Functionally Convergent White Adipogenic Progenitors of Different Lineages Participate in a Diffused System Supporting Tissue Regeneration

Abstract

Pathologies characterized by lipomatous infiltration of craniofacial structures as well as certain forms of lipodystrophies suggest the existence of a distinct adipogenic program in the cephalic region of mammals. Using lineage tracing, we studied the origin of craniofacial adipocytes that accumulate both in cranial fat depots and during ectopic lipomatous infiltration of craniofacial muscles. We found that unlike their counterparts in limb muscle, a significant percentage of cranial adipocytes is derived from the neural crest (NC). In addition, we identified a population of NC-derived Lin⁻/α7⁻/CD34⁺/Sca-1⁺ fibro/adipogenic progenitors (NC-FAPs) that resides exclusively in the mesenchyme of cephalic fat and muscle. Comparative analysis of the adipogenic potential, impact on metabolism, and contribution to the regenerative response of NC-FAPs and mesoderm-derived FAPs (M-FAPs) suggests that these cells are functionally indistinguishable. While both NC- and M-FAPs express mesenchymal markers and promyogenic cytokines upon damage-induced activation, NC-FAPs additionally express components of the NC developmental program. Furthermore, we show that craniofacial FAP composition changes with age, with young mice containing FAPs that are almost exclusively of NC origin, while NC-FAPs are progressively replaced by M-FAPs as mice age. Based on these results, we propose that in the adult, ontogenetically distinct FAPs form a diffused system reminiscent of the endothelium, which can originate from multiple developmental intermediates to seed all anatomical locations.

Disclosure of potential conflicts of interest is found at the end of this article.

Craniofacial, Adipocite progenitor, Muscle injury

Introduction

Craniofacial morphogenesis requires complex interactions between populations of cells derived from the prechordal, lateral, and paraxial mesoderm with cells derived from the cranial neural crest (CNC) cell group [1–3]. The CNC constitutes an early mesenchymal population emerging from the rostral region of the neural tube that subsequently migrates along a subectodermal pathway and aligns with the mesodermal mesenchyme segments [4]. Upon alignment, both mesenchymal populations comigrate, mix extensively, and interact closely to generate major structures including cephalic bones and muscles [2]. At later developmental stages, the close interaction between mesoderm and NC-derived cells decreases as discrete structures generated by each lineage result in well-defined borders that keep the two cell types apart [2]. The cephalic mesenchyme of the adult constitutes an exception to this segregation process in that both NC- and mesoderm-derived cells remain interspersed throughout life.

Previous data have indicated a role for the NC in the generation of white adipose tissue (WAT), one of the major mesenchymal lineages [5]. While the mesoderm has long been considered the only source of WAT, certain forms of WAT dystrophies, such as congenital infiltrating lipomatosis of the face (CIL-F) [6, 7] and the Dunnigan-Kobberling syndrome [8] suggest the existence of different cellular mechanisms for cephalic versus trunk and limb adipogenesis. Recently, others and we have reported the existence of a population of mesenchymal progenitors residing in both WAT and skeletal muscle that contains virtually all the adipogenic activity of both tissues [9–12]. These cells are bipotent, are capable of producing both adipocytes and fibroblasts in vitro [11, 12], and were consequently termed fibro/adipogenic progenitors (FAPs) [13]. FAPs proliferate and undergo differentiation depending on the systemic metabolic and microenvironment signals. In WAT, this process results in the generation of new adipocytes and tissue expansion. In skeletal muscle, FAPs play a dual role; they facilitate regeneration following acute damage but can also contribute to fibrofatty infiltration when regeneration fails [11, 12]. In this study, we identified a population of NC-derived CD31⁻/CD45⁻/α7⁻/Sca1⁺ progenitors that resides in the craniofacial mesenchyme. While NC-derived FAPs constitute virtually the entirety of the FAP population at early postnatal stages, mesodermal FAPs (M-FAPs) copopulate the craniofacial mesenchyme later in postnatal development. In order to study the behavior of NC-FAPs, we investigated the degree of developmental homoplasy between NC-FAPs and M-FAPs both in vitro and in vivo. We found that M-FAPs and NC-FAPs are similar in their developmental potential as well as in their ability to modulate metabolism, to participate in the regenerative response, to acute muscle damage, and to contribute to fatty infiltration following degenerative damage.

Materials and Methods

Animals and Muscle Damage

WNT1-Cre/R26-yellow fluorescent protein (YFP) mice were generated by crossing WNT1-Cre^± (The Jackson Laboratory, Bar Harbor, Maine, www.jax.org) with R26-YFP^± mice (The Jackson Laboratory). R26-YFP littermates were used as controls. Adult (>8 weeks) C57BL/6J-CMV-β actin-enhanced green fluorescent protein (EGFP) transgenic mice were used as donors in transplantation experiments. Platelet-derived growth factor receptor (PDGFR)α-H2B//EGFP mice (The Jackson Laboratory) were used for flow cytometry analysis. A-ZIP/F-1 mice were generated by crossing hemizygous friend virus B-type (FVB)/N males (Dr. Gavrilova, Diabetes Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD) with wild-type C57BL/6J females. In transplantation experiments, >6-week-old A-ZIP/F-1 mice were used as recipients and wild-type littermates were used as controls. For muscle damage experiments, WNT1-Cre/R26-YFP and control mice were anesthetized with 0.5%–2% isoflurane, and the damage was induced by intramuscular injection of 0.15 μg notexin (NTX) (Latoxan, Valence, France, www.latoxan. com) into the masseter and tibialis anterior (TA) muscle. Muscle degeneration was induced by injecting 60 μl of 70% (vol/vol) glycerol into the masseter and the TA muscle. All mice were maintained in a pathogen-free facility and all experiments were performed in accordance to the University of British Columbia Animal Care Committee regulations.

