Long noncoding RNA ANCR inhibits the differentiation of mesenchymal stem cells toward definitive endoderm by facilitating the association of PTBP1 with ID2

The generation of definitive endoderm (DE) cells in sufficient numbers is a prerequisite for cell-replacement therapy for liver and pancreatic diseases. Previously, we reported that human adipose-derived mesenchymal stem cells (hAMSCs) can be induced to DE lineages and subsequent functional cells. Clarifying the regulatory mechanisms underlying the fate conversion from hAMSCs to DE is helpful for developing new strategies to improve the differentiation efficiency from hAMSCs to DE organs. Long noncoding RNAs (lncRNAs) have been shown to play pivotal roles in developmental processes, including cell fate determination and differentiation. In this study, we profiled the expression changes of lncRNAs and found that antidifferentiation noncoding RNA (ANCR) was downregulated during the differentiation of both hAMSCs and embryonic stem cells (ESCs) to DE cells. ANCR knockdown resulted in the elevated expression of DE markers in hAMSCs, but not in ESCs. ANCR overexpression reduced the efficiency of hAMSCs to differentiate into DE cells. Inhibitor of DNA binding 2 (ID2) was notably downregulated after ANCR knockdown. ID2 knockdown enhanced DE differentiation, whereas overexpression of ID2 impaired this process in hAMSCs. ANCR interacts with RNA-binding polypyrimidine tract-binding protein 1 (PTBP1) to facilitate its association with ID2 mRNA, leading to increased ID2 mRNA stability. Thus, the ANCR/PTBP1/ID2 network restricts the differentiation of hAMSCs toward DE. Our work highlights the inherent discrepancies between hAMSCs and ESCs. Defining hAMSC-specific signaling pathways might be important for designing optimal differentiation protocols for directing hAMSCs toward DE.


Introduction
Definitive endoderm (DE) is derived the from mesendoderm 1,2 and is a crucial stage in early embryogenesis and can commit to organs including the thyroid, lungs, pancreas, liver, and intestines 3 . The generation of DE and DE-derived lineages from a variety of genetic backgrounds would be beneficial not only for regenerative medicine applications but also for drug testing and toxicology studies.
Numerous protocols have been set up to direct embryonic stem cells (ESCs) or induced pluripotent stem cells (iPSCs) to differentiate into DE, providing valuable paradigms for the design of induction strategies 4,5 . However, the ethical/legal issues, safety concerns, and risks of teratoma formation hinder their utilization as starting materials in translational medicine. Adult stem cells, such as mesenchymal stem cells (MSCs), hold great promise for clinical applications due to their easy isolation, freedom from ethical issues, hypoimmunogenicity, and multipotent differentiation capacity 6,7 . MSCs derived from adipose, bone marrow, dental pulp, and umbilical cord have been reported to successfully differentiate into DE [8][9][10] , hepatocyte-like [11][12][13][14] and β-cell-like 8,15,16 cells. However, the existing differentiation protocols for directing adult stem cells to DE have mostly been developed by directly mimicking embryonic development programs from primitive streak to DE and its derivative's lineages. Inherent differences between ESCs/iPSCs and MSCs imply that they may undergo different processes toward DE. Thus, understanding the molecular mechanisms specific for DE differentiation from MSCs might help to create more desirable procedures to obtain DE-derived lineages in vitro.
