Commitment of human mesenchymal stromal cells to skeletal lineages is independent of their morphogenetic capacity

Abstract

BACKGROUND

Mesenchymal stromal cells (MSCs) are multipotent cell populations obtained from fetal and adult tissues. They share some characteristics with limb bud mesodermal cells such as differentiation potential into osteogenic, chondrogenic, and tenogenic lineages and an embryonic mesodermal origin. Although MSCs differentiate into skeletal-related lineages in vitro, they have not been shown to self-organize into complex skeletal structures or connective tissues, as in the limb. In this work, we demonstrate that the expression of molecular markers to commit MSCs to skeletal lineages is not sufficient to generate skeletal elements in vivo.

AIM

To evaluate the potential of MSCs to differentiate into skeletal lineages and generate complex skeletal structures using the recombinant limb (RL) system.

METHODS

We used the experimental system of RLs from dissociated-reaggregated human placenta (PL) and umbilical cord blood (UCB) MSCs. After being harvested and reaggregated in a pellet, cultured cells were introduced into an ectodermal cover obtained from an early chicken limb bud. Next, this filled ectoderm was grafted into the back of a donor chick embryo. Under these conditions, the cells received and responded to the ectoderm’s embryonic signals in a spatiotemporal manner to differentiate and pattern into skeletal elements. Their response to differentiation and morphogenetic signals was evaluated by quantitative polymerase chain reaction, histology, immunofluorescence, scanning electron microscopy, and in situ hybridization.

RESULTS

We found that human PL-MSCs and UCB-MSCs constituting the RLs expressed chondrogenic, osteogenic, and tenogenic molecular markers while differentially committing into limb lineages but could not generate complex structures in vivo. MSCs-RL from PL or UCB were committed early to chondrogenic lineage. Nevertheless, the UCB-RL osteogenic commitment was favored, although preferentially to a tenogenic cell fate. These findings suggest that the commitment of MSCs to differentiate into skeletal lineages differs according to the source and is independent of their capacity to generate skeletal elements or connective tissue in vivo. Our results suggest that the failure to form skeletal structures may be due to the intrinsic characteristics of MSCs. Thus, it is necessary to thoroughly evaluate the biological aspects of MSCs and how they respond to morphogenetic signals in an in vivo context.

CONCLUSION

PL-MSCs and UCB-MSCs express molecular markers of differentiation into skeletal lineages, but they are not sufficient to generate complex skeletal structures in vivo.

Key Words: Human mesenchymal stromal cells; Recombinant limbs; Mesenchymal stromal cell morphogenesis; Mesenchymal stromal cell in vivo differentiation; Skeletal tissues

Core Tip: Human mesenchymal stromal cells (MSCs) from umbilical cord blood or placenta can differentiate into osteogenic and chondrogenic lineages in culture systems and have been used in regenerative medicine. Here, we used the recombinant limb (RL) model to provide evidence that MSCs do not have the ability to generate skeletal structures in vivo. MSCs received and responded to the ectoderm’s embryonic spatiotemporal signals in this RL system. However, the expression of differentiation markers of skeletal lineages was not sufficient to generate skeletal structures in vivo.

INTRODUCTION

The limbs have an intricate anatomy that results from highly coordinated morphogenesis and cellular differentiation processes, which establish the correct position and shape of muscles, bones, cartilage, tendons, skin, and nerves[1]. During embryonic development, limb bud mesodermal cells (LBMCs) form a bud covered by a layer of ectodermal cells. The morphogenesis and differentiation of limb tissues are controlled by the response to signals that emanate from three signaling centers: The apical ectodermal ridge (AER), which controls the proximodistal axis; the zone of polarizing activity, which directs the anteroposterior axis; and the ventral and dorsal LB ectoderms, which control the dorsal-ventral axis[1,2]. Fibroblast growth factor 8 (FGF8) and Wnt family member 3a signals emanate from the AER and maintain the MCs in an undifferentiated and proliferative state. Once cells stop receiving AER signals, they begin the differentiation process toward the chondrogenic and tenogenic lineages, followed by endochondral ossification to form the skeletal elements[1,2]. As development progresses, the coordination between signals is essential for patterning and differentiation to fully develop the skeletal elements of a functional limb[1,2].