Flow Cytometry

Sample preparation and flow cytometry were performed as previously described [12]. The following monoclonal primary antibodies were used: anti-CD31 (clones MEC13.3, Becton Dickenson, Mississauga, Canada, www.bd.com, and 390; Cedarlane Laboratories, Burlington, Canada, https://pilot.cedarlanelabs.com), anti-CD34 (clone RAM34; eBioscience), anti-CD45 (clone 30-F11; Becton Dickenson), anti-Sca-1 (clone D7; eBiosciences, San Diego, California, www.ebiosciences.com), and anti-α7 integrin (produced in-house). The antibody dilutions were as previously reported [9, 12]. Analysis was performed on LSRII (Becton Dickenson) equipped with three lasers. Data were collected using FacsDIVA software. Sorts were performed on a FACS Vantage SE (Becton Dickenson) or FACS Aria (Becton Dickenson), both equipped with three lasers. Sorting gates were strictly defined based on isotype control (fluorescence minus one) stains. Data analysis was performed using FlowJo 8.7 (Treestar, Ashland, Oregon, www.treestar.com) software.

Transplantation

Host mice (wild type) were anesthetized using isoflurane. Donor cells were isolated from either WNT1-Cre/R26-YFP mice or C57BL/6J-CMV-β actin-EGFP mice. Cells were sorted into cold Dulbecco's modified Eagle's medium and collected by centrifugation at 450g for 5 minutes. Cells were resuspended in 25 μl Matrigel and loaded into an ice-cold needle and syringe immediately before injection. Cells were injected into the subscapular region. Tissues were collected after 4 weeks.

Limiting Dilution Analysis

From each test population, 1-100 cells were sorted into individual wells of a type 1 collagen-coated 96-well plate directly from the sorter. Cells were grown as described in the text. After 4 weeks, cultures were fixed and stained for perilipin and nuclei. Wells were scored for the presence of colonies (>8 cells) and adipocytes. A minimum of 30 replicate wells was generated for each cell dose. Limiting dilution analysis (LDA) calculations were based on the single hit Poisson model.

Gene Expression Analysis

RNA isolation was performed using RNeasy mini kits (Qiagen, Valencia, California, www.qiagen.com) and reverse transcription was performed using the Superscript Reverse Transcriptase (Applied Biosystems, Foster City, California, www.appliedbiosystems. com). Gene expression analysis was performed using Taqman Gene Expression Assays (Applied Biosystems), on a 7900HT Real Time PCR system (Applied Biosystems). Sequence information for the primers contained in the Taqman assays are not in the public domain, but ordering information is provided in Supporting Information, Table S1. Data were acquired and analyzed using SDS 2.0 and SDS RQ Manager software (Applied Biosystems). Analysis of chondrogenic and osteogenic gene expression was performed using a custom-designed Taqman Gene Signature array (Applied Biosystems). The array contained 384 gene expression assays covering 48 mesenchymal lineage markers. For each sample, 100 ng of cDNA was used, and the reactions were set up following the manufacturer's instructions. Reactions were run on a 7900HT Real Time PCR system (Applied Biosystems) using low-density block. Data were generated using SDS RQ Manager software (Applied Biosystems).

Hormone and Glucose Measurement

Blood samples were collected the day prior to the transplantation procedure and on a biweekly basis thereafter. Plasma insulin was measured using electrochemiluminescent immunoassays following the manufacturer's instructions (Meso Scale Discovery, Gaithersburg, Maryland, www.mesoscale.com), and the signal was detected on a Sector imager 2400 (Meso Scale Discovery). Blood glucose was monitored biweekly using a OneTouch Ultra Meter (Lifescan, Milpitas, California, www.lifescan.com).

In Vivo 5-Bromodeoxyuridine Labeling

5-Bromodeoxyuridine (BrdU) was administered daily by intraperitoneal injection (100 mg/kg) for 3 days, starting on the day of the damage. For flow cytometric analysis, cells were stained for surface markers as previously described [12]. The cells were then fixed in 2% paraformaldehyde, permeabilized with 0.2% saponin for 10 minutes, treated with DNase I (Sigma, Oakville, Canada, www.sigmaaldrich.com) for 1 hour at 37°C, and incubated for 30 minutes with an anti-BrdU antibody (clone PRB-1, Invitrogen, Grand Island, New York, www.invitrogen.com).

Statistics

Statistical analysis was performed using Prism Five for Mac OS (GraphPad Software, La Jolla, California, www.graphpad.com).

Results

NC-Derived Adipocytes Participate in Fat Depot Formation and Craniofacial Muscle Infiltration

To test the hypothesis that NC-derived adipocyte progenitors are present at postnatal developmental stages, we performed lineage tracing in WNT1-Cre/R26-YFP mice, in which NC-derived tissue becomes indelibly labeled upon activation of the WNT1 promoter during NC formation [14]. In order to identify the niche and the anatomical organization of prospective NC-derived adipocyte progenitors, we first studied fat tissue located in the cephalic area. Upon anatomical inspection of 5-week-old mice, we identified a well-defined YFP⁺ fat depot that extends around the neck, between the cervical and the pectoral areas (Fig. 1A). Conversely, we did not detect expression of YFP in the flank (subcutaneous) (Fig. 1A) or in the perigonadal (omental) fat depots (Fig. 1B) of the same WNT1-Cre/R26-YFP mice. We confirmed the specific YFP expression by immunofluorescence using an anti-YFP antibody. As shown in Figure 1C, the cephalic fat depot is constituted exclusively by NC-derived white adipocytes, whereas this lineage was absent in the adipocyte population of the flank depot.