Long noncoding RNAs (lncRNAs) are a subset of RNAs that are longer than 200 nt, have mRNA structure, and are usually polyadenylated but rarely encode proteins 17 . Increasing studies have reported that lncRNAs play multiple roles in the regulation of development and differentiation. Recent studies have confirmed that lncRNAs participate in the differentiation of all three germ layers, including in muscle 18 , lung 19 , dendritic cells 20 , cardiovascular 21 , neural 22 , and epidermal tissue 23 . LncRNAs also have been reported to play critical roles in the DE differentiation of ESCs. For example, lncRNA DENAR1 functions in human endoderm differentiation by regulating FOXA2 expression 24 . The lncRNA DIGIT, a conserved lncRNA in mouse and human, regulates GSC expression to promote DE differentiation in ESCs 25 . To date, little is known about the functions of lncRNAs in the differentiation of MSCs to DE. In this study, we profiled lncRNA expression during human adipose tissue-derived mesenchymal stem cells (hAMSCs) differentiation into DE and found a set of DE differentiation-related lncRNAs. Among them, the lncRNA ANCR (antidifferentiation ncRNA, or DANCR) was significantly downregulated. ANCR was previously found to promote progenitor maintenance and prevent differentiation in epidermal progenitors 23 , osteoblasts 26,27 , and chondrogenesis 28,29 . However, the role of ANCR in the fate conversion of hAMSCs toward DE remains to be discovered. Herein, we provide evidence that ANCR could inhibit the differentiation of hAMSCs to DE by increasing the mRNA stability of ID2 through facilitating its binding with polypyrimidine tract-binding protein 1 (PTBP1).

Results
ANCR was dramatically downregulated during the differentiation of hAMSCs to DE We previously established a stepwise protocol using the combination of Activin A and Wnt3a to generate DE from hAMSCs 8,30 . Teo et al. 31 reported that compared to high doses of Wnt3a, the glycogen synthase kinase-3 inhibitors Chir99021 can induce DE formation from ESCs with comparable efficiency and lower cost. Thus, we set to determine whether Chir99021 could replace Wnt3a in our protocol. As shown in Supplementary Fig. 1, the combination of 5 ng/ml Activin A and 0.3 mM Chir99021 (AC) exhibited a higher expression of key DE marker genes, including SOX17 and FOXA2, compared to our previous combination of 5 ng/ml Activin A and 10 ng/ml Wnt3a (AW). Therefore, we adopted the AC protocol for the induction of DE in the subsequent study. As verified by qRT-PCR, the expression of the DE-specific genes SOX17, FOXA2, and CXCR4 5 were remarkably elevated during DE differentiation (Fig. 1a). Meanwhile, the expression of mesendoderm-related genes, such as EOMES and GSC, were also upregulated (Fig. 1a), while the ectoderm marker PAX6 as well as the mesoderm marker KDR were downregulated (Fig. 1a). Western blot also confirmed the upregulation of SOX17 and FOXA2 after DE induction in hAMSCs (Fig. 1b). Immunofluorescence staining (IF) revealed that double-positive FOXA2/SOX17 cells appeared after DE induction (Fig. 1c). Altogether, these data demonstrated that the AC protocol is effective in converting hAMSCs toward DE, as we reported previously 8,30 .
To identify functional lncRNAs involved in DE differentiation, we performed microarray analyses on days 0, 3, and 5 after DE induction in hAMSCs. Gene Ontology (GO) enrichment analysis of the differentially expressed coding genes indicated that genes involved in extracellular matrix organization, digestive tract development, and the Wnt signaling pathway were significantly changed (Fig. 1d). Next, we analysed the lncRNA expression profiles according to stringent criteria (fold change ≥2, expression value ≥3, and p value < 0.05). We identified 75 lncRNAs (28 upregulated and 47 downregulated) that were differentially expressed in DE cells versus hAMSCs (Fig. 1e). Among the top downregulated lncRNAs in hAMSCs, we noticed that the expression of the lncRNA ANCR, previously reported to play important role in restricting epithelium differentiation, was dramatically decreased during DE differentiation. qRT-PCR also confirmed that ANCR expression levels were decreased in the induced cells (Fig. 1f).
We next induced ESC differentiation toward DE cells using a well-established protocol 4 and examined the expression of ANCR during this progress. The induction efficiency was confirmed by qRT-PCR and IF staining ( Supplementary  Fig. 2). We found that ANCR expression levels were continuously reduced during the differentiation of ESCs toward DE (Fig. 1g). Thus, we focused on the role of ANCR in the generation of DE from hAMSCs as well as ESCs.