SRY-box transcription factor 9 (SOX9) is the earliest molecular marker of chondrogenic commitment[3]. Besides, SOX9 induces the expression of extracellular matrix proteins such as collagen type 2 (Col2α) and aggrecan (ACAN)[4]. On the other hand, growth differentiation factor 5 (GDF5) promotes cell condensation and cell adhesion during the initial stages of chondrogenesis[5,6]. By contrast, hypoxia-inducible factor alpha (HIF-1α) is critical in maintaining the chondrogenic environment[7]. Once cartilage anlagen is formed, ossification begins by expressing Runt-related transcription factor 2 (RUNX2)[8-10]. At the same time, the dorsal and ventral margins of early embryonic LB initiate tendon specification by expressing the earliest known marker of tenocyte progenitors, scleraxis (SCX)[11,12]. SCX induces the transcription of major structural collagens and the glycoprotein tenomodulin[13]. Mohawk (MKX) is involved in tendon maturation by maintaining SCX expression, which restricts cells to the tenogenic lineage[13,14].

Regenerative medicine has emerged as an alternative treatment for degenerative diseases or an adjuvant therapy[15,16]. To this end, the potential use of mesenchymal stromal cells (MSCs) has garnered great interest. Because MSCs share characteristics with LBMCs and have the potential to differentiate into osteogenic or chondrogenic lineages[17-21], in this work, we used the recombinant limb (RL) model to evaluate the capacity of human MSCs to form complex skeletal structures[22]. The RL model recapitulates the differentiation, morphogenesis, and patterning programs that occur during normal limb development[23-25]. RLs consist of an ectodermal cover obtained from an early chicken LB (CK-LB) filled with dissociated-reaggregated cells grafted into the back of a donor chick embryo. Because of the high capacity of human MSCs from umbilical cord blood (UCB-MSCs) or placenta (PL-MSCs) to differentiate into osteogenic and chondrogenic lineages, we evaluated their ability to generate skeletal structures in vivo. Our results demonstrate that although MSCs receive and respond to the ectoderm’s embryonic spatiotemporal signals, they do not generate complex skeletal structures, possibly due to the intrinsic characteristics of MSCs.

MATERIALS AND METHODS

Isolation and culture of MSCs

MSCs were collected from Villa Coapa Hospital, Mexican Social Security Institute (IMSS; Mexico City, Mexico) according to their ethical guidelines, including informed consent. MSCs derived from bone marrow (BM; n = 3) were used to validate the specificity of all primers used in this study. BM cells were obtained from hematologically normal BM transplantation child donors at the Bernardo Sepulveda Hospital (National Medical Center, IMSS). A cell population enriched for MSCs was isolated according to Montesinos et al[25] using a negative selection procedure [RosetteSep^TM system; STEMCELL Technologies Inc. (STI), Vancouver, Canada] and a Ficoll gradient. The cells were resuspended in low-glucose Dulbecco’s modified Eagle medium (Lg-DMEM; Gibco BRL, Rockville, MD, United States) supplemented with 10% fetal bovine serum (FBS) and seeded at a density of 0.2 × 10⁶ cells/cm² into T25 cell culture flasks (Corning Inc., Costar, New York, NY, United States). After 4 d, the nonadherent cells were removed by pipetting, and fresh medium was added. Every 5 d, a medium change was performed. When the cultures reached 80% confluence, they were trypsinized (0.05% trypsin, 0.53 mmol/L EDTA; Gibco BRL, New York, NY, United States) and subcultured at a density of 0.01 × 10⁶ cells/cm² into T75 flasks (Corning). In the second passage, cells were harvested and analyzed.

UCB-MSC and PL-MSC samples were obtained from two volunteer donors from normal full-term deliveries according to the institutional guidelines of Troncoso Hospital, IMSS, as previously described by Montesinos et al[25]. UCB-derived MSCs were obtained using a negative selection procedure (RosetteSep^TM system; STI), as described for BM. Cells were resuspended in Lg-DMEM (Gibco) supplemented with 10% FBS (STI) and seeded at a density of 0.2 × 10⁶ cells/cm² in T25 culture flasks (corning), and subsequent cultures were manipulated as described for BM. PL-derived MSCs were obtained using an enzymatic digestion procedure. The internal area (about 2 cm³) of the central PL lobules was washed with phosphate-buffered saline (PBS) and dissected into small pieces with sterilized scissors and forceps. The chopped tissues were digested with trypsin-EDTA (Gibco) at 37 °C for 10 min. A single-cell suspension was collected by flushing the tissue parts through a 100-μm nylon filter (Falcon; Becton, Dickinson and Company, San Jose, CA, United States) with Lg-DMEM containing 10% FBS. The cell suspension was centrifuged for 10 min at 400 g to collect the cell pellet. Then, the cells were resuspended in Lg-DMEM supplemented with 10% FBS. The total number of nucleated and viable cells was determined, seeded, and manipulated as described for BM. After the third passage, the cells were analyzed and cryopreserved. For all experiments, MSCs were used in passages 4-5.