1

Neural crest (NC)-derived adipocytes in a cervical depot and craniofacial muscle fatty infiltration. (A): Identification of a well-defined cephalic fat depot surrounding the neck of 5-week-old WNT1-Cre/R26-YFP mice that derives from the NC. (B): Only the cephalic depot (center) originates from NC, while the flank (left) and the visceral (right) depots do not. (C): Immunohistochemistry using an anti-green fluorescent protein antibody confirmed that only the cephalic depot is composed of NC-derived white adipocytes (upper panel). (D): Masseter and tibialis anterior (TA) muscles from WNT1-Cre/R26-YFP mice were injected with glycerol and harvested 14 days later and adipocytes were detected by immunofluorescence against perilipin, along with TOTO3 staining. Arrowheads indicate adipocyte-containing clusters. Fatty infiltration in the TA at day 14 is shown in the rightmost panel, where arrows indicate NC-derived nerves. (E): Higher magnification of the adipocyte-containing areas revealed the presence of NC-derived (YFP⁺) adipocytes in the masseter but not in the TA of WNT1-Cre/R26-YFP mice at day 14 after glycerol injection. The inset numbers in the merge box correspond to the percentage of quantified NC-derived adipocytes.

While intramuscular fatty infiltration is observed in CIL-F [6], ectopic fat accumulation also occurs following degenerative skeletal muscle damage [11]. We next studied the origin of craniofacial muscle fibro/fatty infiltration. To this end, we induced degenerative damage by means of glycerol injections delivered into both the masseter and the TA of 5-week-old WNT1-Cre/R26-YFP mice. This procedure has been shown to cause intramuscular adipocyte accumulation [11]. Fourteen days after the injection, adipocyte accumulation was clearly detected in both muscles by means of immunofluorescence against perilipin (Fig. 1D). In the case of the masseter, adipocyte clusters were observed (indicated by arrowheads in Fig. 1D, center panel), where 70% of the adipocytes observed were of NC origin (Fig. 1E). On the contrary, no NC-derived adipocytes were detected in the TAs of the same animals (Fig. 1D, 1E), despite the fact that YFP⁺ NC-derived nerve structures were readily observed (indicated by arrows in Fig. 1D, right panel).

NC-Derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ Cells are FAPs

The existence of NC-derived adipose tissue in adult mice suggested that expansion and maintenance of this tissue is supported by progenitors of NC lineage. In order to isolate NC-derived adipocyte progenitors, we dissected and enzymatically dissociated the cephalic, flank, and perigonadal fat depots from WNT1-Cre/R26-YFP mice to separate the stromal vascular component. To identify prospective NC-derived adipocyte progenitors, we followed a flow cytometry strategy previously developed to identify WAT progenitors in mesodermal fat depots [9]. We first excluded cells from hematopoietic, endothelial, and myogenic lineages by gating the stromal vascular fraction (SVF) based on the expression of CD45 (hematopoietic), CD31 (endothelial), and α₇ integrin (skeletal/smooth muscle) (Fig. 2A). The CD45⁻/CD31⁻ (Lin⁻)/α₇⁻ subpopulation, which comprised approximately 14% of the SVF, was then separated based on the expression of stem cell antigen 1 (Sca1), a marker that labels proliferating cells in fat depots [15] and expression of CD34, a marker expressed by stem/progenitor cells [16] and primary preadipocytes [17]. We found that approximately 27% of the Lin⁻/α₇⁻ population was CD34⁺/Sca1⁺, with up to 95% of the Sca1⁺ cells expressing CD34 (Fig. 2A). Within the Sca1⁺ population, approximately 70% of the cells was YFP⁺, and therefore, of NC lineage (Fig. 2B, left panel). This population comprised approximately 2% of the total cephalic fat depot SVF. No YFP⁺ Lin⁻/α₇⁻/CD34⁺/Sca1⁺ (Sca1⁺) population was observed in the SVF obtained from digestion of the flank depot of WNT1-Cre/R26-YFP mice (Fig. 2B, right panel).