ANCR knockdown improved the differentiation of hAMSCs to DE
To evaluate the effects of ANCR in the generation of DE, we silenced ANCR in hAMSCs using two pairs of specific siRNAs (Fig. 2a). After 5 days of induction, ANCR c Immunofluorescence (IF) staining for DE markers (SOX17 and FOXA2) in ANCR-knockdown hAMSCs or control hAMSCs after DE induction. The nuclei were stained with Hoechst 33342. Scale bar = 100 μm. d Gene Ontology (GO) analysis on the commonly upregulated (top) or downregulated (bottom) coding genes in three donors derived from hAMSCs after DE induction at day 5 compared to hAMSCs at day 0. The y-axis shows the GO terms and the x-axis shows the statistical significance (negative logarithm of p value). e Hierarchical clustering of significantly changed lncRNA on day 3 or 5 after induction compared with day 0 in matched hAMSCs from three donors. f, g qRT-PCR analysis of ANCR levels in hAMSC (f) and ESC (g) at the indicated time points after DE induction. Data are shown as the means ± S.D. (n = 3). *p < 0.05, **p < 0.01, and ***p < 0.001 knockdown resulted in higher expression of the DE markers SOX17, FOXA2, and CXCR4 compared with control siRNA-transfected cells (Fig. 2b). Flow cytometry analysis also indicated that the FOXA2/SOX17 doublepositive population were higher in ANCR-downregulated cells (Fig. 2c). Western blot analysis consistently validated that ANCR depletion enhanced the expression of the DE markers FOXA2 and SOX17 after DE induction (Fig. 2d).
To investigate whether ANCR plays a similar role in the differentiation of ESCs towards DE, we silenced ANCR in Fig. 2 ANCR knockdown improved the differentiation of hAMSCs to DE. a ANCR was silenced in hAMSCs using two independent siRNAs (si-ANCR-1 and si-ANCR-2). The knockdown efficiency was verified by qRT-PCR compared with the negative control (NC). b qRT-PCR analysis detected DE marker genes (SOX17, FOXA2, and CXCR4) in ANCR-knockdown hAMSCs or control hAMSCs on day 5 after DE induction. c Flow cytometry analysis of the FOXA2 + /SOX17 + subpopulation in ANCR-knockdown hAMSCs or control hAMSCs after DE induction. d Western blot detected DE markers (SOX17 and FOXA2) in ANCR-knockdown hAMSCs or control hAMSCs on day 5 after DE induction. e ANCR was silenced in ESCs using two independent siRNAs (si-ANCR-1 and si-ANCR-2). The knockdown efficiency was verified by qRT-PCR compared with the negative control (NC). f, g qRT-PCR analysis of detected stem cell markers genes (OCT4 and NANOG) from ESCs (f) and DE marker genes (SOX17 and FOXA2) after DE induction (g) in ANCR-knockdown ESCs or control ESCs. Data are shown as the means ± S.D. (n = 3). *p < 0.05, **p < 0.01, and ***p < 0.001 ESCs with the above siRNAs (Fig. 2e). We found that ANCR knockdown neither changed the expression of the stemness markers OCT4 and NANOG in ESCs nor the DE markers SOX17 and FOXA2 after DE induction compared to control cells ( Fig. 2f, g). These data demonstrated that ANCR hampered the differentiation of DE in hAMSCs but not in ESCs, which may be because different pathways work in ESCs and hAMSCs. Likewise, BMP and bFGF, which are required for the generation of DE from ESCs 4,32-34 , were dispensable or detrimental to the generation of DE from hAMSCs ( Supplementary Fig. 3).

Overexpression of ANCR inhibited the differentiation of hAMSCs to DE
To further confirm the role of ANCR in the differentiation of hAMSCs to DE, we stably overexpressed fulllength ANCR in hAMSCs using lentivirus (Fig. 3a). Ectopic expression of ANCR significantly impaired the expression of the DE markers SOX17, FOXA2, and CXCR4, as verified by qRT-PCR (Fig. 3b). The western blot assay revealed that overexpression of ANCR (Lenti-ANCR) notably delayed DE differentiation compared to empty vector control-infected cells (Lenti-NC), as indicated by decreased protein levels of the DE markers SOX17 and FOXA2 (Fig. 3c). Flow cytometry analysis also indicated that the FOXA2/SOX17 double-positive population was reduced in ANCR-overexpressing cells (Fig. 3d). Collectively, these data indicated that ANCR plays a negative role in the differentiation of hAMSCs toward DE.
