hMSCs (10). In this direction, signaling factors directly mediating chondrocyte hyper-trophy [i.e., parathyroid hormone-related protein (PTHrP)/Ihh reg-ulatory loop] have been extensively considered (11–13). However, upon removal of the signal, cells typically proceed toward hyper-trophy in vitro and to endochondral ossification in vivo. This suggests that the regulation of downstream signals in the hypertrophy route is not sufficient to modify the intrinsic commitment of hMSCs.

The concept of “developmental engineering” inspired us to operate at an earlier stage, assessing signaling pathways directly involved in embryonic articular cartilage development as po-tential candidates for controlling hMSC fate decision toward stable chondrogenesis. During embryonic joint development, chondroprogenitor cells lying in the interzone have been shown to contribute to either articular or transient cartilage formation (14) based on a tight regulation of bone morphogenetic protein (BMP) signaling (15). The precise spatial organization of the limb mesenchyme indeed results in a band of Noggin-expressing cells, which insulates proliferating chondrocytes in the distal part

Significance

The study offers a pharmacological solution to the challenging target of inducing stable chondrogenesis by human mesen-chymal stromal cells (hMSCs), including protection against vascularization. Cells were reversed from the tendency to fol-low the default differentiation pathway, namely endochondral ossification and osteogenesis. Our findings openperspectives Q:9

in articular cartilage regeneration, as well as in the establish-ment of hMSC-based models of cartilage developestablish-ment, physi-ology, and possibly pathology. Importantly, the results were achieved by mimicking molecular processes occurring during embryonic cartilage formation. This indicates that adult hMSCs share similarities with embryonic mesenchyme and validates the relevance to engineer developmental processes (“de-velopmental engineering”) to control fate specification by adult stem/progenitor cell systems.

Author contributions: P.O., A.B., and I.M. designed research; P.O., S.P., B.D., and A.M. performed research; M.R., T.U., I.K., and S.G.-G. contributed new reagents/analytic tools; S.P. performed in vivo experiments; A.M. provided the cell source; P.O. analyzed data; and P.O., A.B., and I.M. wrote the paper.

The authors declare no conflict of interest. Q:11

This article is a PNAS Direct Submission. Published under thePNAS license.

1_{To whom correspondence should be addressed. Email: ivan.martin@usb.ch.}

This article contains supporting information online atwww.pnas.org/lookup/suppl/doi:10. 1073/pnas.1720658115/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1720658115 PNAS Latest Articles | 1 of 6

APPLIED BIOLOGIC AL SCIENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 Q:1; 2; 3; 4; 5; 6; 7; 8

of the interzone from BMP signaling and specifies their differ-entiation toward articular cartilage (14). Indeed, the BMP an-tagonist Gremlin-1 was found among the most up-regulated genes in human adult articular compared with growth-plate cartilage, suggesting the role of BMP restriction in determining chondroprogenitor commitment (16). On the other hand, BMP signaling is involved in many phases of limb development (17, 18), being a key chondrogenic factor (19) but also a trigger for endochondral ossification (20). Recently, in vitro inhibition of the BMP pathway was exploited to force pluripotent stem cells, either embryonic (21) or induced (22), toward articular-like cartilage in vitro, although with only preliminary evidence of the in vivo phenotype stability.

In this work, we aim to demonstrate that the fate of adult hMSCs can be guided toward stable cartilage by forcing envi-ronmental conditions recapitulating specific events of articular cartilage development. We hypothesize that selective inhibition of BMP signaling during in vitro chondrogenic differentiation of adult hMSC-derived templates is sufficient to prime them toward a stable articular cartilage phenotype. We tested our hypothesis by using two BMP type I receptor-biased kinase inhibitors, dif-fering in their kinase selectivity profile. To screen different doses and temporal stages of supplementation at a high throughput, we introduced a miniaturized microfluidic model, recently de-veloped to allow condensation and subsequent perfusion of 3D microaggregates in each well, crucial to enabling chondrogenesis (23). We then demonstrated the long-term efficacy and the ro-bustness of the selected developmentally inspired protocol both in vitro and in vivo using macroscale tissue models.

Results

BMP Type I Receptor Kinase Inhibitors with Different Selectivity Profiles.Compounds A and B are BMP type I receptor kinase inhibitors with different selectivity profiles. Their activity in biochemical kinase assays and cellular reporter gene assays for BMP and TGF-β signaling are listed in Table S1. Both com-pounds are biased toward ALK2 inhibition. For compound A, the selectivity for ALK2 over ALK1 is only 5-fold (Fig. 1A) and for ALK2 over ALK3 only 4.1-fold (Fig. 1B). With such small selectivity windows, efficacious compound concentrations are expected to always inhibit, at least partially, ALK1 and ALK3 in addition to ALK2. Compound B is a more potent and selective inhibitor of ALK2. Its selectivity over ALK1 is 17.2-fold (Fig. 1D) and 75.3-fold over ALK3 (Fig. 1E). Therefore, at compound

concentrations that are expected to completely inhibit ALK2 (250 nM), the ALK3 receptor kinase is not inhibited (Fig. 1E).

Whereas compounds A and B vary in their ability to inhibit BMP type I receptors, both compounds only poorly inhibit the TGF-β type I receptor ALK5 in biochemical kinase assays (Fig. 1 A and D andTable S1). Their activity in biochemical assays is confirmed in cellular reporter gene assays for BMP and TGF-β signaling. Both compounds potently inhibit BMP6-induced sig-naling in hepatocellular carcinoma cell line (HuH7) cells. Im-portantly, at a concentration of 500 nM, almost completely inhibiting BMP signaling, TGF-β–induced signaling in HEK293 cells is not yet compromised. Only at much higher concentrations is TGF-β signaling also inhibited in vitro, confirming the com-pound’s preference for BMP type I receptor inhibition (Fig. 1 C and F). The specificity in inhibiting BMP signaling was confirmed also in hMSCs (Fig. 1 G and H), whereby both compounds de-creased SMAD1/5/9 phosphorylation compared with controls. Compound A exhibited the highest capacity to down-regulate BMP signaling compared with similar doses of compound B. SMAD2/3 phosphorylation was conversely not affected by the selected doses.

To understand the compound’s mechanism of action on hMSCs, the expression of ALK1, ALK2, and ALK3 receptors was evaluated. Gene-expression analysis revealed the absence of ALK1 in hMSCs, excluding binding to ALK1 as a mechanism involved in this model (Fig. S1A). ALK2 and ALK3 were instead expressed by postnatal day (P0) hMSCs, but not modulated during 14 d of treatment in 3D pellet culture (Fig. S1 B–L). Microfluidic-Based Screening Suggested a Dose- and Timing-Dependent Effect of ALK Inhibitors on Adult hMSCs Commitment. The dose-dependent effect of the BMP signaling inhibitors on adult hMSC commitment was first assessed at the microscale, by exploiting a previously developed microfluidic platform (Fig. S2A) (23). After 14 d, collagen type II deposition was not af-fected by any treatment at any concentration (Fig. 2 A and D). Gene-expression analysis indicates a dose-dependent reduction of ACAN, suggesting that higher doses of both compounds may delay or reduce deposition of cartilaginous matrix (Fig. 2 B and E). A concentration-dependent decrease in collagen type X de-position was detected, with increasing compound concentrations, both at the protein (Fig. 2 A and D) and mRNA (Fig. 2 C and F) levels. This effect was more pronounced upon treatment with compound A, which resulted in complete loss of staining for collagen type X at the highest concentration, correlating with a significant decrease (P < 0.05) in COL10A1 expression com-pared with the control. Results obtained at the microscale level were also confirmed in 3D macropellet culture (Fig. S3 A–D). Different temporal windows for compound A supplementation were then considered. A continuous treatment for 14 d in the presence of TGF-β3 resulted in the most pronounced reduction of hypertrophic matrix deposition (Fig. S2B), correlating with a consistent low expression of Msx2, a transcriptional factor down-stream of the BMP pathway (Fig. S2C). Moreover, to investigate the specificity of the BMP type I receptor LMW kinase inhibitors, 3D pellets were treated with 200 ng/mL of human recombinant Noggin protein, a BMP antagonist. Such conditioning led to the generation of a cartilage matrix rich in collagen type X, thus suggesting that a generic BMP antagonist is not sufficient to avoid a commitment of hMSCs toward hypertrophy (Fig. S4). Thus, a continuous treatment with 500 nM of compound A, which targets both ALK2 and ALK3, was sufficient to maintain hMSCs in a stable cartilage phenotype in a short-term in vitro culture. ALK2 and ALK3 Inhibition on hMSCs-Derived Constructs Is Sufficient to Maintain a Stable Cartilaginous Phenotype in Vitro.We investi-gated how stable the effect of hypertrophy silencing was elicited by ALK2 and ALK3 inhibition in adult hMSC-derived constructs. Three-dimensional hMSC-derived macropellets were incu-bated in standard chondrogenic medium for 28 d with either compound supplemented continuously (Fig. S3 E–H), or for the Fig. 1. Representative IC50curves of both compounds for all assays. (A and