RLs

RLs were established according to the protocol by Marín-Llera et al[24]. PL-MSCs, UCB-MSCs, or 22 Hamburger-Hamilton (HH) CK-LBMCs were collected to stuff 22 HH CK limb ectoderms, according to Hamburger and Hamilton[26]. After obtaining PL-MSCs, UCB-MSCs, or CK-LBMCs, they were centrifuged at 3000 rpm and incubated at 37 °C between 1.5 h and 2 h to form a compact pellet. Separately, to obtain ectoderms, LBs from about 10 of the 22 HH embryos were dissected in PBS, transferred to a tube, and digested in 0.5% trypsin in PBS for 30 min at 37 °C. Then, the LBs were transferred to PBS supplemented with 10% FBS, and the ectoderm was peeled off. Next, the pellet of PL-MSCs, UCB-MSCs, or CK-LBMCs was detached from the bottom of the tube and transferred to a small petri dish containing the ectoderms. Later, a pellet fragment was stuffed into each ectoderm. Ectoderms were individually transferred into a previously windowed 22 HH chick embryo, positioned between somites 15-20, and fixed with palladium wires over a previously scratched wound. Manipulated embryos were incubated for 24 h, 48 h, or 72 h at 37.5 °C until their collection and processing. The leftover pellet from three sources was independently frozen to evaluate its basal expression compared to RLs by quantitative polymerase chain reaction (qPCR). All RL experiments were performed in triplicate.

qPCR

RNA from RLs was extracted with NucleoSpin RNA (Cat. No. 740955; Macherey-Nagel, Düren, Germany) according to the manufacturer’s instructions. Retrotranscription of total RNA was achieved using the RevertAid RT Kit (Cat. K1691; Thermo Fisher Scientific, Waltham, MA, United States). Gene expression levels were quantified using SYBR green (Cat. No. 4309155; Thermo Fisher Scientific), normalizing to the transcript level of ribosomal protein L13. The expression level was evaluated relative to a calibrator according to the 2^-ΔΔCt equation. Each plotted value represents the mean ± SD of at least three independent experiments. For each independent qPCR replicate, RNA was obtained from 25 to 30 RL pools for each condition. In addition, leftover pellets of MSCs were collected and processed for RNA extraction. The student’s t-test was performed to compare the expression levels between pellets and RLs and to determine the statistical significance among RL sources. Statistical significance was set at P < 0.05. To analyze the specificity of the amplification, a melting curve was achieved for each primer. The specificity of all primers was validated with human MSCs-BM before amplification in the RL. The primer sequences used in this study are included in Supplementary Table 1.

Alcian blue and hematoxylin staining

For Alcian blue staining of RLs, samples were fixed in 5% trichloroacetic acid (Cat. No. A-5268; Sigma-Aldrich, St. Louis, MO, United States) for 24 h and stained with 1% Alcian blue in ethanol-hydrogen chloride for 24 h. Then, the RLs were transferred to 100% ethanol for 24 h and cleared with methyl salicylate (Cat. No. M-2047; Sigma-Aldrich) until the skeleton was observed. RL images were acquired with the AxioZoom v16 microscope (Carl Zeiss, Oberkochen, Germany) using ZEN lite software (Carl Zeiss). After image acquisition, Alcian blue-stained RLs were dehydrated in ethanol and xylol before embedding in paraffin. Ten-micrometer sections were obtained with a microtome (RM2125 RTS; Leica, Wetzlar, Germany). For hematoxylin and eosin (H&E) staining, samples were rehydrated with an ascending gradient of xylol-ethanol, followed by incubation with H&E dyes. Next, slides were dehydrated with ethanol-xylol and mounted with DPX medium (Cat. 44581; Sigma-Aldrich). Images were acquired with the Olympus BX51-WI microscope equipped with a fluorescence and gyratory disc unit (Tokyo, Japan) using Stereo Investigator 9 software (MicroBrightField Inc., Colchester, VT, United States).