ANCR regulated ID2 mRNA stability To investigate the mechanism by which ANCR impacted DE differentiation in hAMSCs, we performed microarray analysis to compare the gene expression profiles in ANCR-knockdown hAMSCs and control cells. The data showed that 172 genes were differentially expressed after ANCR knockdown (Fig. 4a) (fold change ≥2 and p < 0.05). GO enrichment analysis revealed that the differentially expressed genes were significantly enriched for endoderm differentiation (Fig. 4b). Among the most significantly changed genes, we focused on inhibitor of DNA binding 2 (ID2), which was notably downregulated after ANCR knockdown (Fig. 4a). The ID family has been reported to play key roles in fate determination, inhibition of differentiation, and maintenance of self-renewal in multipotent stem cells 35,36 . We then aimed to validate whether ANCR could regulate ID2 levels in hAMSCs. As shown in Fig. 4c, ANCR silencing notably reduced the expression of ID2 and ANCR overexpression slightly upregulated its expression. The western blot assay also verified the positive correlation between ANCR and ID2 (Fig. 4d).
Specific subcellular localization and cell fractionation are essential for understanding the function and mechanism of lncRNAs. Since ANCR largely displayed a cytoplasmic distribution (>90%) (Fig. 4e), we speculate that ANCR might promote downstream effectors at the posttranscriptional level. Thus, we next determined The y-axis shows GO terms and the x-axis shows statistical significance (negative logarithm of p value). c qRT-PCR analysis of ID2 in hAMSCs transfected with si-ANCR versus si-NC (left) or in hAMSCs overexpressing ANCR (Lenti-ANCR) versus empty vectors (Lenti-NC) (right). d Western blot detected the expression of ID2 in hAMSCs with ANCR knockdown (left) and overexpression (right). e Fractionation of ANCR in hAMSCs followed by qRT-PCR. GAPDH served as cytoplasmic mRNA control. U1 served as a nuclear RNA control. f mRNA decay detected by qRT-PCR in ANCR-knockdown hAMSCs (top), overexpression hAMSCs (bottom), and their corresponding controls. Data are shown as the means ± S.D. (n = 3). *p < 0.05, **p < 0.01, and ***p < 0.001 whether ANCR could regulate ID2 mRNA decay. To evaluate the effects of ANCR on ID2 mRNA stability, hAMSCs transduced with Lenti-ANCR or control vector (Lenti-NC) were treated with actinomycin D (5 μg/ml), which inhibits RNA polymerase and blocks transcription. The half-life of ID2 mRNA was shorter than 2h based on our preliminary results (data not shown). Knockdown of ANCR resulted in a shortening in the half-life of ID2 mRNA, whereas overexpression of ANCR prolonged the ID2 mRNA half-life (Fig. 4f), indicating that ANCR stabilized ID2 mRNA. Collectively, these data revealed that ANCR positively regulated the expression of ID2 at least partly by regulating its mRNA stability.

ID2 negatively regulated the differentiation of hAMSCs toward DE
To investigate the role of ID2 during the differentiation of hAMSCs to DE, we first silenced the expression of ID2 using two siRNAs against ID2 (Fig. 5a). Knockdown of ID2 resulted in an elevated induction of the DE markers SOX17, FOXA2, and CXCR4 after DE induction compared to control cells (Fig. 5b). Consistently, The western blot assay confirmed that ID2 depletion was associated with increased expression of DE markers (Fig. 5c). Moreover, flow cytometry analysis also showed that the percentage of FOXA2/SOX17 double-positive cells was elevated in ID2-depleted cells after DE induction (Fig. 5d).