2

NC- and mesoderm-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells contain functionally comparable adipogenic progenitors. (A): Flow cytometry analysis of the cephalic depot stromal vascular fraction from 5-week-old WNT1-Cre/R26-YFP mice. The stromal fraction was analyzed for expression of CD31/CD45 (Lin), CD34, and Sca1. (B): Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells from the cephalic fat depot (left panel) and the flank fat depot (right panel) were separated based on YFP expression. (C): Lin⁻/α₇⁻/CD34⁻/Sca1⁻ and Lin⁻/α₇⁻/CD34⁻/Sca1⁺/YFP⁺ cells purified from the cephalic depot and Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells purified from the flank depot of WNT1-Cre/R26-YFP mice were cultured in medium containing adipogenesis-inducing factors. Macroscopic (upper panels) and microscopic (lower panels) views of the oil red O-stained wells are shown. Lin⁻/α₇⁻/CD34⁻/Sca1⁺/YFP⁺ cells purified from the cephalic depot were cultured in chondrogenic medium (D) and osteogenic medium (E). MEFs were used as positive control. (F): Expression of adipocyte markers was measured by Taqman RT-PCR on total RNA extracted from Lin⁻/α₇⁻/CD34⁺/Sca1⁺/YFP⁺ cells freshly sorted and cultured in either GM or AM for 14 days. Results are shown as mean ± SD for each condition (n = 3). NC-FAPs and M-FAPs, isolated from the cephalic and flank depot of WNT1-Cre/R26-YFP mice, respectively, were cultured in adipogenic medium. Purified Lin⁻/α₇⁻/CD34⁻/Sca1⁻ cells were used as control. Macroscopic (upper panels) and microscopic (lower panels) views of the oil red O-stained wells are shown. (G): Lin⁻/α₇⁻/CD34⁺/Sca1⁺/YFP⁺ cells were isolated from the cephalic depot of WNT1-Cre/R26-YFP mice and incubated with TGF-β1 for 7 days. Expression of fibrogenic markers α-SMA, Vimentin, Desmin, Col1, FN1, and SM-MHC was measured by Taqman RT-PCR. Results are shown as mean ± SD for each condition (n = 3). Abbreviations: AM, adipogenic medium; GM, growth medium; MEF, mouse embryonic fibroblast; M-FAP, mesoderm-derived fibro/adipogenic progenitor; NC-FAPs, neural crest-derived FAPs; Sca1, stem cell antigen 1; TGF, transforming growth factor.

Next, we tested the ability of NC-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells to differentiate into adipocytes in vitro. Lin⁻/α₇⁻/CD34⁺/Sca1⁺/YFP⁺ cells isolated from 5-week-old WNT1-Cre/R26-YFP mice were cultured in proadipogenic media containing insulin, troglitazone, isobutylmethylxanthine (IBMX), and dexamethasone. The presence of mature adipocytes in cultures of the Lin⁻/α₇⁻/CD34⁺/Sca1⁺/YFP⁺ and the Lin⁻/α₇⁻/CD34⁻/Sca1⁻ populations from the cephalic depot, along with the Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells from the flank depot of adult were assessed in parallel by oil red O staining in vitro. We found that both the NC- and M-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cell populations efficiently differentiated into adipocytes with unilocular lipid-containing vacuoles that stained positive for oil red O (Fig. 2C). In contrast, we observed little or no adipogenic activity in the Lin⁻/α₇⁻/CD34⁻/Sca1⁻ population cultured under the same adipogenic conditions (Fig. 2C). In order to test whether NC-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells could generate other mesenchymal lineages, we exposed the NC- and M-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells to osteogenic and chondrogenic conditions. After 20 days in culture, NC-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells failed to differentiate into either bone or cartilage (Fig. 2D, 2E) or to express chondrogenic and osteogenic markers (Supporting Information Fig. S1).

In order to further confirm adipogenic differentiation of NC-FAPs, the expression of adipocyte-associated markers was measured by qPCR. After 10 days in adipogenic conditions, we detected the upregulation of the master adipogenic regulators PPARγ and C/EBP along with the increased expression of lipoprotein L and adiponectin (Fig. 2F). Simultaneously, we observed a decrease in the expression of Pref-1, a factor secreted by preadipocytes that inhibits adipogenesis [18]. We could not detect the expression of UCP1, indicating that these NC-derived progenitors do not differentiate into brown adipocytes under these culture conditions (Fig. 2F).

It has been reported that TA resident Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells express fibroblast markers such as SMA, SM22, Calponin, and sm-MHC upon incubation with transforming growth factor (TGF)-β1 [11]. In order to study whether NC-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells can adopt a fibrogenic lineage, Lin⁻/α₇⁻/CD34⁺/Sca1⁺/YFP⁺ cells were purified from adult WNT1-Cre/R26-YFP mice and cultured in the presence of TGF-β1 for 7 days. We found that TGF-β1 upregulated the fibroblast markers α-SMA, Vimentin, and Desmin, the myofibroblast marker SM-MHC, and the fibrosis-associated markers Col1a1 and FN1 (Fig. 2G). Altogether, these data indicate that NC-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ mesenchymal cells constitute a population of progenitors of the fibro/adipogenic lineage, and hence we termed them NC-FAPs (NC-FAPs). Importantly, our data do not exclude the possibility that these cells may adopt additional mesenchymal fates upon exposure to specific environmental stimuli.

NC-FAPs Constitute a Resident Population Localized to the Interstitial Spaces of Cephalic Tissues that Decreases with Age

Our group as well as others have recently established that within the Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cell population, cells expressing PDGFRα account for virtually all the adipogenic potential present in skeletal muscle [11, 12]. In order to verify that the same holds true in the craniofacial mesenchyme, we analyzed the SVFs of both the cephalic fat depot and the masseter of PDGFRα-H2B/EGFP mice by flow cytometry. We found that 95%–97% of the Sca1⁺ cells are PDGFRα⁺ both in adipose tissue and skeletal muscle (Supporting Information Fig. S2). Based on those data, we used an anti-PDGFRα antibody to anatomically localize NC-derived PDGFRα⁺ cells in craniofacial structures such as the cephalic fat depot and the masseter muscle. PDGFRα⁺ cells that were YFP⁺ were readily observed in the interstitial spaces of the cephalic fat depot but not in the flank fat depot of 5-week-old WNT1-Cre/R26-YFP mice (Fig. 3A). Similarly, PDGFRα⁺ YFP⁺ cells were identified in the interstitial spaces between myofibers in the masseter muscle WNT1-Cre/R26-YFP mice (Fig. 3B), while only PDGFRα⁺ YFP^- cells were observed in the TA muscle of the same mice (Fig. 3B). Altogether, these results show that, similarly to what reported for M-FAPs, NC-FAPs reside in the interstitial mesenchyme of both craniofacial muscle and cephalic fat depot in the adult mouse.