D) Biochemical peptide phosphorylation assay panel. (B and E) Biochemical autophosphorylation assays. (C and F) BMP (on HuH7 cells) and TGF-β (on HEK293 cells) reporter gene assays. (G) BMP and (H) TGF-β pathway activa-tion in adult hMSCs assessed in response to inhibitor treatment by Western blot analysis of SMAD1/5/9 and SMAD2/3 phosphorylation, respectively.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1720658115 Occhetta et al.

125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248

initial 14 d only, followed by incubation in standard chondro-genic medium (“recovery” phase) (Fig. S3 I–L). All constructs treated for 28 d or for 14 d + 14 d of recovery displayed a car-tilaginous extracellular matrix rich in GAG and collagen type II, although chondrogenesis was less pronounced for tissues treated with compound A. Regarding early hypertrophic commitment, compound A treatment was the only condition leading to con-structs negative for collagen type X after 28 d of treatment (Fig. S3E). Interestingly, no collagen type X was deposited even after removal of the compound, suggesting the stability of the estab-lished effect (Fig. S3I). To further prove the sustained effect of initial ALK2/ALK3 inhibition on favoring the generation of stable cartilage, compound A- and compound B-treated hMSC-derived macropellets were forced toward hypertrophy, following a previously established protocol (4). Hypertrophic induction after preincubation with compound A led to deposition of a cartilaginous extracellular matrix rich in GAG and collagen type II, but with negative collagen type X immunostaining (Fig. 3A). Collagen type X was instead detected in constructs treated with compound B and in the control (Fig. 3 B–D). Alizarin red staining indicated the presence of calcified areas in all condi-tions, but upon compound A treatment. Quantitative RT-PCR analyses indicated that compound A treatment did not change COL2A1 expression compared with compound B and control (Fig. 3E), while it significantly reduced COL10A1 and MMP13 expression compared with control (Fig. 3 F and H). IHH ex-pression was also reduced, even if not statistically significantly (Fig. 3G). Taken together, these data indicate that inhibition of both ALK2 and ALK3 receptors (achieved by compound A treatment) is sufficient to prevent hMSCs from expressing both early and late hypertrophic markers in vitro.

Simultaneous Inhibition of ALK2 and ALK3 Down-Regulates the BMP Pathway and Triggers the Activation of Genes Characteristic for Articular Cartilage in Adult Human MSCs.The modulation of key players of the BMP pathway was monitored upon treatment with BMP type I receptor inhibitors. Among endogenous BMP an-tagonists, GREM was up-regulated by both compounds, with compound A inducing the most pronounced increase. hMSCs cultured in standard chondrogenic medium instead down-regulated GREM during the whole culture period (Fig. 4A). NOG uniformly increased and no significant differences were detected among the different conditions, suggesting that Noggin is not modulated by inhibiting ALK2 and ALK3 (Fig. 4B). BMP2 was strongly up-regulated in standard chondrogenic conditions compared with treatment with compound A (Fig. 4C). This trend

was confirmed by detection of secreted BMP2 protein in the culture medium using a specific ELISA, which was consistently reduced by compound A treatment (Fig. S5).

To characterize the nature of the hMSC-derived cartilaginous tissues obtained by treatment with the different compounds, gene-expression changes of a panel of articular or transient cartilage signature genes were assessed byquantitative RT-PCR.Q:12 Expression of Lubricin (PRG4) and Wnt antagonist Frizzled-related protein (FRZb ), typical genes expressed in articular chondrocytes (16), was increased in compound A-treated cells (Fig. 4 D and E). Interestingly, FRZb expression was down-regulated in the control conditions compared with basal levels during the whole culture period. GDF5, a gene associated with joint interzone development (14), was up-regulated upon treat-ment with compound A (Fig. 4F). COL2A1 expression was up-regulated in all conditions compared with basal levels, with compound A-treated constructs showing the least increase of COL2A1 (Fig. 4G). Expression of genes associated with the Fig. 2. A microfluidic platform (23) for culturing hMSCs 3D micromasses was used to screen the concentration-dependent effect of (A–C) compound A and (D–F) compound B. Concentrations spanning over four orders of magnitude (0, 0.5, 5, 50, 500 nM) were assessed. (A and D) Collagen type II (green) and collagen type X (red) stainings were performed directly within the platform after 14 d. All of the images were taken at the same magnification. (Scale bars, 75 μm.) (B–F) Quantitative PCR was performed to assess the mRNA levels of ACAN and COL10A1, by extracting the RNA from the microfluidic platform. All ΔΔCt values are normalized relative to GAPDH expression and refer to basal expression in day 0 hMSCs; all fold-changes in transcript levels are shown in logarithmic scale (n = 3, *P < 0.05, **P < 0.005).

Fig. 3. (A–D) In vitro maturation and stability of 3D macropellets derived from adult hMSC cultured in chondrogenic medium, either alone or supplemented compound A or B, and forced after 14 d of treatment toward hypertrophy. Samples were histologically analyzed for GAG, collagen type II, and collagen type X deposition and Alizarin red. All images were taken at the same mag-nification. (Scale bar, 300 μm.) Quantitative RT-PCR analyses on (E) COL2A1, (F) COL10A, (G) IHH, and (H) MMP13 confirmed the immunohistochemistry stain-ing. All ΔΔCt values are normalized relative to GAPDH expression and refer to basal expression in day 0 hMSCs; all fold-changes in transcript levels are shown in logarithmic scale(n = 3, *P < 0.05, ****P < 0.0001). Q:19

Occhetta et al. PNAS Latest Articles | 3 of 6

APPLIED BIOLOGIC AL SCIENCES 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372

initiation of hypertrophy and angiogenesis/vascularization— namely COL10A1, IHH, MMP13, and VEGFA—was significantly higher in hMSC-derived constructs cultured in standard chon-drogenic medium, compared with compound A-treated samples (Fig. 4 H–K). PTHRP exhibited

Q:13 instead a variable trend (Fig.

4L), which would be challenging to interpret due to its in-volvement in complex feedback loops (20). Thus, inhibition of ALK2 and ALK3 contributed to the in vitro generation of hMSC-derived constructs with increased expression of genes related to articular cartilage development and function, while downregulating expression of genes associated with growth-plate development.

Preincubation with an ALK2/ALK3 Inhibitor in Vitro Favors a Stable Cartilaginous Phenotype of hMSC-Derived Constructs upon Ectopic Implantation in Vivo.To assess whether an in vitro preincubation with BMP type I receptor inhibitors is sufficient to prevent remodeling of hMSC-derived cartilaginous constructs in vivo, compounds A and B pretreated and control samples were implanted ectopically in immunodeficient mice.

