Transcription factor 21 (TCF21) promotes proinflammatory interleukin 6 expression and extracellular matrix remodeling in visceral adipose stem cells

The visceral (VIS) and subcutaneous (SQ) fat pads are developmentally distinct white adipose tissue depots and contribute differently to inflammation and insulin resistance associated with obesity. The basic helix–loop–helix transcriptional regulator, transcription factor 21 (TCF21), is a marker gene for white adipose tissues and is abundantly expressed in VIS-derived adipose stem cells (ASCs), but not in SQ-derived ASCs. However, TCF21's role in regulating fat depot–specific gene expression and function is incompletely understood. Here, using siRNA-mediated Tcf21 knockdowns and lentiviral gene transfer of TCF21 in mouse ASCs, we demonstrate that TCF21 is required for the VIS ASC–specific expression of interleukin 6 (IL6), a key cytokine that contributes to the proinflammatory nature of VIS depots. Concurrently, TCF21 promotes MMP-dependent collagen degradation and type IV collagen deposition through the regulation of the extracellular matrix (ECM) modifiers, matrix metalloproteinase (MMP) 2, MMP13, and tissue inhibitor of MMP1 (TIMP1), as well as collagen type IV α1 chain (COL4A1) in VIS ASCs. We also found that although IL6 mediates the expression of Mmp13 and Timp1 in VIS ASCs, the TCF21-dependent expression of Mmp2 and Col4a1 is IL6-independent. These results suggest that TCF21 contributes to the proinflammatory environment in VIS fat depots and to active ECM remodeling of these depots by regulating IL6 expression and MMP-dependent ECM remodeling in a spatiotemporally coordinated manner.

inflammatory tissue damage relative to SQ WAT as exemplified by increased macrophage infiltration and extracellular matrix (ECM) remodeling (3). Among a cohort of adipokines secreted from expanding adipose tissues, the expression of interleukin-6 (IL6) from VIS WAT is associated with chronic inflammation in obesity (4). Moreover, the level of adipose IL6 expression is correlated with the degree of insulin resistance in humans (5,6), suggesting a potential role for VIS WAT-derived IL6 in adipose tissue inflammation and insulin resistance associated with visceral adiposity.
Adipose tissue ECM remodeling is another key pathological process observed in obesity (7,8). Transcriptomic profiling of VIS and SQ WAT-derived stem cell antigen-1 (Sca1)-high adipose-derived stem cells (ASCs) demonstrates the differential expression of inflammatory cytokines, matrix metalloproteinases (MMPs), and tissue inhibitors of MMPs (TIMPs) (9). VIS ASCs display increased expression of MMP8 and MMP13 as well as TIMP1, all of which are highly induced under inflammatory and obesogenic conditions (10 -13). Elevated VIS ASC expression of MMP8 and MMP13 was associated with the expression of proinflammatory cytokines, including IL6 and CXCL1 (9). Despite these observations, the transcriptional regulation underlying VIS ASC-specific expression of inflammatory cytokines and MMP/TIMP family members has not been well established.
To define the transcriptional network of VIS ASCs responsible for the co-expression of IL6 and a subset of MMPs and TIMPs, we focused on transcription factor 21 (TCF21), a basic helix-loop-helix (bHLH) transcription factor abundantly expressed in VIS ASCs but barely in SQ ASCs. A human TCF21 gene variant is associated with larger pericardial fat mass (14), and TCF21 promotes epithelial-mesenchymal transition (EMT) of epicardial cells to cardiac fibroblasts (15). Gene targeting of Tcf21 results in perinatal lethality of mice, demonstrating its indispensable role in organ development (16,17).
Even though TCF21 has been recognized as a WAT-specific marker gene (18), its potential role in regulating VIS WATspecific inflammatory gene expression and ECM remodeling has not been fully explored. Here, we identify a novel role for TCF21 in driving the expression of IL6, as well as a subset of MMP/TIMP family members (MMP2, MMP13, and TIMP1) and type IV collagen in IL6-dependent and -independent manners. TCF21-dependent regulation of IL6 expression and ECM remodeling may contribute to the unique characteristics of visceral WAT.