3

Neural crest (NC)-derived fibro/adipogenic progenitors reside in craniofacial fat and muscle and decline with age. Tissue sections from WNT1-Cre/R26-YFP mice WAT depots and muscles were subjected to immunofluorescence using antibodies against PDGFRα and YFP. (A): NC-derived PDGFRα⁺ cells reside in the interstitial spaces of the cephalic depot (upper panel) but not in the flank depot (lower panel) of WNT1-Cre/R26-YFP mice. (B): NC-derived PDGFRα⁺ cells lie in close apposition to masseter fibers (upper panel), while no YFP⁺ PDGFRα⁺ cells were observed in the interstitial space of the tibialis anterior (TA). Scale bar = 10 μM. (C): Flow cytometry analysis of the Lin⁻/α₇⁻/Sca1⁺ fraction in the masseter of 1-, 5-, 8-, and 20-week-old mice. The percentages of YFP⁻ and YFP⁺ cells are indicated within the histograms. The gray-lined histogram shown in the 1-week-old mice plot represents the Lin⁻/α₇⁻/ Sca1⁺ fraction in the TA of an age-matched WNT1-Cre/R26-YFP mice. Abbreviations: Sca1, stem cell antigen 1; WAT, white adipose tissue.

It has been reported that in the mouse embryo, the neuroepithelium supplies a transient wave of mesenchymal stem cells that declines throughout postnatal development [19]. In order to test the effect of aging on NC-FAPs, we used flow cytometry to measure the percentage of NC-derived Lin⁻/α₇⁻/Sca1⁺ progenitors in the intact masseters of 1-, 5-, 8-, and 20-week-old WNT1-Cre/R26-YFP mice. We found that NC-FAPs account for 100%, 70%, 30%, and 20% of the Lin⁻/α₇⁻/Sca1⁺ fraction in 1-, 5-, 8-, and 20-week-old mice, respectively (Fig. 3C). This reduction in NC-FAP content was concomitant with an increase in the percentage of what presumably are mesoderm-derived FAPs, although the origin of the Lin⁻/α₇⁻/Sca1⁺/YFP⁻ was not verified in this study.

NC-FAPs and M-FAPs Constitute Convergent Adipogenic Progenitor Populations with Equivalent Metabolic Properties In Vivo

To further functionally compare these populations and allow a clear distinction between quantitative and qualitative differences, we compared the frequency of colony-forming cells (CFCs) as well as adipogenic CFCs (A-CFC) within NC- and M-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cell populations. To this end, we performed LDA in vitro of NC-FAP cells purified from the cephalic depot and M-FAP from the flank depot of WNT1-Cre/R26-YFP mice. The cells were cultured in growth media for 14 days, followed by exposure to adipogenic conditions for 1 week. The data revealed that the clonogenicity of both NC-FAPs and M-FAPs was similar with average values of 1 in 14.1 and 1 in 15.7 CFCs, respectively (Table 1). Importantly, the adipogenicity of the NC-derived population was approximately 3.5 times lower with average values at 1 in 139 cells for NC-FAPs and 1 in 38.9 cells for M-FAPs, respectively (Table 1).

Table 1

Limiting dilution assay of neural crest- and mesoderm-derived Lin⁻/α7⁻/CD34⁺/Sca1⁺ progenitors

Next, we tested whether NC-FAPs and M-FAPs are equally able of generating functional fat in vivo. To this end, we purified NC-FAPs (YFP⁺) from the cephalic depot of WNT1-Cre/R26-YFP mice and M-FAPs from the flank depot of CAG-EGFP⁺ mice by cell sorting and transplanted them subcutaneously into the subscapular region of lypodystrophic A-ZIP/F-1 recipient mice. These mice lack white fat and provide a strong proadipogenic environment for incoming transplanted cells. Four weeks postinjection, flat circular fatty lumps of approximately 4–5 mm² in diameter were identified in 100% (8/8) of the recipient animals in the subcutaneous regions where the FAPs had been injected. The engraftments were harbored between the subcutaneous layer and the dermis of the animals that received donor progenitor cells, while no engraftment was observed in mice that received sham injections of support matrix. Histological analysis showed that the depots contained either YFP⁺ or green fluorescent protein positive (GFP⁺) unilocular adipocytes that stained positive for perilipin (Fig. 4A).

4

NC-FAPs and M-FAPs possess equivalent metabolic properties. (A): Both Lin⁻/α₇⁻/CD34⁺/Sca1⁺/YFP⁺ cells purified from the cephalic depot of WNT1-Cre/R26-YFP mice or Lin⁻/α₇⁻/CD34⁺/Sca1⁺/GFP⁺ cells purified from the flank depot of β actin/EGFP mice differentiated into mature adipocytes in A-ZIP/F-1 mice. Scale bar = 20 μM. (B): Analysis of plasma insulin and glucose in WT and A-ZIP/F-1 mice, 4 weeks after injection of equal numbers of adipogenic progenitors. The Sham group received vehicle only. Results are shown as mean ± SEM for each group (n = 4). *, p, .05. Abbreviations: GFP, green fluorescent protein; M-FAP, mesoderm-derived fibro/adipogenic progenitor; NC-FAPs, neural crest-derived FAPs; WT, wild-type.