Upon retrieval (8 wk after implant), constructs displayed dif-ferent morphologies. In vitro pretreatment with compound A favored the maintenance of stable cartilaginous matrices after 8 wk of implantation, characterized by high GAG content and absence of bone remodeling (Fig. 5A andFig. S6A). Constructs treated in vitro with compound B progressed differently depending on the concentration used. The higher concentration (500 nM) resulted in a reduction of GAG accumulation over time; however, this still did not lead to complete cartilage remodeling during the observation timeframe (Fig. 5B andFig. S6B). Constructs treated with the lower concentration (250 nM) underwent a clear remodeling, resulting after 8 wk in complete cartilage resorption and bone formation (Fig. 5C andFig. S6C). As expected, in the control condition a complete remodeling of the tissue was observed, with the cartilage entirely replaced by bone and bone marrow after 8 wk (Fig. 5D andFig. S6D). Taken together, these data indicate that in vitro inhibition of both ALK2 and ALK3, achieved with 500 nM of compound A, is sufficient to prevent hMSCs-derived cartilaginous constructs from remodeling in vivo. Instead, even complete inhibition of ALK2 alone with only marginal or no inhibition of ALK3, as

seen with 500 and 250 nM of compound B, respectively, can only delay the remodeling process.

Simultaneous ALK2 and ALK3 Inhibition Prevents hMSCs from Bone Remodeling by Activating a Protective Mechanism Against Vascularization. Immunofluorescence staining for vessel invasion (CD31, αSMA and DAPI) (Fig. 6 A–D) revealed different patterns of vasculature 8 wk after construct implantation. Constructs pretreated with 500-nM compound A remained avascular after 8 wk in vivo (Fig. 6A). Con-sistent with the observed bone remodeling (Fig. 5 C and D), both samples pretreated with ALK2-selective doses of compound B (250 nM) and control constructs resulted in efficient vascularization, exhibiting deep-vessel colonization (Fig. 6 C and D). Samples pre-treated with compound B at high doses showed formation of vessels principally localized at the periphery (Fig. 6B). To further elucidate the underlying mechanism, the effect of in vitro pretreatment on regulating vascularization was assessed. Chondromodulin, an anti-angiogenic factor stabilizing the chondrocyte phenotype (24), was not expressed by hMSCs cultured in standard chondrogenic medium, which committed them toward hypertrophy (Fig. 6H). In vitro pre-treatment with either compound A or high doses of compound B induced Chondromodulin expression in hMSCs after 14 d (Fig. 6 E and F). Nasal chondrocytes, which were shown to readily differentiate into articular cartilage in vitro (25), were used as a positive control. In chondrogenic medium, nasal chondrocyte-derived constructs form stable cartilaginous-like matrices and high levels of Chondromodulin were confirmed (Fig. S7). Interestingly, Chondromodulin was still detectable in compound A pretreated constructs even 8 wk after implantation in vivo (Fig. 6A), while absent with all other conditions (Fig. 6 B–D). In addition, gene-expression levels of leukemia in-hibitory factor (LIF), whose deficiency has been reported to enhance vascular endothelial growth factor (VEGF) levels in developing neonatal bone (26), were significantly higher in compound A treated constructs compared with both compound B treatments and control (Fig. 6I). VEGF amounts secreted by hMSCs were unaltered by any of the culture conditions, suggesting that ALK2 or ALK3 inhibition does not prevent the expression of proangiogenic factors (Fig. 6J). Taken together, these data suggest the initiation of a protective mechanism against vascularization of the cartilaginous constructs upon inhibition of ALK2 and ALK3 during the early differentiation events.

Fig. 4. Gene-expression levels of BMP antagonists (A) GREM and (B) NOG and of (C) BMP-2. (D–G) Expression of articular cartilage related genes PRG4, FRZb, GDF5, COL2A1. (H–L) Expression of growth plate development asso-ciated genes COL10A1, IHH, MMP13, VEGF, PTHRP. All ΔΔCt values are normalized relative to GAPDH expression and refer to basal expression in day 0 hMSCs; all fold-changes in transcript levels are shown in logarithmic scale (n = 3; *P < 0.05, **P < 0.005).

Fig. 5. Development of hMSCs-derived cartilage tissues following 8 wk of ectopic implantation in vivo. Safranin O staining identifies GAG deposition; Masson Trichrome and H&E staining allow identify the presence of bone remodeling and bone marrow. (A) Compound A pretreatment was the only condition allowing the maintenance of a cartilaginous matrix still rich in GAG after 8 wk. (B–D) Allof the other conditions exhibited a progressionQ:20

toward remodeling in bone. All images were taken at the same magnifica-tion (n = 3 donors). (Scale bar, 300 μm.)

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1720658115 Occhetta et al.

373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496

Discussion

Following the concept of developmental engineering (1), we investigated in this study the role of BMP signaling on the dif-ferentiation of adult mesenchymal progenitor cells, namely hMSCs, toward articular cartilage. The combination of (i) a microfluidic system for time- and dose-dependent screening of soluble factors on 3D microaggregates and of (ii) synthetic compounds selectively and specifically silencing the BMP path-way by targeting ALK2 and ALK3 receptors allowed us to suc-cessfully design a strategy effective in guiding hMSCs toward stable articular-like cartilage, both in vitro and in vivo. We also postulate the role of ALK2 and ALK3 inhibition in triggering a protective mechanism against vascularization, eventually favor-ing the maintenance of stable and avascular cartilage tissue.

Recently, the achievement of an organized cartilage template based on hMSCs was reported by recreating in vitro the native ar-ticular cartilage spatial organization. However, the only partial in-hibition of hypertrophic markers achieved in this model suggests that key players are still missing to drive hMSCs toward stable cartilage (27). In this direction, few studies have recently focused on manipulating key signaling factors involved in growth-plate devel-opment. Exposure to Wnt3a (either alone or in combination with FGF2) was indeed shown to enhance MSCs undifferentiated pro-liferation, while priming cells toward a more efficient chondrogenic differentiation (28, 29). While providing promising results in forming stable cartilage, the expression of hypertrophic markers could not be completely inhibited by following these approaches. Sequential exposure to FGF-2, -9, or -18 was also reported as a strategy to delay, but not to avoid hypertrophic differentiation of hMSCs (30). PTHrP/Ihh regulatory loop inhibition was also pro-posed. However, both continuous (11) and intermittent (12, 13) supplementation of PTHrP was not sufficient to diminish exoge-nously induced enhancement of hypertrophy in hMSC and finally resulted in unstable in vivo cartilage formation. To overcome these issues, instead of working on downstream signals of the hypertrophy route, our approach focused on earlier stages considering signaling pathways directly involved in articular cartilage fate specification during embryonic development (14, 17). Recent evidence demon-strated the requirement of a BMP signaling restriction during mouse articular cartilage development, achieved through up-regulation of

BMP inhibitors in specific spatial location (14, 15). Moreover, BMP inhibition was exploited together with manipulation of other sig-naling pathways to guide embryonic and induced pluripotent stem cells differentiation through mesoderm intermediates to a final chondrocyte population (21, 22). In this study, we translate this concept to an adult human population of mesenchymal pro-Q:14 genitor cells, being in an advanced stage of commitment and typically considered intrinsically programmed on a default pathway to osteogenesis.

By using two ALK2-biased BMP type I receptor kinase inhibi-tors, we were able to selectively silence the BMP signaling pathway on 3D hMSC-based cultures, while minimizing unwanted off-tar-get effects (i.e., inhibition of the TGF-β pathway). The exploita-tion of such synthetic low molecular-weight compounds gave us a more reproducible effect compared with the supplementation of a native BMP antagonist (e.g., Noggin), which might trigger nega-tive feedback loops (17, 31), in turn making the system more complex and less predictable. Moreover, a microfluidic platform for generating, culturing and conditioning 3D hMSC-derived micromasses (32) allowed us to design an efficient and fine-tuned BMP inhibition strategy. Specifically, the most efficacious doses and timing of inhibitors were preselected based on their capacity to down-regulate hypertrophic markers, while maintaining high levels of cartilage-specific markers, in hMSC-derived micromass cultures. Notably, the microscale model was highly predictive of results subsequently obtained at the macroscale and in vivo. This confirms the potential of such microfluidic platforms as a powerful tool to initially set up differentiation protocols in a more efficient and medium-throughput fashion.