TCF21 is co-expressed with IL6 and specific ECM remodeling factors in visceral adipose stem cells
Our previous work showed the depot-specific transcriptomes of mouse WAT Sca1 high ASCs using whole-genome RNA-seq (9). The gene encoding basic helix-loop-helix transcription factor Tcf21 was among the most differentially expressed; its transcript is present at high level in VIS ASCs but nearly undetectable in SQ ASCs (Fig. 1A). TCF21 is an established regulator of epicardial fibroblast EMT and development of kidney, lung, and spleen (16,17); however, its function in WAT is presently undefined.
The mRNA levels of Il6 and Timp1 were also preferentially expressed in VIS ASCs, whereas those of Mmp2 and Mmp14 were similar between ASCs of both depots (Fig. 1A). Notably, Mmp13 transcript level in VIS ASCs was more than double the level observed in SQ ASCs (Fig. 1A).
Real-time qPCR confirmed our RNA-seq findings, showing VIS ASC expression of TCF21 180-fold, and expression of IL6 50-fold that seen in SQ ASCs (Fig. 1B). The mRNA analysis of human preadipocytes derived from VIS and SQ WAT revealed a similar pattern (Fig. 1C). In agreement with gene expression, immunocytochemistry of cultured ASCs revealed nuclear TCF21 staining in VIS ASCs but not in SQ ASCs (Fig. 1D). IL6 staining was similarly restricted to VIS ASCs and showed substantial co-expression with TCF21 among VIS WAT-derived stromal cells (Fig. 1E). Indeed, IL6 was detected in 89% of TCF21-positive, but only in 34% of TCF21-negative VIS WATderived vascular stromal cells (Fig. 1E, right panel).
Within mouse VIS (perigonadal) adipose depot, TCF21 protein was seen exclusively in cells positive for platelet-derived growth factor receptor ␣, a cell-surface marker of fibroblastadipocyte progenitors (Fig. 1F) (19,20). On the contrary, TCF21 was undetectable in SQ (inguinal) depots isolated from the same animals (Fig. 1F). TCF21 and IL6 were expressed in the same cells specifically in VIS fat depot (Fig. 1F). Taken together, these data show that in WAT, TCF21 is restricted to ASCs of VIS depot and mostly co-expressed with IL6 in these cells. Genes encoding specific ECM remodeling factors, MMP13 and TIMP1, were also preferentially expressed in VIS ASCs relative to SQ ASCs.

TCF21 is necessary for IL6 expression in VIS ASCs
We hypothesized that TCF21 could transcriptionally regulate genes expressed in VIS ASCs, including IL6. To test this, we

TCF21 in adipose IL6 and ECM remodeling
examined the effect of siRNA-mediated TCF21 knockdown on IL6 transcription and protein expression in isolated mouse VIS ASCs. Two independent siRNA oligonucleotides (designed within Tcf21 coding regions) suppressed mRNA levels by 40 and 75%, respectively ( Fig. 2A). Correspondingly, Il6 transcript levels declined to 36 and 46% of baseline (Fig. 2B). We next evaluated the impact of Tcf21 knockdown on levels of IL6 protein in the culture medium. Western blotting revealed progressive accumulation of IL6 protein over a 3-day culture period in control siRNA-treated group, and the time-dependent accumulation of IL6 was comparable with that of type I collagen (Fig. 2C). IL6 protein content was decreased in the media of VIS ASCs treated with Tcf21 siRNAs at each time point, whereas no significant effect was observed for type I collagen (Fig. 2C). IL6 protein was nearly undetectable in the medium of cultured SQ ASCs (Fig. 2C). Together, these data demonstrate that Tcf21 is necessary for IL6 expression specifically in mouse VIS ASCs, and the minimal expression of Tcf21 accounts for the very low IL6 expression in SQ ASCs. Conversely, when we overexpressed human TCF21 in mouse VIS ASCs using lentiviral gene transfer, we observed that Il6 expression increased in parallel with TCF21 expression (Fig. 2D).