Detection of cell death with LysoTracker staining

RLs were incubated in 1 μM LysoTracker Red DND-99 (Cat. L7528; Molecular Probes, Eugene, OR, United States) at 37 °C for 15 min. Then, the samples were rinsed with PBS and fixed in 4% paraformaldehyde overnight at 4 °C. Next, the RLs were dehydrated in an increasing methanol-PBS-Tween series and cleared with 2:1 benzylic alcohol-benzyl benzoate solution for 1 h (following the 1999 recommendations by Parish). Images were acquired with an Olympus BX51-WI microscope equipped with a fluorescence and gyratory disc unit using Stereo Investigator 9 software (MicroBrightField).

Scanning electron microscopy

RLs were washed twice with PBS and fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH = 7.2) for 2 h. The fixation solution was washed three times with 0.1 M sodium cacodylate at 4 °C. For postfixation, samples were immersed in 1% osmium tetroxide in cacodylate solution for 2 h before dehydration in graded concentrations of ethanol. The samples were placed in microporous capsules submerged in pure ethanol and placed in the Critical Point drying equipment (SPI) for desiccation. Ethanol was replaced with liquid carbon dioxide until its critical constants (31.3 °C and 1072 PSI) were met. After drying, the sample was mounted on an aluminum specimen holder with a silver-based adhesive. The coating was performed by ionizing the sample with a layer of gold in a low vacuum ionizer. Images were acquired using the DSM-950 Carl Zeiss microscope.

In situ hybridization

RNA antisense probes were labeled with UTP-digoxigenin (Cat. No. 11209256910; Roche Applied Science, Indianapolis, IN, United States) and used for whole-mount in situ hybridization as previously described[6]. Samples were treated with 15 μg/mL proteinase K for 20 min at 21 °C, and the hybridization temperature was 68 °C. The Fgf8 signal was visualized with BM Purple substrate for alkaline phosphatase (Roche). Images were acquired with the AxioZoom v16 microscope (Carl Zeiss).

RESULTS

RLs serve as a model to evaluate the morphogenesis and differentiation processes of human MSCs

We used RLs as an experimental model to evaluate the capacity of PL-MSCs and UCB-MSCs to differentiate into limb tissues and to determine their capacity to generate well-organized tissues in an in vivo context. In the first instance, we evaluated the competence of human MSCs to respond to embryonic signals present in the limb ectoderm (Figure 1). The results showed that 24 h postimplantation (hpi), the PL-MSCs and UCB-MSCs were successfully integrated and formed RLs, with an efficiency of 62.95% and 76.87%, respectively (Figure 1A). Next, we measured the expression and distribution of major histocompatibility complex 1 (HLA1) to determine whether the RLs were exclusively formed by MSCs. We observed that both MSCs-RLs expressed significantly higher levels of HLA1 compared with their own pellet before transplantation. In addition, the cells in the center of the RLs were HLA1⁺ (Figure 1B). These results demonstrated the successful establishment of RLs composed of human MSCs for the evaluation of MSC behavior in vivo.

Figure 1 Validation of the recombinant limb system using human mesenchymal stromal cells. A: Implantation efficiency of umbilical cord blood (UCB)-recombinant limb (RLs) and placenta (PL)-RLs in chicken embryos after 24 h. Scanning electron microscopy of PL-RLs (n = 3) and UCB-RLs (n = 2) at 24 h postimplantation (hpi). The embryonic ectoderm and apical ectodermal ridge were conserved in RLs from both sources; B: Levels of human leukocyte antigen (Hla) detected by quantitative polymerase chain reaction at 24 hpi RLs from PL and UCB relative to its own pellet. Data represent three independent experiments. The major histocompatibility complex (HLA) expression pattern is shown in RLs from PL (n = 3) and UCB (n = 2). HLA expression was observed in the center of the RLs, indicating that mesenchymal stromal cells constituted the RLs. Statistical significance was set as follows: ^aP < 0.05; ^cP < 0.0005. UCB: Umbilical cord blood; PL: Placenta; RL: Recombinant limb; HLA: Human leukocyte antigen.