Complete reversion of hyperglycemia and hyperleptinemia after transplantation of adipocyte progenitors in lypodystrophic mice has been recently reported [10]. Following the same principle, we tested the ability of NC-FAPs and M-FAPs to revert the diabetic phenotype of A-ZIP/F-1 mice. Based on the differences in A-CFC content between Lin⁻/α₇⁻/CD34⁺/Sca1⁺ purified from cephalic and flank adipose depots detected in our in vitro limiting dilution assay, we devised two transplantation assays, one in which equal numbers of cells were transplanted into A-ZIP/F-1 recipients, and another in which the number of cells was normalized to ensure that the same number of adipogenic progenitors was transplanted. Our hypothesis was that if no cell-intrinsic differences exist between the two populations, transplantation of equivalent numbers of adipogenic CFCs should have similar effects on the metabolic parameters of A-ZIP/F-1 mice.

Four weeks after the injection of 5 × 10⁴ sorted cells, M-FAP recipient A-ZIP/F-1 mice displayed significant reductions in plasma insulin and glucose levels when compared with both pretransplantation levels and levels in the sham-transplanted group (Supporting Information Fig. S3). In contrast, no significant differences were found in plasma insulin or glucose levels compared to pretransplantation levels in NC-FAP recipient A-ZIP/F-1 mice (Supporting Information Fig. S3).

In the second transplantation experiment, either 1.75 × 10⁵ (YFP⁺) NC-derived or 5 × 10⁴ (GFP⁺) mesoderm-derived Lin⁻/α₇⁻/CD34⁺/Sca1⁺ cells, both corresponding to approximately 1,300 adipogenic CFCs, were transplanted into A-ZIP/F-1 mice. In this case, plasma insulin and glucose levels significantly dropped in the NC-FAP recipient animals at week 4, compared to their pretransplantation values, and were significantly lower than the values of the sham group at week 4 (Fig. 4B). Similar to the first experiment, the M-FAP recipients had lower plasma insulin and glucose concentrations at week 4 compared to the pretransplantation values of the same group (Fig. 4B). Altogether, these data suggest that observed differences in the efficiency with which NC-FAPs and M-FAPs rescue the metabolic defect of lypodystrophic mice can be primarily attributed to a lower frequency of adipogenic clones in the NC-FAPs cell population rather than to cell-intrinsic differences between the two subpopulations, and therefore provide support to the notion that these two developmentally distinct progenitors populations possess convergent metabolic roles.

The NC-FAP Population Becomes Activated upon Acute Craniofacial Muscle Damage

Results obtained in limb skeletal muscle show that FAPs become activated and proliferate following injury [12]. We next investigated the response to acute muscle damage of the YFP⁺ (NC-derived) and YFP⁻ FAP populations residing in the craniofacial region. To that end, both the masseter and the TA of 5-week-old WNT1-Cre/R26-YFP mice were injected with NTX—a myotoxin that causes muscle damage specifically to differentiated myofibers. Based on our previous data indicating that virtually all Sca1⁺ cells are CD34⁺ [12], CD34 labeling was omitted in these experiments. The flow cytometry profiles obtained for the markers studied (CD31, CD45, α7, and Sca1) were similar to those observed in the adipose tissue, with a Lin⁻/α₇⁻/Sca1⁺ subpopulation clearly defined in both the masseter and the TA (Supporting Information Fig. S4). In both undamaged masseter and TA, FAPs represented ∼13% of the Lin⁻ population. Similar to what we observed for the cephalic adipose tissue, the Sca1⁺ fraction residing in the masseter contained two subpopulations: a major YFP⁺ one that accounted for 75% of total FAPs and ∼10% of the Lin⁻/α₇⁻ population, and a YFP⁻ subpopulation that accounted for 25% of FAPs and ∼3.4% of the Lin⁻/α₇⁻ fraction. Conversely, NC-derived Sca1⁺ cells were not observed in the TAs of the same animals (Supporting Information Fig. S4).

Regardless of the origin, and in agreement with our previously reported data [12], we found similar expansion of FAPs at day 3 post-NTX damage (D3) both in the masseter and in the TA, which was reflected by a significant increase in the percentage of FAPs in the Lin⁻ fraction (Supporting Information Fig. S4). Lineage tracing of the activated FAP populations revealed that in the masseter YFP⁺, FAPs expanded efficiently, rising from 10% to 30% of the Lin⁻/α₇⁻ population on D3 (Fig. 5A, 5B). The frequency of YFP⁺ FAPs returned to predamage levels by postdamage day 7 (D7) (Fig. 5A, 5B). Similarly, the percentage of YFP⁻ cells in the masseter also increased albeit by a lower amount and returned to basal levels by D7 (Fig. 5 A, 5B). In the TA, damage led to an increase in the number of YFP⁻ FAPs on D3 followed by restoration of predamage values on D7, as previously described (Supporting Information Fig. S4).