TCF21 is required for the expression of collagenolytic MMPs and TIMP1 in VIS ASCs
RNA-seq identified unique expression profiles of ECM remodeling factors (MMPs and TIMPs) in mouse VIS and SQ ASCs (9). Among collagenolytic MMPs, Mmp13 was highly expressed in VIS ASCs, whereas transcripts encoding gelatinolytic MMP2 and MMP9 were more abundant in SQ ASCs. Tcf21 knockdown by siRNA in VIS ASCs markedly suppressed MMP13 mRNA (Fig. 3A). On the contrary, expression of the gene encoding MMP14 (MT1-MMP), a major pericellular collagenase equally expressed in VIS and SQ ASCs (Fig. 1A), was unaffected by Tcf21 siRNA (Fig. 3A). Expression of tissue inhibitor of matrix metalloproteinase-1 (Timp1), abundant in VIS ASCs but not in SQ ACSs, was suppressed by Tcf21 knockdown (Fig. 3A). By contrast, Tcf21 siRNA treatment did not change the expression of Timp2 or Timp3, which showed no predominance in VIS ASC over SQ ASC in RNA-seq data (Fig. 3A). VIS ASC-specific expression of Mmp8 was not impacted by Tcf21 knockdown (Fig. 3A). Unexpectedly, Tcf21 was found to be required also for the expression of Mmp2 in VIS ASCs and to a lesser extent in SQ ASCs ( Fig. 3A and Fig. S2). Consistent with the effect on Mmp2 expression, Tcf21 siRNA reduced MMP2 protein level as detected by gelatin zymography (Fig. 3B).

TCF21 is responsible for type IV collagen deposition by VIS ASCs
In cultured VIS ASCs, Tcf21 knockdown specifically reduced the expression of Col4a1 but had no impact on Col1a1 or Col3a1 (Fig. 4A). This suggests a specific role for TCF21 in Figure 2. TCF21 drives IL6 expression in VIS ASCs. A and B, the siRNA oligonucleotides targeting Tcf21 and control siRNA were transfected to mouse VIS ASCs and cultured for 3 days. Total RNA was extracted to assess Tcf21 and Il6 expression by RT-qPCR (n ϭ 4 independent biological replicates, means Ϯ S.E.). *, p Ͻ 0.05; ***, p Ͻ 0.001; ****, p Ͻ 0.0001 by one-way ANOVA. C, IL6 and type I collagen (COL1) protein levels in the conditioned media of siRNA transfected VIS and SQ ASCs. Conditioned media were collected at indicated time points for Western blotting. Independent experiments were performed twice. D, mouse VIS ASCs were transduced with human TCF21 cDNA by lentiviral gene transfer. Total RNA was extracted for RT-qPCR analysis of TCF21 and Il6 expression (n ϭ 3 independent biological replicates, means Ϯ S.E.). *, p Ͻ 0.05 by ratio paired t test. E, schematic diagram of human IL6 promoter region cloned into pGL3-basic reporter plasmid. Known cis-elements (black boxes) and predicted TCF21-binding E-box sites (hatched boxes) are shown along with respective sequences. F, IL6 promoter in CHO cells regulated by co-transfected full-length and mutant TCF21. Reconstituted IL6 promoter activity (firefly luciferase) was normalized by co-transfected thymidine kinase promoter activity (Renilla luciferase; n ϭ 7, means Ϯ S.E.). ****, p Ͻ 0.0001 by one-way ANOVA. The experiment was independently repeated. The siRNA oligonucleotides targeting Tcf21 and control siRNA were transfected to mouse VIS ASCs and cultured for 3 days. A, total RNA was extracted for RT-qPCR analysis (n ϭ 4, means Ϯ S.E.). **, p Ͻ 0.01; ***, p Ͻ 0.001; ****, p Ͻ 0.0001 by one-way ANOVA. B, to assess MMP2 activity, the media of transfected cells were changed to serum-free DMEM a day after transfection; conditioned media were collected 3 days later and subjected to gelatin zymography. This is a representative figure (n ϭ 3, means Ϯ S.E.). ****, p Ͻ 0.0001 by one-way ANOVA.