Morphology of MSCs changes as RL develops

To assess the capacity of human MSCs to form well-organized tissues, the phenotype of MSC-RLs and their cell organization were evaluated at 24 hpi, 48 hpi, and 72 hpi and compared with RLs formed with CK-LBMCs as a reference of the morphogenetic behavior under RL conditions. In contrast to CK-RL, no evidence of central skeletal elements was observed in RLs from either MSC source (Figure 2). Histological analyses revealed that the ectodermal cover in all RL conditions was maintained as a monostratified epithelium, while cellular condensation in the center of the MSCs-RL was observed at all time points (Figure 2). Interestingly, the morphology of cell condensations in MSCs-RLs from both sources changed as the development of the RLs progressed, but UCB-RLs showed more evident progressive elongation of a central structure than PL-RLs. Although Alcian blue staining around the condensed cells was observed in MSCs-RLs from both sources, cell morphology analyses showed no cartilage anlage formation after 72 hpi as observed in CK-RLs (Figure 2). The lack of well-formed skeletal elements and cartilage morphology in MSCs-RLs suggests that MSCs cannot form complex skeletal structures because they can remain undifferentiated or in an early committed state. These results also suggest that PL-MSCs and UCB-MSCs respond differently to embryonic ectoderm signals.

Figure 2 Phenotypic and histological characterization of the mesenchymal stromal cells-recombinant limbs. Alcian blue and hematoxylin and eosin staining of recombinant limbs from the placenta (n = 8), umbilical cord blood (n = 7), and chicken (n = 4) at 24, 48, and 72 h postimplantation. hpi: Hour postimplantation; UCB: Umbilical cord blood; PL: Placenta; RL: Recombinant limb; CK: Chicken; H&E: Hematoxylin and eosin.

Molecular and in situ characterization of cell differentiation in MSC-RLs

In contrast to CK-RL, histological analyses of MSCs-RLs did not indicate specific tissue formation at any evaluated time. Thus, to determine if the PL-MSCs and UCB-MSCs committed to limb cellular lineages during MSCs-RL formation, we evaluated the expression profile of chondrogenic (Sox9, Gdf5, Col2a1, Acan, and Hif1a), osteogenic (Runx2), and tenogenic (Scx and Mkx) differentiation genes at 24 hpi in MSCs-RLs from both sources (Figure 3A). The results confirmed that MSCs commit to limb lineages but respond differentially to embryonic ectodermal signals as the expression of molecular markers of chondrogenic, osteogenic, and tenogenic lineages was induced at different levels in MSCs-RLs from both sources compared with its initial pellet (Table 1).

Figure 3 Mesenchymal stromal cells differentially commit to chondrogenic, osteogenic, and tenogenic lineages in the recombinant limb system. A: Quantitative polymerase chain reaction analysis of cartilage (a-e), bone (f), and tendon (g and h) cell fate-regulating genes of the mesenchymal stromal cells-recombinant limbs (MSCs-RLs) at 24 h postimplantation (hpi) relative to its own pellet set to 1.0. Statistical significance comparing MSCs-RLs from the two sources is shown; B: Protein expression pattern of master genes of the three differentiation lineages in MSCs-RLs at 24 hpi. Data represent three independent experiments. Statistical significance was set as follows: ^aP < 0.05; ^bP < 0.005; ^cP < 0.0005; ^dP < 0.0001. For more details see Table 1. Sox9: SRY-box transcription factor 9; Gdf5: Growth differentiation 5; Col2α1: Collagen type 2 alpha 1; Acan: Aggecan; Hif1α: Hypoxia-inducible factor 1 alpha; Runx2: Runt-related transcription factor 2; Scx: Scleraxis; Mkx: Mohawk; UCB: Umbilical cord blood; RL: Recombinant limb; PL: Placenta; NS: Not significant.

Table 1 Analysis of gene expression fold-changes of differentiation genes between pellets and their corresponding recombinant limbs of different mesenchymal stromal cell sources at 24 h postimplantation.