The different selectivity profiles of the two tested compounds allowed to unravel the contribution of ALK2- and ALK3-mediated signaling toward guiding hMSC commitment. Recent findings demonstrated how a synergistic inhibition of ALK2 and ALK3 is essential for reducing heterotopic ossification, a progressive for-mation of ectopic bone in soft tissue (33). In accordance, we demonstrated that inhibition of only ALK2 by using selective doses of compound B is not sufficient to sustainably silence the hyper-trophic phenotype in hMSC-derived 3D constructs. Conversely, dual inhibition of ALK2 and ALK3, achieved by either nanomolar doses of compound A or by micromolar doses of compound B (Fig. S8) reduced the expression of hypertrophic markers and contrib-uted to the maintenance of an articular-like cartilage phenotype of hMSC-derived constructs. Interestingly, a pretreatment for 14 d in vitro was sufficient for stabilizing this antihypertrophic effect, in accordance with previous reports (33). This suggests the possible induction of permanent modifications triggered within hMSCs by inhibition of ALK2 and ALK3, warranting further investigations to elucidate the responsible mechanism.

Simultaneous inhibition of ALK2 and ALK3 by compound A clearly resulted in a reduced expression of genes associated with endochondral ossification. The result could lead to the in-terpretation that such treatment induces a delay in chondro-genesis, rather than a specific reprogramming toward stable cartilage. However, the former possibility is not consistent with the findings that compound A resulted in (i) significantly higher ex-pression of genes characterizing articular cartilage (e.g., Grem1, PRG4, and FRZb) (Fig. 4), and most importantly, (ii) long-term in vivo stability of the generated cartilage matrix (Fig. 5). Thus, we may conclude that cartilaginous-like constructs obtained through continuous dual inhibition of ALK2 and ALK3 in vitro contained all of the “biological instructions” necessary to prevent entering the endochondral ossification pathway. We also considered that the process of hypertrophic cartilage remodeling into bone in an ectopic in vivo environment is typically associated with invasion by new vessels. Chondromodulin is a protein expressed in the avas-cular zone of prehypertrophic cartilage and its expression de-creases during chondrocyte hypertrophy and vascular invasion (24). Interestingly, in vitro hMSCs expressed Chondromodulin only upon a dual inhibition of ALK2 and ALK3, and maintained high-expression levels even after implantation in vivo. A previous study demonstrated the importance of silencing vascularization for Fig. 6. Effect of treatments with compounds A and B on vascularization

in vitro and in vivo. (A–D) After

Q:21 8 wk in vivo, construct vascularization was assessed by immunofluorescence staining for CD31 (red), αSMA (gray), and Chondromodulin (green). (E–H) Chondromodulin staining after 14 d of pretreatment in vitro. All images taken at the same magnification. (Scale bar, 300 μm.) (I) Quantitative RT-PCR was performed to assess the mRNA levels of LIF in all conditions. All ΔΔCt values are normalized relative to GAPDH expression and refer to basal expression in day 0 hMSCs; all fold-changes in transcript levels are shown in logarithmic scale (n = 3, *P < 0.05, **P < 0.005). (J) VEGF secreted by hMSCs was measured by ELISA at day 2, day 7 and day 14 of culture.

Occhetta et al. PNAS Latest Articles | 5 of 6

APPLIED BIOLOGIC AL SCIENCES 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620

generating stable cartilage templates, by genetically modifying hMSCs to continuously produce soluble VEGF receptor-2 (sFlk1) (34). We here demonstrated that an in vitro pretreatment tar-geting ALK2 and ALK3 is sufficient to trigger a protective mechanism against vascularization within hMSC-derived con-structs, which eventually concurs in stabilizing their cartilaginous phenotype in vivo.

The implantation of hMSC-derived constructs primed toward articular cartilage in an orthotopic in vivo model will demon-strate if this system can potentially be used as a cartilage

Q:15

re-generation strategy. In perspective, our study may open the possibility to use hMSC-derived and primed tissues as grafts for cartilage tissue engineering. Similarly, hydrogel-based strategies could be envisaged, in which the BMP type I receptor kinase inhibitors are delivered directly at the site of the injury, poten-tially in combination with mobilization of progenitor cells from subchondral bone, which typically include MSC populations.

In conclusion, we demonstrate that restriction of BMP sig-naling can program adult hMSC to stable chondrogenesis, in-cluding protection from vessels invasion, and thus revert the postulated default commitment to osteogenesis. The strategy, which is inspired by recapitulation of critical aspects of normal articular cartilage formation during embryonic development, can be exploited to design pharmacological cartilage regeneration approaches, as well as to generate adult hMSC-based models of cartilage development, physiology, and possibly pathology. As a broader perspective, our achievements indicate that adult hMSCs share regulatory processes of embryonic mesenchyme, and strengthen the validity of developmental engineering as a paradigm to control the fate of adult progenitor cell systems.

Materials and Methods

All human samples were collected with informed consent of involved indi-viduals, and mouse experiments were performed in accordance with Swiss law. All studies were approved by the ethics board of the canton Basel, Switzerland and by the Swiss Federal Veterinary Office. hMSCs were ex-panded for two passages (average of 13–15 doubling), to minimize the loss of chondrogenic potential according to previously described protocols(35).Q:16

BMP type I receptor kinase inhibitors with different selectivity profiles (compounds A and B) were obtained from Novartis Pharma AG, Basel and they may be provided upon requestunder a material transfer agreementQ:17

specifying nondisclosure conditions, and restrictions regarding the use of the materials and the results obtained therewith. The dose- and timing-de-pendent effects of inhibitors were tested by culturing hMSC-derived micromasses for 14 d in a previously developed microfluidic platform (23). Micromasses were analyzed by immunofluorescence directly within the platform and RNA was extracted for quantitative RT-PCR analyses. Cells were then either cultured in a 3D scaffold-free macroscale pellet model (0.25 × 106

cells per pellet) or seeded onto type I collagen meshes at a density of 7 × 106

cells/cm3_{and cultured for up to 4 wk in serum-free chondrogenic medium,}

supplemented or not with compound A or B. After 2 wk of pretreatment in vitro, samples were implanted in the subcutaneous pouches of nude mice (four samples per mouse) and retrieved after 8 wk postimplantation. The generated tissues were analyzed histologically, immunohistochemically, and by quantitative RT-PCR; specific ELISAs were used to quantified the amounts of secreted proteins. A more complete and detailed description of the methods is included inSI Materials and Methods.

ACKNOWLEDGMENTS. We thank S. Schären and M. Haug for the kind pro-vision of bone marrow aspirates and nasal cartilage biopsies. This work was partially funded by the Swiss National Science Foundation Grant NBM 1579 (to I.M.).

1. Lenas P, Moos M, Jr, Luyten FP (2009) Developmental engineering: A new paradigm for the design and manufacturing of cell-based products. Part I: From three-di-mensional cell growth to biomimetics of in vivo development. Tissue Eng Part B Rev 15:381–394.

2. Tonnarelli B, Centola M, Barbero A, Zeller R, Martin I (2014) Re-engineering devel-opment to instruct tissue regeneration. Curr Top Dev Biol 108:319–338.

3. Scotti C, et al. (2010) Recapitulation of endochondral bone formation using human adult mesenchymal stem cells as a paradigm for developmental engineering. Proc Natl Acad Sci USA 107:7251–7256.

4. Scotti C, et al. (2013) Engineering of a functional bone organ through endochondral ossification. Proc Natl Acad Sci USA 110:3997–4002.

5. Sheehy EJ, Vinardell T, Buckley CT, Kelly DJ (2013) Engineering osteochondral con-structs through spatial regulation of endochondral ossification. Acta Biomater 9: 5484–5492.

6. Farrell E, et al. (2011) In-vivo generation of bone via endochondral ossification by in-vitro chondrogenic priming of adult human and rat mesenchymal stem cells. BMC Musculoskelet Disord 12:31.