TCF21 in adipose IL6 and ECM remodeling
regulation of basement membrane structure. Type IV collagen deposition by VIS ASCs detected by immunohistochemistry was not much changed with Tcf21 knockdown (Fig. 4B). Nonetheless, whereas type IV collagen fibers appeared robust and well organized in control siRNA-treated samples, they appear diffuse and finer in Tcf21 siRNA-treated samples (Fig. 4B, high magnification). The Tcf21 knockdown also reduced the amount of degraded type I collagen products (Fig. 4C), suggesting that TCF21 mediates both degradation of type I collagen fibers and the deposition of a basement membrane component, type IV collagen. TCF21-dependent degradation of type I collagen was MMP-dependent, as shown by the suppression of the collagen degradation by a MMP inhibitor, GM6001 (Fig. 4D). We next assessed the impact of TCF21 overexpression on ASC collagen deposition. Lentiviral gene transfer of TCF21 resulted in enhanced TCF21 signals in VISC ASCs and novel presence of TCF21 in SQ ASCs (Fig. S3). Elevated TCF21 levels markedly enhanced the deposition of type IV and I collagens in VIS ASCs (Fig. 4E). SQ ASCs, which demonstrate a low staining of type IV collagen at baseline, also showed considerably increased deposition of type IV collagen with TCF21 overexpression (Fig. 4E). The effect of TCF21 was collagen type-selective, and no discernable effect was observed on the level of type VI collagen (Fig. 4E). TCF21-dependent accumulation of type IV collagen was pronounced more in VIS ASCs than in SQ ASCs (Fig. 4E). Not only the deposition of collagens, but also TCF21 overexpression augmented the degradation of type I collagen by VIS ASCs (Fig. 4F). Together, these findings suggest that TCF21 actively engages in type I collagen turnover and type IV collagen deposition by VIS ASCs through the regulation of MMP activities.

IL6 is responsible for the expression of MMP13 and TIMP1 but not for collagen IV and MMP2 and ECM remodeling
Given positive regulation of IL6 by TCF21, we sought to determine whether TCF21-dependent ECM remodeling requires IL6 effector function. To test this, we incubated VIS and SQ ASCs with IL6-neutralizing antibody with and without recombinant IL6. We found that mRNA levels of Mmp13 and Timp1 were reduced in the presence of anti-IL6 antibody and that the effect was rescued by the presence of excess recombinant IL6 (Fig. 5A). On the contrary, expression of Col4a1 and Mmp2 were unaffected by either IL6-neutralizing antibody or IL6 protein (Fig. 5A). To determine whether TCF21 regulates type IV collagen deposition and type I collagen degradation through IL6 and its potential effectors, i.e. MMP13 and TIMP1, VIS  way  ANOVA). B, extracellular type IV collagen was detected (red) and quantified, normalized by the number of nuclei (blue). Scale bar, 50 m (low mag) and 10 m (high mag) (n ϭ 5, means Ϯ S.E., not significant per two-way ANOVA). C, extracellular degraded collagen was stained (red) and quantified per nuclei (blue) (n ϭ 5, means Ϯ S.E.) Scale bar, 50 m. *, p Ͻ 0.05; ****, p Ͻ 0.001 by two-way ANOVA. D, ASCs isolated from VIS WAT were cultured in the presence or absence of 10 M GM6001 for 2 days, stained for degraded collagen (red), and quantified per nuclei (blue) (n ϭ 5, means Ϯ S.E.). Scale bar, 50 m. *, p Ͻ 0.05 by Student's t test. E and F, ASCs isolated from VIS and SQ WATs were transduced with TCF21 cDNA by lentiviral gene transfer. The cells were cultured for 3 days and analyzed by immunostaining for collagens type IV, I, and VI and degraded type I collagen (all in red) along with DAPI (nuclei, blue). Scale bar, 50 m. Signal intensities of staining were quantified and normalized by nucleus number (n ϭ 5, means Ϯ S.E.). *, p Ͻ 0.05; **, p Ͻ 0.01; ****, p Ͻ 0.001 by two-way ANOVA (E) or by Student's t test (F).

TCF21 in adipose IL6 and ECM remodeling
ASCs transfected with siTcf21 or control siRNA were cultured in the presence and absence of IL6-neutralizing antibody. Despite the inhibitory effects on the expression of MMP13 and TIMP1, the presence of IL6-neutralizing antibody did not change either type IV collagen deposition or type I collagen degradation (Fig. 5B), suggesting that TCF21 promotes collagenolytic ECM remodeling in a IL6-independent manner.