Source	Sox9	Col2a1	Acan	Gdf5	Hif1α	Runx2	Scx	Mkx
UCB-RL vs UCB-pellet	3.99 ± 0.4, P < 0.0004	ND	3.58 ± 0.10, P < 0.0001	0.01 ± 00.003, P < 0.0001	1.58 ± 0.79, P < 0.0002	9.96 ± 0.65 P < 0.0001	377.61 ± 78.24, P < 0.002	0.24 ± 0.03, P < 0.0001
PL-RL vs PL-pellet	3.79 ± 0.4, P < 0.0007	0.61 ± 0.12, P < 0.002	0.55 ± 0.21, P < 0.052	0.56 ± 0.12, P < 0.01	1.61 ± 0.28, P < 0.03	2.18 ± 0.59, P < 0.03	0.61 ± 0.33, P < 0.14	1.29 ± 0.22, P < 0.21

Gene expression fold-change was evaluated comparing the pellet of mesenchymal stromal cells (MSCs) after forming the recombinant limb (RL) with MSCs-RLs from both sources. Data from the pellet were set to 1.0. Statistical significance comparing placenta-RL vs umbilical cord blood-RL is shown in Figure 3A. All data are presented as mean ± SD. Statistical significance was set at P < 0.05. Acan: Aggecan; Col2α1: Collagen type 2 alpha 1; Gdf5: Growth differentiation 5; Hif1α: Hypoxia-inducible factor 1 alpha; Mkx: Mohawk; ND: Not detected; Scx: Scleraxis; Sox9: SRY-box transcription factor 9; Runx2: Runt-related transcription factor 2; UCB: Umbilical cord blood; RL: Recombinant limb; PL: Placenta.

Comparative analyses of Sox9 expression levels between MSCs-RLs showed no significant differences (P < 0.8421), but remarkably, it was highly induced in both MSCs after receiving ectodermal signals (Figure 3A, Table 1). By contrast, low Gdf5 expression was observed in both sources compared with its initial levels. In addition, it was significatively reduced in UCB-RLs compared with PL-RLs (P < 0.0205; Figure 3A, Table 1). Regarding later chondrogenic-associated genes, expression of Col2a1 was not detected in UCB-RLs, whereas its expression was significatively reduced in PL-RLs (Figure 3A, Table 1). On the other hand, Acan was overexpressed only in UCB-RLs and was significatively reduced in PL-RLs (P < 0.0001; Figure 3A, Table 1). Interestingly, the levels of Hif1a observed in both MSCs-RLs suggest that a hypoxic environment is favored under these conditions (Figure 3A, Table 1). Regarding osteogenic commitment, Runx2 was positively regulated in both PL-RLs and UCB-RLs; however, the expression levels of this gene in UCB-RLs were significatively higher than those observed in PL-RLs (P < 0.0001; Figure 3A, Table 1). The tenogenic marker Scx was markedly induced in UCB-RLs compared with PL-RLs (P < 0.0004), in which expression was inhibited (Figure 3A, Table 1). By contrast, Mkx expression was significantly downregulated in UCB-RLs compared with PL-RLs (P < 0.0002; Figure 3A, Table 1). To evaluate the spatial organization of committed cells, the protein distribution of SOX9, SCX, and RUNX2 was evaluated in MSC-RLs (Figure 3B). Importantly, the high protein expression was correlated with gene expression in all RLs, as assessed by qPCR (Figure 3B, Table 1). However, these results demonstrated that the localization of SOX9, SCX, and RUNX2 was not restricted to the center of the MSCs-RLs (Figure 3B). Together, these results suggest that at 24 hpi, cells in MSCs-RLs are committed early to the chondrogenic lineage. UCB-RLs can also commit to osteogenic lineage, although preferentially to a tenogenic cell fate, but this cell fate is not promoted in PL-RLs.

Evaluation of Fgf-8 expression and cell death in MSCs-RLs

Signals from the ectoderm and AER are indispensable for maintaining cell survival and proliferation during limb development and RL formation[23]. During limb development, the undifferentiated cellular state and cell commitment balance depend on signals from the AER[2]. To determine whether the lack of ectodermal signals or massive cell death was the mechanism underlying the inability of MSC to form skeletal structures, ectoderm integrity by Fgf8 expression and lysotracker staining were evaluated in RLs (Figure 4). The results showed that the cell death process was present in the ectoderm of all evaluated RLs; UCB-RLs showed a higher number of dead cells than PL-RLs and CK-RLs. Interestingly, in PL-RLs and CK-RLs, dead cells were concentrated in the apical area of the ectoderm, whereas in UCB-RLs, dead cells were distributed homogeneously on the ectoderm of all samples evaluated (Figure 4). Despite the death of ectodermal cells, Fgf8 expression was maintained in some MSCs-RLs (see details in the figure legend) or its expression in PL-RLs was disrupted in the center of the ectoderm (Figure 4). These results suggest that at 24 hpi, signals continue to emanate from the AER, which are received by the MSCs or CK-LBMCs beneath the AER.