7. Freeman FE, Haugh MG, McNamara LM (2016) Investigation of the optimal timing for chondrogenic priming of MSCs to enhance osteogenic differentiation in vitro as a bone tissue engineering strategy. J Tissue Eng Regen Med 10:E250–E262. 8. Pelttari K, et al. (2006) Premature induction of hypertrophy during in vitro

chon-drogenesis of human mesenchymal stem cells correlates with calcification and vas-cular invasion after ectopic transplantation in SCID mice. Arthritis Rheum 54: 3254–3266.

9. Gaut C, Sugaya K (2015) Critical review on the physical and mechanical factors in-volved in tissue engineering of cartilage. Regen Med 10:665–679.

10. Occhetta P, Stüdle C, Barbero A, Martin I (2016) Learn, simplify and implement: De-velopmental re-engineering strategies for cartilage repai. Swiss Med Wkly 146: w14346.

11. Mueller MB, Tuan RS (2008) Functional characterization of hypertrophy in chondro-genesis of human mesenchymal stem cells. Arthritis Rheum 58:1377–1388. 12. Fischer J, Aulmann A, Dexheimer V, Grossner T, Richter W (2014) Intermittent PTHrP

(1-34) exposure augments chondrogenesis and reduces hypertrophy of mesenchymal stromal cells. Stem Cells Dev 23:2513–2523.

13. Weiss S, Hennig T, Bock R, Steck E, Richter W (2010) Impact of growth factors and PTHrP on early and late chondrogenic differentiation of human mesenchymal stem cells. J Cell Physiol 223:84–93.

14. Ray A, Singh PNP, Sohaskey ML, Harland RM, Bandyopadhyay A (2015) Precise spatial restriction of BMP signaling is essential for articular cartilage differentiation. Development 142:1169–1179.

15. Brunet LJ, McMahon JA, McMahon AP, Harland RM (1998) Noggin, cartilage mor-phogenesis, and joint formation in the mammalian skeleton. Science 280:1455–1457. 16. Leijten JCH, et al. (2012) Gremlin 1, frizzled-related protein, and Dkk-1 are key

reg-ulators of human articular cartilage homeostasis. Arthritis Rheum 64:3302–3312. 17. Pignatti E, Zeller R, Zuniga A (2014) To BMP or not to BMP during vertebrate limb bud

development. Semin Cell Dev Biol 32:119–127.

18. Bénazet J-D, et al. (2012) Smad4 is required to induce digit ray primordia and to initiate the aggregation and differentiation of chondrogenic progenitors in mouse limb buds. Development 139:4250–4260.

19. Karamboulas K, Dranse HJ, Underhill TM (2010) Regulation of BMP-dependent chondrogenesis in early limb mesenchyme by TGFbeta signals. J Cell Sci 123: 2068–2076.

20. Kronenberg HM (2003) Developmental regulation of the growth plate. Nature 423: 332–336.

21. Craft AM, et al. (2013) Specification of chondrocytes and cartilage tissues from em-bryonic stem cells. Development 140:2597–2610.

22. Craft AM, et al. (2015) Generation of articular chondrocytes from human pluripotent stem cells. Nat Biotechnol 33:638–645.

23. Occhetta P, et al. (2015) High-throughput microfluidic platform for 3D cultures of mesenchymal stem cells, towards engineering developmental processes. Sci Rep 5: 10288.

24. Hiraki Y, et al. (1997) Identification of chondromodulin I as a novel endothelial cell growth inhibitor. Purification and its localization in the avascular zone of epiphyseal cartilage. J Biol Chem 272:32419–32426.

25. Pelttari K, et al. (2014) Adult human neural crest-derived cells for articular cartilage repair. Sci Transl Med 6:251ra119.

26. Poulton IJ, McGregor NE, Pompolo S, Walker EC, Sims NA (2012) Contrasting roles of leukemia inhibitory factor in murine bone development and remodeling involve re-gion-specific changes in vascularization. J Bone Miner Res 27:586–595.

27. Ng JJ, et al. (2017) Recapitulation of physiological spatiotemporal signals promotes in vitro formation of phenotypically stable human articular cartilage. Proc Natl Acad Sci USA 114:2556–2561.

28. Centola M, et al. (2013) Priming 3D cultures of human mesenchymal stromal cells toward cartilage formation via developmental pathways. Stem Cells Dev 22: 2849–2858.

29. Narcisi R, et al. (2015) Long-term expansion, enhanced chondrogenic potential, and suppression of endochondral ossification of adult human MSCs via WNT signaling modulation. Stem Cell Reports 4:459–472.

30. Correa D, et al. (2015) Sequential exposure to fibroblast growth factors (FGF) 2, 9 and 18 enhances hMSC chondrogenic differentiation. Osteoarthritis Cartilage 23:443–453. 31. Freeman M (2000) Feedback control of intercellular signalling in development.

Nature 408:313–319.

32. Occhetta P, Visone R, Rasponi M (2017) High-throughput microfluidic platform for 3D cultures of mesenchymal stem cells. 3D Cell Culture: Methods and Protocols, ed Koledova Z (Springer, New York), pp 303–323.

33. Agarwal S, et al. (2017) Strategic targeting of multiple BMP receptors prevents trauma-induced heterotopic ossification. Mol Ther 25:1974–1987. Q:18 34. Marsano A, et al. (2016) Spontaneous in vivo chondrogenesis of bone marrow-derived

mesenchymal progenitor cells by blocking vascular endothelial growth factor sig-naling. Stem Cells Transl Med 5:1730–1738.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1720658115 Occhetta et al.

621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744

Q: 1_Please contact PNAS_Specialist.djs@sheridan.com if you have questions about the editorial

changes, this list of queries, or the figures in your article. Please include your manuscript number in

the subject line of all email correspondence; your manuscript number is 201720658.

Q: 2_Please (i) review the author affiliation and footnote symbols carefully, (ii) check the order of the

author names, and (iii) check the spelling of all author names, initials, and affiliations. To confirm

that the author and affiliation lines are correct, add the comment “OK” next to the author line. This

is your final opportunity to correct any errors prior to publication. Misspelled names or missing

initials will affect an author’s searchability. Once a manuscript publishes online, any corrections (if

approved) will require publishing an erratum; there is a processing fee for approved erratum. Please

check with your coauthors about how they want their names and affiliations to appear.

Q: 3_Please review the information in the author contribution footnote carefully. Please make sure that the

information is correct and that the correct author initials are listed. Note that the order of author

initials matches the order of the author line per journal style. You may add contributions to the list in

the footnote; however, funding should not be an author’s only contribution to the work.

Q: 4_You have chosen not to pay for the PNAS open access option. Please confirm that this is correct. If

this is incorrect, and you would like to pay for the PNAS open access option, please note this in the

margin.

Q: 5_Supporting Information (SI) that is not composed in the main SI PDF (appendices, datasets, movies,

and “Other Supporting Information Files”) has not been altered in any way from your originally

submitted file. The composed portion of your SI is included in this proof as subsequent pages

following the last page of the main text. The noncomposed pieces of SI are available via separate

links in the Author Center. If you did not receive the proofs with your composed SI or if you are

unable to view your noncomposed SI in the Author Center, please contact PNAS_Specialist.

djs@sheridan.com.

Q: 6_Please check the order of your keywords and approve or reorder them as necessary. Note that PNAS

allows up to five keywords; please do not add new keywords unless you wish to replace others.

Q: 7_Per PNAS style, certain compound terms are hyphenated when used as adjectives and unhyphenated

when used as nouns. This style has been applied consistently throughout where (and if) applicable.

Q: 8_The article title “Developmentally inspired programming of adult human mesenchymal stromal cells

toward stable chondrogenesis” given in the manuscript is different from the article title given in the

article metadata (“towards” has an “s” in the metadata.) Okay as is?

Q: 9_PNAS does not allow claims of priority or primacy, hence the term “new” has been deleted in the

Significance statement.