Discussion
In this report, we have demonstrated for the first time that TCF21, a bHLH transcription factor, drives the expression of IL6 concurrently of MMP2 and type IV collagen in VIS ASCs. TCF21 has been known to regulate the EMT of epicardial fibroblast progenitors (15) and to be essential for the development of the heart, kidney, and spleen (17). Although TCF21 was recognized as a gene highly expressed in white adipose tissues (18), its role in adipose tissue biology, particularly in WAT depot-dependent gene expression, had been unknown.
WAT actively expands in response to excess calorie intake; however, the cellular mechanisms underlying the expansion of two WAT depots, i.e. visceral and subcutaneous WAT, appear different (22,23). Visceral WAT expansion involves not only hypertrophy of adipocytes but ASC proliferation and adipogenesis, particularly in male mice (22,23). Our previous work suggests that a cohort of transcription factors along with cytokines and ECM remodeling genes are differentially expressed between VIS and SQ ASCs (9). Jeffery et al. (23) suggest that visceral WAT provides a microenvironment permissive for ASC proliferation and adipogenesis under a high-fat diet condition. Nonetheless, it is unclear whether the pro-adipogenic microenvironment of VIS WAT is conferred by paracrine cytokines, ECM molecules, or physical interaction between stromal cells and other cell types. Our data suggest that TCF21, which is specifically expressed in VIS ASCs, is responsible for the expression of IL6 along with ECM remodeling genes in VIS ASCs. It is likely that the heightened expression of IL6 coupled with type IV collagen deposition and type I collagen degradation contributes to a unique VIS WAT microenvironment permissive for inflammatory adipose tissue expansion in vivo (22,23).
IL6 is expressed most abundantly in adipose tissues, followed by the heart, kidney, and spleen (24). In parallel, TCF21 expression is found not only in adipose tissues, but also in the heart, kidney, and spleen. TCF21 plays an indispensable role for the development of these organs (15)(16)(17). Our study shows the role of TCF21 in regulating IL6 expression and ECM remodeling in white adipose tissues; however, it is conceivable that TCF21 may play a similar role in regulating IL6 expression and ECM remodeling in the vital organs, such as the heart and kidney. The potential role for TCF21 in defining pro-inflammatory environment and ECM remodeling may need to be further investigated in these organs.
The role of IL6 in obesity has been somewhat controversial. Although visceral WAT-derived IL6 is correlated with the presence of insulin resistance (4), skeletal muscles respond to IL6 favorably by improving whole-body glucose metabolism (25). In mice, genetic loss of IL6 paradoxically increases the risk of age-dependent obesity (26), insulin resistance, and steatohepatitis (27). Our findings suggest that a bHLH transcriptional factor, TCF21, is essential for IL6 expression in VIS ASCs. The expression of TCF21 and IL6 in visceral WAT did not change during a short-term, high-fat diet challenge in mice (data not shown); however, macrophages infiltrating WAT in chronic obesity express high levels of IL6 in the late stage of high-fat diet-induced obesity (28). Despite these findings, the specific contribution of stromal versus macrophage-derived IL6 to insulin resistance in the early and late stages of obesity remains unclear. Our study has not addressed whether TCF21-dependent expression of stromal IL6 expression is metabolically beneficial or detrimental for glucose metabolism in obesity. Nonetheless, because TCF21 and IL6 are expressed in VIS WAT under both physiological and obesogenic conditions, it is likely that TCF21-dependent IL6 expression and ECM remodeling may play an active role in defining visceral WAT function in both development and obesity.

Animals
C57BL/6J and 129SvEv male mice 8 -12 weeks of age were purchased from the Jackson Laboratory (Bar Harbor, ME) and Taconic (Rensselaer, NY), respectively. The animals were housed in a pathogen-free environment with 12-h light-dark cycle and free access to food and water. Animal studies and procedures were approved by the University of Michigan Institutional Animal Care and Use Committee.

TCF21 in adipose IL6 and ECM remodeling Isolation of Sca1 high ASCs
Mouse primary vascular stromal cells were isolated as described previously (9). VIS and SQ WATs were isolated, minced, and digested by 5 mg/ml type 3 collagenase (Worthington Biochemical Corp., Freehold, NJ) in Hanks' balanced salt solution with calcium and magnesium (Invitrogen/Thermo Fisher Scientific) at 37°C for 20 min. DMEM containing 10% FBS was added to the cell suspension to inactivate collagenase, and then cells were passed through a 100-m cell strainer. After centrifugation for 10 min at 1,500 rpm (412.5 ϫ g), the pellet was resuspended in 5 ml of distilled water for 30 s to lyse red blood cells. After adding growth medium (DMEM containing 10% FBS, 20 mM glutamine, 100 units/ml penicillin, 100 units/ml streptomycin, and 250 ng/ml amphotericin B), the cell suspension was passed through a 100-m cell strainer again and centrifuged for 10 min at 1,500 rpm and seeded onto a culture plate. The cultured primary cells were applied to MACS separator (Miltenyi Biotec, Auburn, CA) to enrich for Sca1 high ASCs as previously reported (9,29).