We then stably overexpressed ID2 in hAMSCs using lentivirus transduction and verified the results by qRT-PCR (Fig. 5e). After DE induction, the expression levels of the DE markers SOX17, FOXA2, and CXCR4 were reduced in ID2-overexpressing hAMSCs (Lenti-ID2) compared with control cells (Lenti-NC) (Fig. 5f). Consistently, the protein levels of SOX17 and FOXA2 were decreased in ID2-overexpressing cells after DE induction (Fig. 5g). Flow cytometry analysis also confirmed that DE differentiation was suppressed after ID2 overexpression (Fig. 5h). Collectively, these data demonstrated that ID2 negatively regulated the DE differentiation of hAMSCs, which was consistent with the function of ANCR.

ID2 was responsible for ANCR-mediated DE differentiation in hAMSCs
To examine whether ANCR regulated DE differentiation of hAMSCs in an ID2-dependent manner, we overexpressed ID2 in ANCR-knockdown hAMSCs or control cells (Fig. 6a). As mentioned above, ANCR knockdown resulted in higher expression of DE markers, whereas ID2 overexpression could significantly decrease DE marker expression. Overexpression of ID2 could decrease DE marker expression even under ANCR knockdown, indicating that ID2 strongly inhibited DE differentiation. The promoting effects of ANCR knockdown on DE differentiation of hAMSCs were partially reversed when ID2 was overexpressed, as indicated by the downregulated the expression of key DE markers (Fig. 6b, c). These data demonstrated that ANCR regulates hAMSCs differentiation to DE in an ID2-dependent manner.
ANCR regulates ID2 mRNA stability by binding to PTBP1 LncRNAs often regulate target genes through interactions with specific protein partners 37,38 . To explore the underlying mechanism whereby ANCR affects ID2 mRNA stability, we conducted RNA pull-down assays in hAMSCs to identify ANCR-interacting proteins. Biotinylated sense and antisense ANCR were synthesized and incubated with hAMSC extracts. A discrepant band between sense and antisense lanes (red arrow) was excised, digested, and subjected to mass spectrometry (LC-MS) (Fig. 7a). We chose candidates according to LC-MS scores and validated the interaction by pull-down western blot. The results showed that PTBP1 was identified as a binding partner of ANCR (Fig. 7b). PTBP1 (also known as PTB or hnRNP I) is an RNA-binding protein that regulates mRNA stability and pre-mRNA splicing 39 . In contrast, Vimentin, which received the highest score in the LS-MS assay, did not differentially bind to sense and antisense of ANCR (Fig. 7b).
To further verify the interaction between PTBP1 and ANCR, we performed an RNA immunoprecipitation (RIP) assay in hAMSCs. As shown in Fig. 7c, PTBP1 antibody (anti-PTBP1) significantly precipitated ANCR as well as ID2 compared with an anti-IgG control. Using siRNA silencing to decrease the expression of PTBP1 resulted in decreased levels in both ANCR and ID2 (Fig. 7d). Moreover, knockdown of PTBP1 in hAMSCs could decrease the mRNA stability of ID2 and ANCR when treated with actinomycin D (5 μg/ml) (Fig. 7e). Interestingly, biotinlabelled ID2 mRNA can also pull down PTBP1 (Fig. 7f). More importantly, overexpression of ANCR can significantly increase the binding of PTBP1 and ID2 mRNA (Fig. 7f). These data collectively revealed that ANCR physically binds to PTBP1 and enhanced the interaction between PTBP1 and ID2 mRNA.

Discussion
The generation of DE cells with high safety and efficiency is a prerequisite for potential clinical applications. We previously reported a protocol to induce adult adipose-derived MSCs differentiation toward DE and their derived functional cells 8,30 . In this study, we optimized the induction protocol and identified key lncRNAs during the process. We demonstrated that ANCR played an important role in the fate conversion of hAMSCs toward DE. ANCR could regulate ID2 mRNA stability by binding to PTBP1 and subsequently suppress the differentiation of hAMSCs into DE in an ID2-dependent manner.