AUTHOR QUERIES

Q: 10_PNAS italicizes genes and alleles. All genes, alleles, and proteins will appear as indicated in your

proofs. Please carefully check use of italics and capital letters throughout the proof and correct as

necessary. If, by “

■■■ gene,” you mean “the gene that encodes protein ■■■,” then italic type

is not necessary. (NOTE: If all instances of a gene/allele should be changed, please make only one

correction and indicate that it is a global change.)

Q: 11_Please confirm that there are no conflicts of interest related to authors’ employment at Novartis.

Q: 12_Does edit preserve your intent? Please verify the expansion of RT-qPCR as “quantitative RT-PCR,”

as it is used elsewhere in the paper.

Q: 13_As meant? Please verify “PTHRP” in the sentence beginning “PTHRP exhibited ...” as it is written

“PTHrP” most other places in the paper.

Q: 14_PNAS does not allow claims of priority or primacy, hence the term “for first the first time” has been

deleted in the sentence beginning “In this study, we ...”

Q: 15_PNAS does not allow claims of priority or primacy, hence the term “new” has been deleted in the

sentence beginning “The implantation of ...”

Q: 16_Please note in the sentence beginning “hMSCs were expanded . . .” that there is no ref. 35 in the

reference list. Please either provide the correct reference in this sentence, or provide a ref. 35 in the

reference list. Was the same ref. as SI ref. 1 (Martin et al.) intended instead?

Q: 17_All data relevant to an article should be made publicly available to readers. The data designated as

“

available upon request” should be deposited in a publicly available repository before publication

and indicated in a data deposition footnote. Please let us know if you have concerns or issues with

this course of action.

Q: 18_Please verify ref. 33 has amended.

Q: 19_Please note that in Fig. 3 there are triple asterisks, not quadruple (four). Please amend the P values

or asterisks accordingly.

Q: 20_In the legend to Fig. 5, the callout for “B, D” has been changed to “B-D” to account for panel “C.”

Okay? If this is in error, please provide a description for panel C as well.

Q: 21_Please provide a value for the scale bar in Fig. 6 A–D.

AUTHOR QUERIES

Supporting Information

Occhetta et al. 10.1073/pnas.1720658115

SI Materials and Methods

The kinase activity was either determined using an autophos-phorylation assay (ALK2 and ALK3) or a peptide phosautophos-phorylation assay (ALK1, ALK2, ALK5, and ALK6). For all biochemical assays, human recombinant proteins were expressed in and purified from baculo virus-transfected insect cells. The constructs comprised the GS-domain and kinase domain of ALK1 (amino acids 166–493), ALK2 (amino acids 172–499), ALK3 (amino acids 198–525), or ALK5 (amino acids 162–503). The kinase activity of ALK2 and ALK3 and its inhibition were measured using the ADP-Glo Kinase Assay (Promega) to determine the enzyme autophosphorylation. The assays were performed in 384-well, low-volume microtiter assay plates in a final reaction volume of 6 μL. Dose–response curves were generated by incubating 10 nM of each kinase in 50 mM Hepes pH 7.5, 0.02% Tween 20, 0.02% BSA, 1 mM DTT, 10 μM Na3VO4, 10 mM β-Glycerolphosphate, 1 mM MgCl2, 12 mM MnCl2, and 15 μM ATP for 60 min at 32 °C in the presence or absence of compound diluted in DMSO. The amount of gen-erated ADP is a measure of kinase activity and is quantified using the ADP-Glo Kinase Assay according to the manufacturer’s in-structions. ADP is converted to ATP, which is subsequently con-verted into a bioluminescent signal by the luciferase assay reagents, which is measured using a PHERAstar Multilabel Reader. The luminescent signal positively correlates with kinase activity.

An additional kinase-selectivity panel, which measures substrate peptide phosphorylation, was set up for ALK1, ALK2, and ALK5. A fluorescein-labeled peptide substrate is phosphorylated by the respective kinase, which results in the introduction of two addi-tional negative charges. Because of this difference in charge, the phosphorylated and unphosphorylated peptides migrate with dif-ferent velocities in an electrical field. In the applied method, the separation takes place inside a chip that contains a complex capillary system for simultaneous analysis of 12 samples (“12-sipper chip”; Caliper Technologies). The separated labeled peptides can be quantified by fluorescence intensity through the instruments laser and detection system (LC3000; Caliper Life Sciences). The assays were performed in 384-well, low-volume microtiter assay plates in a final reaction volume of 9 μL. Dose–response curves were generated by incubating 10 nM of each kinase together with 2 μM of the fluorescently labeled substrate peptide 5-Fluo-Ahx-KKYQAEEN-T-YDEYENKK-amid (10 mM stock solution in DMSO) in 50 mM Hepes pH 7.5, 0.02% Tween 20, 0.02% BSA, 1 mM DTT, 10 μM Na3VO4, 10 mM β-Glycerolphosphate, 1 mM MgCl2, 12 mM MnCl2and 15 μM ATP for 60 min at 30 °C in the presence or absence of compound diluted in DMSO.

Kinase reactions were terminated by adding 15 μL STOP buffer (100 mM Hepes pH 7.5, 5% DMSO, 0.1% Caliper coating reagent, 10 mM EDTA, and 0.015% Brij35). Plates with terminated kinase reactions were transferred to the Caliper LC3000 workstation (Caliper Technologies) for reading. The relative amount of phos-phorylated peptide r, was calculated using the heights of the sub-strate peak, s, and the product peak, p: r = p/(p + s).

BMP Signaling Reporter Gene Assay.A human liver hepatocellular carcinoma cell line (HuH7) stably transfected with a reporter plasmid consisting of the BMP response element (BRE) from the Id1 promoter fused to a luciferase reporter gene was generated through lentiviral transduction. This cell line was used to de-termine the inhibitory effect of the compounds on BMP signaling. Cells were maintained in DMEM (high glucose plusL-Glutamine; GIBCO), 10% FCS (Amimed), 1% Pen/Strep (Amimed), and 5 μg/mL Blastidicin (InvivoGen) at 37 °C, 5% CO2. Assays were

performed in sterile 384-well flat bottom polystyrene microtiter plates. The cells were starved through medium exchange in Blasticidine- and FCS-free medium 16 h before the assay. Next, 2 × 104_{cells in a total volume of 40 μL were added to each well} of a plate already containing serial dilutions of each compound in DMSO (final DMSO concentration 0.5%). Cells and com-pound were incubated for 1 h at 37 °C, 5% CO2before stimu-lation with 5 μL per well recombinant BMP6 (R&D Systems) at a final concentration of 100 ng/mL. Assay plates are incubated for another 5 h at 37 °C, 5% CO2 before luciferase levels were quantified using the Steady-Glo Luciferase Assay System (Promega). Five microliters of the Steady-Glo Reagent were added to each well, the samples were mixed through vigorous shaking of the plate before measuring the luminescence in a PHERAstar Multilabel Reader.