5

NC-FAPs proliferate following craniofacial muscle damage. (A): Representative flow cytometry profiles of the Lin⁻/α₇⁻/Sca1⁺ population and its lineage composition in the masseter of WNT1-Cre/R26-YFP mice at D0, D3, and D7 after damage. (B): Frequency of NC-FAPs and M-FAPs in the masseter of WNT1-Cre/R26-YFP mice at D0, D3, and D7 after damage. Results are shown as mean ± SEM for each group (n = 5). *, p, .05. (C): BrdU incorporation in masseter-resident NC-FAPs and M-FAPs at D0, D3, and D7 after damage. (D): Frequencies of proliferating NC-FAPs and M-FAPs in the masseter. Results are shown as mean ± SEM for each group (n = 4). (E): Immunofluorescence staining of masseter for YFP and PDGFRα following notexin (NTX) damage in 5-week-old WNT1-Cre/R26-YFP mice. Representative images of masseters from nondamaged (D0) and day 3 post-NTX injection (D3) mice are shown. Scale bar = 50 μM. Higher magnification images are depicted on the right. Abbreviations: M-FAP, mesoderm-derived fibro/adipogenic progenitor; NC-FAPs, neural crest-derived FAPs; Sca1, stem cell antigen 1.

To further characterize FAP expansion in craniofacial muscle, we measured BrdU incorporation following masseter damage. To this end, 5-week-old WNT1-Cre/R26-YFP mice received daily i.p. BrdU injections, starting at the time of damage and continuing until the animals were sacrificed. Muscle was collected prior to damage (D0), 3 days (D3) and 7 days (D7) post-NTX injection, enzymatically dissociated, and analyzed by flow cytometry. The percentage of BrdU⁺ NC-FAPs in the masseter increased from 0.81% ± 0.44% at D0 to 70.7% ± 12.2% at D3 (Fig. 5C, 5D) and returned to basal levels by D7 (Fig. 5C, 5D). The percentage of BrdU⁺ YFP⁻ FAPs significantly increased from 0.93% ± 0.92% to 48.3% ± 8.3% following masseter damage (Fig. 5C, 5D). Altogether, the data indicate that NC-FAPs enter cell cycle and participate in the FAP response following acute muscle damage in the craniofacial region of 5-week-old mice. To further characterize the NC-FAP response following acute muscle damage, we injected NTX into the masseter of WNT1-Cre/R26-YFP mice and studied the distribution of NC-FAPs by immunofluorescence. Our results show that PDGFRα⁺ NC-FAPs (YFP⁺) infiltrate the masseter at D3 postdamage in a mesh-shaped pattern (Fig. 5E). As expected in the TA, no YFP⁺ NC-FAPs were evident either before or after damage, at a time in which the M-FAPs profusely infiltrated the interstitial spaces (Supporting Information Fig. S4).

Regulation of Promyogenic Cytokines and Lineage Markers During FAP Activation

We previously showed that limb muscle FAPs facilitate myogenesis by providing trophic signals to myogenic progenitors. By means of quantitative reverse transcription-polymerase chain reaction (qRT-PCR), we assessed the transcriptional activity of cytokines that enhance myoblast differentiation and fusion both in NC-FAPs purified from the masseter muscle as well as M-FAPs purified from the TAs of WNT1-Cre/R26-YFP mice. Expression of IL-6 was significantly higher in both activated (postdamage day 3) NC-FAPs (approximately threefold) and M-FAPs (approximately fivefold) (Fig. 6A). Expression of IL-10—an anti-inflammatory cytokine that facilitates myoblast differentiation by preventing the antimyogenic activity of tumor necrosis factor (TNF)α and IL-1β [20]—was also upregulated in both FAP populations upon activation (Fig. 6A). Thus, limited to these promyogenic factors, NC- and mesoderm-derived cells display a similar response.

6

Activated NC-FAPs upregulate the expression of promyogenic cytokines and NC migration markers. (A): NC-FAPs and M-FAPs were purified from nondamaged (D0) and damaged (D3) masseters and tibialis anteriors, respectively, of WNT1-Cre/R26-YFP mice. Total RNA was purified and the expression of the IL-6, IL-10, Snail, Twist1, EdnrA, and Ecad was measured by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) using specific primers and probes. Results are shown as mean ± SEM for each group (n = 4). a, b, p, .05 versus D0. Abbreviations: EdnrA, endothelin receptor type A; M-FAP, mesoderm-derived fibro/adipogenic progenitor; NC-FAPs, neural crest-derived FAPs.

Recapitulation of developmental gene expression programs has been reported during the regeneration of tissues such as muscle [21] and the nervous system [22]. To assess whether genes involved in NC migration are reactivated during the NC-FAP response to masseter damage, we measured the expression of the transcription factor Twist1, which regulates NC cell proliferation, and the transcriptional repressor Snail, which regulates NC cell migration thru E-cadherin repression [23]. The levels of Snail and Twist1 mRNAs were significantly higher in D3 NC-FAPs (∼2.9 and ∼2.3-fold, respectively) (Fig. 6A). Remarkably, both markers were also upregulated in activated M-FAPs (∼3.2-fold and approximately threefold, respectively) (Fig. 6A). Similar to what is observed in embryonic development during the migration of NC cells, we measured a significant decrease in Ecad expression in NC-FAPs. In these experimental conditions, we did not detect Ecad expression in M-FAPs (Fig. 6A). Finally, expression of endothelin receptor type A was significantly higher (approximately sixfold) in activated NC-FAPs, while no significant changes were found in activated M-FAPs (Fig. 6A). Altogether, these data indicate that similar to mesodermal-derived FAPs, NC-FAPs are capable of producing promyogenic cytokines following craniofacial muscle injury. However, unlike in M-FAPs, the response of NC-FAPs to damage involves the reactivation of genes involved in NC cell migration in early stages of development.