Cell culture
Isolated mouse ASCs and human primary preadipocytes from VIS and SQ WAT (Lonza, Fair Lawn, NJ) were cultured in DMEM containing 10% FBS.

siRNA treatment, lentiviral transduction, IL6/IL6-neutralizing antibody treatment, and gene expression analysis
For TCF21 knockdown, ASCs were transfected with control siRNA (siControl), CCUGCGGUAGAUGAACCAUUCA-CAA, siRNA against Tcf21 (siTcf21) #1, CCUCAGCGAUGU-AGAAGACCUUCAA, or siTcf21 #2, CCGGCAAACCAGAG-AATGACCTGAA. Briefly, siRNA (10 nM) was incubated with Lipofectamine RNAiMAX (Invitrogen) in Opti-MEM (Gibco/ Thermo Fisher Scientific) and added to cell suspension at the beginning of cell culture as per the manufacturer's recommendation. For TCF21 overexpression, lentivirus carrying TCF21 gene in pLenti-GIII-CMV-GFP-2A-Puro vector (Applied Biological Materials Inc., Richmond, Canada) was transduced with 8 g/ml Polybrene (American Bioanalytical, Natick, MA) for 8 h twice. GM6001 (10 M final, Millipore-Sigma) was added to culture for 2 days. VIS and SQ ASCs were incubated in the presence or absence of 100 ng/ml IL6-neutralizing antibody (catalog no. MAB406, R&D systems) and 30 ng/ml recombinant mouse IL6 (Cell Signaling) in serum-free DMEM supplied 1 g/ml. RNA was extracted from cultured cells using RNeasy mini kit (Qiagen) and subjected to reverse transcription with SuperScript II (Invitrogen). cDNA samples were then subjected to qPCR by using either Power SYBR green or Universal Taq-Man Mastermix in StepOnePlus (Applied Biosystems/Thermo Fisher Scientific). Relative cDNA quantities were normalized to the expression of a housekeeping gene, 36B4 (Rplp0), and are shown as fold change relative to control. Sequences of primers and TaqMan probes are shown in Table S1.

Promoter activity assay
Synthesized human IL6 promoter region (Ϫ1479 to ϩ21) was cloned into pGL3-basic plasmid (Promega, Madison, WI). Synthesized human TCF21 ORF (GenBank TM accession no. NM_198392) was cloned into pmCherry-C1 (TaKaRa Clontech, Kyoto, Japan). Amino acids 78 -132 in TCF21 were predicted as basic helix-loop-helix by BLAST and was eliminated TCF21 in adipose IL6 and ECM remodeling by inverse PCR using 5Ј-GGGGAAGCAGCAGATCCT-GGCTAACGACA-3Ј (forward) and 5Ј-CCAGGATCTGC-TGCTTCCCCTCCTGGCTGA-3Ј (reverse) primers and a QuikChange site-directed mutagenesis kit (Agilent Technologies, Palo Alto, CA). mCherry-tagged protein expression was confirmed by transfecting the plasmid into COS7 cells followed by Western blotting with the cell lysates. For promoter activity assay, 500 ng of pmCherry-TCF21 and 500 ng of pGL3-IL6 promoter or their empty vector were transfected together with 20 ng of pRL-TK Renilla luciferase plasmid (Promega) and 2.5 l of Lipofectamine 2000 (Invitrogen) into CHO cells in each well of 12-well plate. Two days later, firefly and Renilla luciferase activity was assayed by using a dual-luciferase reporter assay system (Promega) and a Synergy NEO microplate reader (BioTek Instruments, Inc., Winooski, VT).

Statistical analysis
All data were statistically analyzed using Student's t test or paired t test, one-way ANOVA with post hoc multiple comparison by Tukey's procedure, or two-way ANOVA with post hoc multiple comparison by Sidak's procedure as specified in the figure legends.