The lncRNA ANCR was first characterized as a functional suppressor that represses differentiation in epidermal progenitors 23 . ANCR has also been reported to modulate osteogenic and adipogenic differentiation in stem cells derived from different tissues [40][41][42] . In hepatocellular carcinoma, ANCR/DANCR was reported to increase stemness features and contribute to tumor progression 43 . Herein, we provided evidence that ANCR plays a significant role in suppressing the differentiation of hAMSCs toward DE. We demonstrated that ANCR can affect ID2 mRNA stability during the differentiation of hAMSCs into DE. Members of the ID family are negative regulators of transcription factors with a basic helix-loop-helix motif 36 . It has been widely observed that ID proteins are abundant in proliferating multipotent cells including stem cell populations, but low or undetectable in differentiated and senescent cells 44,45 . Accumulating studies have revealed that ID proteins play key roles in lineage commitment, cell fate decisions, and oncogenic transformation in cancer 44,[46][47][48][49][50] . In MSCs, blocking the degradation of ID proteins through the deubiquitinating enzyme USP1 led to inhibited osteoblastic differentiation and enhanced proliferation 51 . In this study, we found that ID proteins are enriched in hAMSCs and play a negative role in the fate conversion of hAMSCs, which is consistent with the function of ANCR. Knockdown of ID2 by siRNAs promoted DE differentiation from hAMSCs, while ectopic expression of ID2 exerted opposite effects. Moreover, overexpression of ID2 could significantly reverse the promoting effects of ANCR downregulation on DE differentiation in hAMSCs. The precise mechanism by which ID2 affects the differentiation of hAMSCs to DE remains to be determined in the future. PTBP1, which shuttles between the nucleus and cytoplasm, is a multifunctional protein involved in all steps of RNA biogenesis 52 . In the nucleus, PTBP1 forms ribonucleoprotein complexes and regulates alternative splicing, polyadenylation, and mRNA export. In the cytoplasm, PTBP1 participates in localization, translation initiation, and mRNA stability 39 . Here, we demonstrated that lncRNA ANCR is mostly distributed in the cytoplasm and could form a complex with PTBP1. We found that PTBP1 knockdown decreased the RNA levels of both ID2 and ANCR. Moreover, ANCR overexpression can significantly increase the binding of PTBP1 and ID2 mRNA. Our findings revealed that PTBP1 is involved in regulating the effects of ANCR on ID2. Mechanistically, ANCR functions as an RNA scaffold to recruit PTBP1 to ID2, which modulates ID2 mRNA stability. Consistently, several lncRNAs have been reported to function in a similar way. LncRNA H19 interacts with PTBP1 to facilitate its association with SREBP1c mRNA and protein, leading to increased stability 53 . LncRNA UCA1 regulates heme biosynthesis and erythrocyte development by recruiting PTBP1 to ALAS2 mRNA 54 .
Interestingly, the expression of ANCR is abundant in both ESCs and hAMSCs and remarkably decreased during differentiation into DE. ANCR knockdown in hAMSCs enhanced the differentiation efficiency of hAMSCs towards DE. However, ANCR depletion in ESCs did not cause similar effects as in hAMSCs. Currently, the precise Fig. 6 ID2 was responsible for ANCR-mediated DE differentiation in hAMSCs. a Lenti-ID2 hAMSCs and Lenti-NC hAMSCs were transfected with siRNAs targeting ANCR or NC. The levels of ANCR and ID2 were detected by qRT-PCR. b qRT-PCR analysis detected DE marker genes (SOX17, FOXA2, and CXCR4) in Lenti-ID2 hAMSCs and Lenti-NC hAMSCs transfected with siRNAs targeting ANCR or NC, respectively, on day 5 after DE induction. c Western blot detected the expression of SOX17 and FOXA2 in Lenti-ID2 hAMSCs and Lenti-NC hAMSCs that transfected with siRNAs targeting ANCR or NC, respectively, on day 5 after DE induction. Data are shown as the means ± S.D. (n = 3). *p < 0.05, **p < 0.01, and ***p < 0.001 role of ANCR in the regulation