Discussion

In this study, we identified a novel population of NC-derived adipogenic progenitors that resides within both cephalic adipose tissue and craniofacial muscles in mice. Our data show that NC-derived adipocytes form a well-defined subcutaneous fat depot in the cervical region of adult WNT1-Cre/R26-YFP mice, confirming previous data reporting the presence of NC-derived adipocytes surrounding the salivary gland in adult Sox10-Cre/R26-YFP mice [5]. Furthermore, neural-crest-derived adipocytes accumulate ectopically during fatty infiltration of craniofacial muscles, accounting for 70% of the adipocytes generated following degenerative damage of the masseter in 5-week-old mice.

Craniofacial WAT dystrophies such as CIL-F, characterized by abnormal hyperplasia of the craniofacial fat tissue, with massive infiltration of facial muscles [24], and Dunnigan-Kobberling syndrome, characterized by a lack of body fat with the concomitant presence of adipose tissue on the face, retro-orbital space and periserous sites, suggest a distinct adipogenic program in the craniofacial region [8]. In support of this notion, we identified a distinct population of neuroepithelium-derived Lin⁻/CD34⁺/Sca1⁺ cells that resides exclusively in the mesenchyme of both cephalic adipose tissue and craniofacial muscle. Our in vitro lineage analysis indicated that NC-FAPs are able to adopt either a fibrogenic or an adipogenic lineage, a characteristic that would support a role for those progenitors in fibro/fatty infiltration. Thus, in this respect, NC-FAPs are functionally similar to their mesodermal counterpart. Our in vitro comparative analysis further indicated that while purified NC-FAPs and M-FAPs contain similar numbers of clonogenic cells, NC-FAPs contain less progenitors capable of forming adipocytes. Our in vivo transplantation assays validated this finding and further indicated that unequal content of adipogenic clones rather than cell-intrinsic differences underlie a lower ability of NC-FAPs to ameliorate the diabetic phenotype of A-ZIP/F-1 mice. Indeed, injecting the same number of adipogenic clones of either lineage into these lipodystrophic recipients resulted in similarly reduced hyperinsulinemia and hyperglycemia. Thus, on a per cell basis, NC- and M-FAPs are functionally indistinguishable.

Two NC-derived cell groups contribute to the establishment of ectoderm-derived mesenchyme (ectomesenchyme), namely the cardiac NC and the CNC. Both groups give rise to NC-derived mesenchyme that closely interacts with their mesoderm-derived counterpart and participate in the development of specific cardiac structures. A subgroup of cardiac NC mesenchymal cells interacts with second heart field mesoderm to develop the outflow tract [25]. Likewise, CNC cells interact closely with cranial mesoderm-derived mesenchyme during the formation of different craniofacial tissues. In line with this notion, we show here that NC-FAPs, a CNC-derived population, reside within craniofacial muscles, which are of mesodermal origin [26]. Our results show that similarly to what we previously reported for limb M-FAPs, NC-FAP expands following acute muscle injury, supporting the case for homoplasy. Further confirmation of this hypothesis stems from experiments showing that both lineages enter cell cycle upon masseter damage. Interestingly, despite the fact that the vast majority of fibroblastic-CFCs found in developing and perinatal muscle are NC-derived, others and we failed to detect any lineage-traced myofibers either spontaneously or after muscle damage. This strongly suggests that, in contrast with what was recently proposed [27], mesenchymal progenitors do not normally contribute to the myogenic lineage, at least in the head.

As indicated by previous data, the neuroepithelium supplies an initial wave of stem/progenitor cells that account for virtually all the multipotent cells observed in the embryo mesenchyme [19]. Those neuroepithelium-derived multipotent cells decline with age and are replaced by cells of a different embryonic source in the adult mouse [19]. Our data provide further evidence for a lineage-succession model. Analysis of the FAP fraction in 1-, 5-, 8-, and 20-week-old mice revealed that while NC-FAPs are the only FAP population in the early postnatal craniofacial mesenchyme, this population declines with age, constituting the minority FAP subpopulation in 8-week-old mice.

Our results indicated that Ecad, Twist1, and Snail are expressed with patterns that resemble those observed during NC detachment and migration. In particular, our data show the concomitant upregulation of Snail and the downregulation of E-cad in NC-FAPs postdamage, a result consistent with the fact that Snail directly represses E-cad expression [28]. In activated M-FAPs, Ecad was not detected, but similarly to NC-FAPs, these cells upregulated both Twist1 and Snail. This suggests that FAPs use similar molecular pathways following muscle damage, regardless of their lineage. We found that EndrA expression increases significantly in activated NC-FAPs after acute masseter damage, while expression of this mRNA was not detected in activated M-FAPs. Signaling through the endothelin system has been directly implicated in craniofacial development [29], and abnormal craniofacial development has been reported in endothelin 1 (ET-1) knockout mice [29]. Since ET-1 secretion by the mesoderm plays a crucial role in NC migration and cell survival [29, 30], a role for endothelin signaling could be proposed in craniofacial muscle regeneration, perhaps via ET-1 secretion by the mesoderm-derived myofibers of the damaged masseter.

Altogether, our data disclose a novel origin for WAT and reveals a new role for the neuroepithelium in adipogenesis and tissue regeneration in the adult mouse.

Conclusion

These results also indicate that FAPs form a diffused system reminiscent of the endothelium, which can originate from multiple developmental intermediates to seed many diverse anatomical locations. The identification of NC-FAPs may provide a new target for the study of pathologies associated with mesenchymal aberrations in the craniofacial region.

Acknowledgements

We thank the BRC Animal Facility, BRC Core Staff, and UBC FACS Facility for their technical assistance. This research was supported by grants from CIHR (MOP-82864) and Heart and Stroke Foundation of Canada (HSF) to F.M.V.R.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.