Figure 4 Cell death and fibroblast growth factor 8 expression pattern in mesenchymal stromal cells and chicken recombinant limbs at 24 h postimplantation. Lysotracker staining (Ltk) to evaluate cell death in recombinant limbs (RLs) from the placenta (PL-RLs; n = 6), umbilical cord blood (UCB-RLs; n = 6), and chicken (CK-RLs; n = 4) is shown. In situ hybridization of fibroblast growth factor 8 (Fgf8) in PL-RLs showed disrupted pattern of this gene in 2/4 RLs; Fgf8 expression in UCB-RLs was detected in 3/6 evaluated samples, while 5/5 CK-RLs continued to express Fgf-8 at 24 h postimplantation. UCB: Umbilical cord blood; PL: Placenta; RL: Recombinant limb; CK: Chicken; Ltk: Lysotracker staining; Fgf8: Fibroblast growth factor 8.

DISCUSSION

Cells commit to specific lineages once the expression of master genes occurs in response to inducing factors. However, the ability to interpret signals and acquire a particular fate and behavior depend on the developmental history of the cells. The capacity to organize complex structures depends on cell differentiation, cell recruitment, cell movement, differential cell proliferation, and cell death, which occur during the morphogenetic process of tissues and organs.

In this study, we used the RL model to evaluate the morphogenetic abilities of MSCs other than those observed in cell culture. Because RLs provide limb spatial-temporal signals, we evaluated the ability of MSCs to generate complex skeletal structures[23-25]. We found that human MSCs can integrate into an RL implanted in the CK embryo. PL- or UCB-MSCs from RLs respond differently to ectodermal signals, suggesting that the source of MSCs might be important to delineate a particular lineage in vivo (Figure 5). Although MSCs from both sources start the chondrogenic program with high levels of SOX9 expression, they do not necessarily equally follow and complete the cell differentiation process (Figure 5). In vitro studies have shown that UCB-MSCs preferentially differentiate into osteogenic lineages[17] and present a higher in vivo osteogenic and chondrogenic differentiation potential than BM-MSCs when grafted subcutaneously with a scaffold of hydroxyapatite/tricalcium phosphate[19]. By contrast, PL-MSCs preferentially differentiate into an osteogenic lineage when seeded into microcarriers or nanofiber scaffolds[20,21]. These findings suggest that depending on the tissue source, MSCs can represent a cell population with a different number of osteogenic or chondrogenic progenitors or both. Similarly, the molecular markers from chondrogenic and osteogenic lineages were induced at higher levels in UCB-RLs than in PL-RLs. However, the high expression levels of Runx2 and Scx genes did not lead to the formation of well-defined bone tissue or tendons. In addition, the gene expression pattern of Sox9, Scx, and Rux2 did not associate with a specific tissue pattern or arrangement inside the RLs. The expression of molecular markers was not sufficient to promote the formation of well-defined structures, likely because the morphogenetic program was not triggered. Interestingly, PL-RLs expressed high levels of Mkx while UCB-RL expressed high levels of Scx. It is unknown if in this system both genes are needed to form well-structured tendons. Nevertheless, the expression of Mkx or Scx can drive the in vitro differentiation of BM-MSCs to a tenocyte fate[13,27].

Figure 5 Mesenchymal stromal cells commit to limb mesodermal lineages but are not capable of forming skeletal structures. Mesenchymal stromal cells (MSCs) from umbilical cord blood (UCB) and placenta (PL) differentiated into chondrogenic, osteogenic, and tenogenic lineages in vitro. However, under the influence of differentiation and morphogenetic signals from limb ectoderm in the recombinant limb (RL) system, MSCs committed to limb lineages but failed to organize into tridimensional structures and form skeletal elements as observed with limb bud mesodermal cells under the same conditions. Similarly, the response to ectodermal signals depended on the source of MSCs: UCB-RLs had higher expression of osteogenic and tenogenic markers than PL-RLs, whereas in PL-RLs, tenogenic markers were downregulated after receiving ectodermal signals. UCB: Umbilical cord blood; PL: Placenta; LBMC: Limb bud mesodermal cell; RL: Recombinant limb.