TGF-β Signaling Reporter Gene Assay.Human HEK293 cell were transiently transfected with a reporter plasmid consisting of a TGF-β response element (CAGA12) from the PAI-1 promoter fused to a luciferase reporter gene. These cells were used to determine the inhibitory effect of the compounds on TGF-β signaling. Cells were maintained in DMEM (high glucose plus L-glutamine; GIBCO), 10% FCS (Amimed), and 1% Pen/Strp (Amimed) at 37 °C, 5% CO2. Assays were performed in sterile 384-well flat-bottom polystyrene microtiter plates. The cells were starved through medium exchange in FCS-free medium 16 h before the assay. Next, 2 × 104_{cells in a total volume of 40 μL} were added to each well of a plate already containing serial di-lutions of each compound in DMSO (final DMSO concentration 0.5%). Cells and compound were incubated for 1 h at 37 °C, 5% CO2before stimulation with 5 μL per well recombinant TGF-β2 (amino acids 303–414, dimer) at a final concentration of 9 ng/mL. Assay plates are incubated for another 5 h at 37 °C, 5% CO2 before luciferase levels were quantified using the Steady-Glo Luciferase Assay System (Promega # E2520). Five microliters of the Steady-Glo Reagent were added to each well, the samples were mixed through vigorous shaking of the plate, before mea-suring the luminescence in a PHERAstar Multilabel Reader. hMSC Isolation and Expansion. hMSCs were isolated from bone marrow aspirates and processed as previously described (1). Briefly, marrow aspirates (20 mL volume) were harvested from healthy donors (three males and one female, 17–49 y old) during routine orthopedic procedures involving exposure of the iliac crest. A bone-marrow biopsy needle was inserted through the cortical bone, and the aspirate was immediately transferred into plastic tubes containing 15,000 IU heparin. Isolated hMSCs were ex-panded in α-MEM containing 10% FBS, 4.5 mg/mLD-glucose, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 mM Hepes buffer, 100 Ul/mL penicillin, 100 μg/mL strepto-mycin, and 0.29 mg/mLL-glutamate, further supplemented with 5 ng/mL FGF2 (R&D Systems). Medium was changed twice a week and hMSCs were used at P2 for the following experiments. Western Blot Assay.hMSCs were treated with compounds A and B for 30 min prior stimulation with TGF-β3 alone or in combination with BMP-2 and BMP-4 for an additional 30 min. Cells were rinsed with fresh PBS and protein were extracted using lysis buffer containing 25 mM Tris·HCl, 150 mM NaCl, 1% Nonidet P-40, 1 mM EDTA, 5% glycerol, and supplemented with phos-phatase and protease inhibitor mixture (Roche). Protein con-centration was determined using BCA protein assay kit (Thermo

Occhetta et al.www.pnas.org/cgi/content/short/1720658115 1 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124

Scientific). Five micrograms of protein from each sample were subjected to 4–15% SDS/PAGE and then transferred onto PVDF membranes. After blocking (5% fat-free powdered milk in TBS with 0.1% tween for 1 h), membranes were probed with primary antibodies against phospho-Smad1/5/9, phosphor-Smad2/3, Smad1, and Smad2/3 (Cell Signaling). Anti-GAPDH antibody (Origen) was used as loading control. Blots were then exposed to appropriate peroxidase-coupled secondary antibodies (SouthernBiotech). Protein detection was done using Immobilon Western Chemiluminescent HRP Substrate (Millipore) and signal was acquired with Gel Doc XR+ Imager system (Bio-Rad). Microfluidic Platform Fabrication, Cell Seeding and Culturing, and on Chip Analyses.The microfluidic platform was fabricated as pre-viously described (2, 3). In detail, the microfluidic layouts (Fig. S2A) were designed by means of a CAD software (AutoCAD; Autodesk Inc.) and the corresponding master molds were realized through photolithography techniques (4). A thin microfluidic layer (1-mm thickness) was obtained by replica molding of poly-dimethylsiloxane (PDMS; Sylgard Dow Corning). Briefly, liquid PDMS was cast on the mold in ratio 10:1 [(wt/wt) prepolymer to curing agent], degassed and cured at 80 °C for 3 h. The obtained PDMS microstructured layer was permanently bonded facing up to a flat PDMS slab (5-mm thickness), upon 1 min of air plasma treatment (Harrick Plasma), thus closing the channels. Input and output ports were obtained through a 0.5-mm biopsy puncher (Harris Uni-CoreTM). The chip was finally permanently bonded to a glass-cover slide (50-mm diameter, 150-μm thickness; VWR International Inc.) upon an additional 1-min exposure to air plasma. External connections were realized through tygon tubing (ID = 0.02″; Qosina) and stainless steel couplers (23 gauge; Instech Laboratories). Upon sterilization by autoclaving (121 °C, 20 min, wet cycle) and drying overnight at 80 °C, devices were submerged in sterile PBS, degassed and filled with serum free medium (SFM). After two passage-expansions, hMSCs were di-luted to the concentration of 107 _{cells/mL in SFM, and seeded} into the culture units by perfusing 0.1 μL/min of cell suspension from the seeding channels (C1–C5). The outlet channels (D1–D5) were used as waste lines both during the seeding procedure. Once the culture chambers were filled with hMSCs, the seeding chan-nels were clamped and a continuous flow rate of 5 μL/h per line was applied at the primary inlets (A1–A2) throughout the entire experiment. Secondary inlets (B1–B2) were used at the beginning to facilitate the connection of the platform to the syringe pump used for perfusing the system, while closed during the culturing. hMSC-derived micromasses generated spontaneously in each culture chamber within 3 h from seeding; exploiting the integrated logarithmic serial-dilution generator, micromasses were then cul-tured under continuous perfusion with chondrogenic medium (SFM enriched with of 10–7 M dexamethasone, 10 ng/mL TGF-β3, and 10–4 M ascorbic acid 2-phosphate), while serial dilutions of compound A or B were established spanning over four orders of magnitude (0, 0.05, 0.5, 50, 500 nM). In detail, by perfusing chondrogenic medium from A1 and chondrogenic medium con-taining the highest concentration of compounds from A2, and setting the flow rate at A1 three times higher than the flow rate at A2, logarithmic dilutions were achieved (2). After 14 d in culture, micromasses were analyzed for protein content by performing im-munofluorescence directly inside the device. Furthermore, RNA was extracted by flowing RNA lysis buffer solution from dedicated input/output ports (E1–E5 and F1–F5, respectively) and gene-expression levels analyzed by quantitative RT-PCR.

Generation of MSC-Derived Constructs, in Vitro Culture, and in Vivo Implantation.hMSCs were expanded for two passages and then differentiated toward the chondrogenic lineage by either gener-ating scaffold-free 3D macropellet cultures (0.25 × 106_{cells per} pellet) or seeding them onto type I collagen meshes (Ultrafoam;

Davol) at a density of 7 × 106_cells/cm3_{. Constructs were cultured} for 2 or 4 wk in chondrogenic medium, supplemented with com-pound A (500 nM), comcom-pound B (250, 500, or 1,000 nM) or DMSO as control. For selected samples, 2 wk of chondrogenic differentiation were followed by 2 wk of hypertrophic differentia-tion by adding serum-free hypertrophic medium, supplemented with 50 nM thyroxine, 7 mM β-glycerophosphate, 10 nM dexa-methasone, 0.25 mM ascorbic acid, and IL-1β (50 pg/mL) (5). After 2 wk of pretreatment in vitro, samples were implanted in the sub-cutaneous pouches of nude mice (CD-1 nude/nude; Charles River Laboratories, four samples per mouse) and retrieved after 8 wk. Real-Time RT-PCR Quantitation of Transcript Levels.Total RNA ex-traction, cDNA synthesis and real-time reverse transcriptase-PCR (RT-PCR; 7300 AB Applied Biosystem) were performed to quan-titate expression levels of the following genes of interest (Applied Biosystems): GREM1 (Hs01879841_s1), NOG (Hs00271352_s1), BMP2 (Hs00154192_m1), PRG4 (Hs00981633_m1), FRZB (Hs00173503_ m1), GDF5 (Hs00167060_m1), COL2A1 (Hs00264051_m1), COL10A1 (Hs00166657_m1), IHH (Hs01081800_m1), MMP13 (Hs00233992_m1), VEGF (Hs00900055_m1), ACAN (Hs00153936_m1), LIF (Hs01055668_m1), ALK1 (Hs00163543_m1), ALK2 (Hs00153836_m1), ALK3 (Hs01034913_g1), MSX2 (Hs00741177_m1).

The housekeeping gene GAPDH was used as reference (Hs02758991_g1).

Histological Staining and Immunohistochemistry.After in vitro and in vivo cultures, the constructs were fixed in 4% paraformaldehyde, if necessary decalcified with 15% EDTA solution (Sigma) and embedded in paraffin. Sections (5-μm thick) were stained for Safranin-O (Fluka), Toluidine Blue, H&E (Baker), Masson’s Trichrome (RAL Diagnostic), and Alizarin red. Immunohisto-chemical analyses were performed to characterize the extracellu-lar matrix over the time using the following antibodies: type II (Col II; MPBiomedicals) and type X (Col X; AbCam) collagens, MMP13 (AbCam), Chondromodulin (AbCam). Upon rehydration in ethanol series, sections were treated as previously described for antigen retrieval (6), or according to the manufacturer’s instruc-tions. The immunobinding was detected with biotinylated sec-ondary antibodies and using the appropriate Vectastain ABC kits. The red signal was developed with the Fast Red kit (Dako Cytomation) and sections counterstained by Hematoxylin. Nega-tive controls were performed during each analysis by omitting the primary antibodies. Histological and immunohistochemical sec-tions were analyzed using an Olympus BX-61 microscope. Immunofluorescence Images. Following implantation, in vivo samples were fixed in 4% paraformaldehyde (Sigma), decalcified with 15% EDTA (Sigma) solution and embedded in paraffin. Sections (5-μm thick) were incubated with the primary antibodies against Chondromodulin (Abcam), CD31 (PECAM-1; BD Pharmingen), and αSMA (BD Pharmingen). As appropriate, secondary antibodies labeled with Alexa Fluo 488, Alexa Fluo 546, and Alexa Fluo 647 (Invitrogen) were used and DAPI was used to stain nuclei. Confocal fluorescence images were acquired using a Nikon A1R Nala Con-focal microscope (Nikon).

Quantification of Secreted Proteins by ELISA.During culture, cell supernatants were collected after 7, 14, and 28 d of incubation and analyzed for their content of BMP-2 and VEGF, according to the ELISA manufacturer’s instructions (R&D Systems).

Statistical Analysis. The results of quantitative RT-PCR and ELISA analyses are presented as mean ± SD (n = 3). Statistical analysis was performed using the unpaired Student’s t test (normality test P > 0.05). P values of 0.05 or less were considered as statistically significant.

Occhetta et al.www.pnas.org/cgi/content/short/1720658115 2 of 9

125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248

1. Martin I, Muraglia A, Campanile G, Cancedda R, Quarto R (1997) Fibroblast growth factor-2 supports ex vivo expansion and maintenance of osteogenic precursors from human bone marrow. Endocrinology 138:4456–4462.

2. Occhetta P, et al. (2015) High-throughput microfluidic platform for 3D cultures of mes-enchymal stem cells, towards engineering developmental processes. Sci Rep 5:10288. 3. Occhetta P, Visone R, Rasponi M (2017) High-throughput microfluidic platform for 3D

cultures of mesenchymal stem cells. 3D Cell Culture: Methods and Protocols, ed Koledova Z (Springer, New York), pp 303–323.

4. McDonald JC, Whitesides GM (2002) Poly(dimethylsiloxane) as a material for fabri-cating microfluidic devices. Acc Chem Res 35:491–499.

5. Scotti C, et al. (2010) Recapitulation of endochondral bone formation using human adult mesenchymal stem cells as a paradigm for developmental engineering. Proc Natl Acad Sci USA 107:7251–7256.

6. Dickhut A, et al. (2009) Calcification or dedifferentiation: Requirement to lock mesenchymal stem cells in a desired differentiation stage. J Cell Physiol 219: 219–226.

Fig. S1. The expression of ALK2 and ALK3 in P0 adult human MSCs in vitro expanded cells was evaluated by quantitative RT-PCR and immunohistochemical staining. (A) The expression of ALK1, ALK2, and ALK3 in P0 in vitro-expanded adult human MSCs was evaluated by quantitative RT-PCR. Immunohistochemistry confirmed that both (B) ALK2 and (C) ALK3 are expressed in P0 hMSCs at a protein level. (D) A negative staining control was applied without primary antibody. (E and I) ALK2 and ALK3 gene expression was monitored after 14 d of treatment with either compound in 3D pellet culture. All ΔΔCt values are normalized relative to GAPDH expression and refer to basal expression in day 0 hMSCs; all fold-changes in transcript levels are shown in logarithmic scale

Q:1 (n = 3, n.s.d.). (F–

H) ALK2 and (J–L) ALK3 protein expressions were not modulated by 14 d of treatment with compound A, compound B, or standard chondrogenic medium in 3D pellet culture. All of the images were taken at the same magnification. (Scale bars, 50 μm.)

Occhetta et al.www.pnas.org/cgi/content/short/1720658115 3 of 9

249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372

Fig. S2. (A) Schematic representation of the microfluidic platform integrating a logarithmic serial dilution generator (blue), a 3D culture area for generation and culture of 3D hMSC-derived micromasses (red), and a system of bypass channels for RNA extraction (green). The operational procedure is briefly described. Cell suspension is injected into the culture units from the seeding lines (C1–C5), while using the outlets (D1–D5) as waste lines. Upon spontaneous generation of hMSC-derived micromasses in each culture chamber (about 3 h from seeding), both seeding channels (C1–C5) and secondary inlets are clamped (B1–B2), and a continuous flow rate is applied at the main inlets (A1–A2) throughout the entire experiment. Controlled concentrations of compounds, spanning over four orders-of-magnitude, are administered to cells by exploiting the logarithmic serial dilution generator (i.e., chondrogenic medium is perfused in A1 and chondrogenic medium containing the highest concentration of compound through A2). Furthermore, inlets E1–E5 and outlets F1–F5 allow flowing RNA lysis buffer solution and collecting RNA, respectively, for subsequent analyses of gene-expression levels through quantitative RT-PCR. (B) Taking advantages of this platform, different temporal windows of treatment were investigated for compound A at 500 nM, maintaining a total in vitro culture time of 14 d. Every condition was tested with or without TGF-β3 perfusion during the time of the compound supplementation. Immunofluorescence analysis showed a compa-rable deposition of collagen type-II (green) in all temporal windows, except for the 14-d treatment without TGF-β. TGF-β is indeed known to be essential for hMSC chondrogenic differentiation (1, 2), and absence of TGF-β during the whole culture period of 14 d led, as expected, to a lack of chondrogenesis. Collagen type X (red) staining was detected in the control and samples only treated for the initial 2 d, while absent when the treatment was provided continuously for 14 d or in the last 2 d. All pictures were taken at the same magnification. (Scale bar, 75 μm.) (C) Quantitative RT-PCR analysis indicated the maintenance of a low MSX2 expression status both at the beginning (day 2) and at the end (day 14) of the culture only for the 14-d continuous treatment. Msx2 is a tran-scriptional factors downstream the BMPs pathway. All ΔΔCt values are normalized relative to GAPDH expression and refer to basal expression in day 0 hMSCs; all fold-changes in transcript levels are shown in logarithmic scale (n = 3, n.s.d.).

1. Johnston B, Hering TM, Caplan AI, Goldberg VM, Yoo JU (1998) In vitro chondrogenesis of bone marrow-derived mesenchymal progenitor cells. Exp Cell Res 238:265–272. 2. Tuli R, et al. (2003) Transforming growth factor-beta-mediated chondrogenesis of human mesenchymal progenitor cells involves N-cadherin and mitogen-activated protein kinase and

Wnt signaling cross-talk. J Biol Chem 278:41227–41236.

Occhetta et al.www.pnas.org/cgi/content/short/1720658115 4 of 9

373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496

Fig. S3. In vitro maturation of 3D macropellets derived from adult human MSCs and cultured in chondrogenic medium, either (D, H, and M) alone or sup-plemented with (A, E, and I) compound A or (B–D, F, J, and L) compound B

Q:2 . In vitro culture conditions determined the composition of tissues generated after

treatment for 14 d (A–D), 28 d (E–H), or after treatment for 14 d followed by 14 d in standard chondrogenic medium (“recovery”) (I–L). Samples were his-tologically analyzed for Glycosaminoglycan, collagen type II, and collagen type X deposition. Samples were hishis-tologically analyzed for Glycosaminoglycan, collagen type II, and collagen type X deposition. All of the images were taken at the same magnification. (Scale bars, 300 μm.)

Occhetta et al.www.pnas.org/cgi/content/short/1720658115 5 of 9

497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620