The inability of MSCs to generate complex structures suggests that MSC populations, in addition to being composed of a wide variety of cell types, are unable to initiate or follow a morphogenetic program in vivo. Although MSCs can differentiate into skeletal lineages, it does not guarantee that they will organize into skeletal elements. It is possible that the RL model lacks signals to support the formation of skeletal elements. However, CK-LBMCs respond to differentiation signals, resulting in the generation of skeletal elements. Thus, the RL system provides signals that instruct cells to form skeletal elements.

We did not observe exacerbated cell death in the MSCs-RLs, but in some MSCs-RLs the loss of Fgf8 expression in the AER was evident. Additional studies are needed to determine if this cell death results from the loss of FGF8 loop signaling between the AER and MSCs, or if MSCs cannot recover and maintain the loop to maintain FGF8 expression. Our data suggest that the differentiation process in MSCs is detached from the process of morphogenesis and patterning, although it is possible that additional signals may be needed to promote the integration of morphogenetics with differentiation signals in MSCs. Accordingly, other studies have shown that after implantation in the kidney subcapsular region or nude mice[28-31], MSCs organize better than in vitro but never originate complex structures such as skeletal elements or other limb musculoskeletal tissues.

Although MSCs are considered relevant in regenerative medicine, their intrinsic cellular properties may explain why the reported therapeutic effects of MSCs are mostly indirect through immunomodulation or paracrine mechanisms rather than reliable integration into adult tissues or de novo tissue formation[32-39].

CONCLUSION

This study demonstrates that the expression of differentiation markers of skeletal lineages in MSCs is not sufficient to generate skeletal structures in vivo, possibly due to the intrinsic characteristics of MSCs. In regenerative medicine, cells must be incorporated into well-defined tissues or generate new complex structures with a well-defined patterning. Thus, further application of MSCs in regenerative medicine needs to focus on understanding their biological characteristics to gain insights into how MSCs can integrate into adult tissues or properly rebuild tissues and organs.

ARTICLE HIGHLIGHTS

Research background

Mesenchymal stem cells (MSCs) differentiate in vitro to different skeletal lineages; however, it is unknown if they have the ability to form complex skeletal structures.

Research motivation

Although MSCs are considered relevant in regenerative medicine, reliable integration into adult tissues or de novo tissue formation has not been demonstrated. The application of MSCs in regenerative medicine needs to focus on understanding their biological characteristics to gain insights into how MSCs can integrate into adult tissues or properly rebuild tissues and organs.

Research objectives

To evaluate the ability of MSCs to organize and form complex skeletal structures in vivo under the influence of embryonic signals.

Research methods

The recombinant limb (RL) is an experimental system that recapitulates the embryonic environment and its influence on cells to generate skeletal structures. Here, umbilical cord blood (UCB)-MSCs or placenta (PL)-MSCs were placed in an RL to assess their ability to form skeletal structures. The evaluation was conducted by Alcian blue staining, immunofluorescence, and quantitative polymerase chain reaction of molecular markers of skeletal lineages.

Research results

MSCs expressed molecular markers of skeletal lineages but were unable to generate complex skeletal structures. PL-MSCs or UCB-MSCs integrated into an RL implanted in a chicken embryo. They responded differently to ectodermal signals starting the chondrogenic, osteogenic, or tenogenic program with high SRY-box transcription factor 9, Runt-related transcription factor 2, and scleraxis gene expression levels. However, PL-MSCs or UCB-MSCs did not complete the cell differentiation and morphogenetic processes, likely due to the intrinsic characteristics of MSCs.

Research conclusions

PL-MSCs or UCB-MSCs express molecular markers of skeletal lineages but do not organize into complex skeletal structures. The use of RLs is an excellent model for determining the ability of cells from other origins than limb bud mesodermal cells to generate skeletal structures. The inability of MSCs to form skeletal structures might be due to their intrinsic characteristics. Additional studies are needed to understand the properties of MSCs and whether they can integrate into adult tissues or properly rebuild tissues and organs.

Research perspectives

The application of MSCs to regenerative medicine needs to focus on understanding their biological characteristics to gain insights into how MSCs can integrate into adult tissues or properly rebuild tissues and organs.

ACKNOWLEDGEMENTS

The authors thank Maria Valeria Chimal-Montes de Oca for her artwork.