of pluripotency and differentiation in ESCs remains unclear. It is suggested that hAMSCs and ESCs might go down different paths and utilize different signaling pathways during the generation of DE. In support of this opinion, we found that BMP and bFGF, which are required for the generation of DE from ESCs 4,32-34 , were dispensable or detrimental to the generation of DE from hAMSCs. BMP signaling promotes DE formation while simultaneously suppressing pluripotency in ESCs/iPSCs 31 . bFGF promotes the epithelial to mesenchymal transition (EMT) 55 , which is a critical step during the acquisition of DE from iPS 56 . We speculated that exit from pluripotency and EMT progress were crucial for DE generation from ESCs/iPSCs but not MSCs. What's more, the optimal concentration of Activin A for directing MSCs towards DE is relatively low (5 ng/ml), Fig. 7 ANCR regulates ID2 mRNA stability by binding to PTBP1. a In vitro the RNA pull-down assay. hAMSCs lysates were incubated with biotinlabelled sense or antisense ANCR RNAs. After pull down, the proteins were subjected to SDS-PAGE and staining. The band indicated by the arrow was subjected to mass spectrometry. b Western blot analysis for PTBP1 or Vimentin following RNA pull down with biotin-labelled sense or antisense ID2 or ANCR. Antisense RNAs incubated with hAMSCs were used as nonspecific controls. c The RIP assay for PTBP1 enriched with ANCR (left) and ID2 (right) in hAMSCs. IgG was used as a negative control. All relative abundances were compared to 1% input. d PTBP1 was silenced in hAMSCs using two independent siRNAs (si-PTBP1-1 and si-PTBP1-2). The knockdown efficiency and expression of ANCR and ID2 were verified by qRT-PCR compared with si-NC. e hAMSCs transfected with si-PTBP1s or si-NC were treated with actinomycin D (5 μg/mL) and RNA was extracted at different time points (0, 2, and 4 h). The levels of ANCR and ID2 were analysed by qRT-PCR and normalized to GAPDH. mRNA at 0 h served as a reference. f The RNA pull down assay to determine the interaction between PTBP1 protein and ID2 mRNA in Lenti-Ctrl or Lenti-ANCR hAMSCs. Data are shown as the means ± S.D. (n = 3). *p < 0.05, **p < 0.01, and ***p < 0.001 while the optimal concentration for ESCs or iPSCs is high (100 ng/ml) 8 . Most currently available protocols are based on mimicking the DE formation signaling pathway from embryonic development. Our study highlights the importance of initial cell and developing cell sourcespecific protocols when designing induction schemes.
Collectively, we demonstrate for the first time that the lncRNA ANCR can negatively regulate hAMSC differentiation toward DE by binding with PTBP1, enhancing the interaction between PTBP1 and ID2 mRNA and subsequently increasing ID2 mRNA stability. LncRNAs are ideal targets for small molecules and nucleic acids because of their specific expression patterns and unique sequences and secondary structure 57 . Further identification of functional lncRNAs as well as cell type-specific signaling in the generation of DE cells will help to develop new strategies to enhance the efficiency of cell fate conversion or differentiation.

Materials and methods
Isolation, culture, and differentiation of hAMSCs hAMSCs were isolated from human adipose tissues obtained from donors undergoing liposuction according to our previous studies 8,30 . hAMSCs at passage 3 were used in our experiments. All experiments and procedures were approved by the Ethics Committee at the Chinese Academy of Medical Sciences and Peking Union Medical College.
For DE differentiation, as described previously 8 and in Supplementary Fig. 1a, hAMSCs at passage 3 were seeded in six-well plates in regular culture medium. The next day, the cells were changed to differentiation basic medium DMEM (Gibco, Grand Island, NY) supplemented with 0.5% FBS (Gibco, Grand Island, NY), 5 ng/ml Activin A (Peprotech, USA), and 50 ng/ml Wnt3a (Peprotech, USA)(AW) or 0.3 µM Chir99021(AC) on the first days, with Wnt3a or Chir99021 being withdrawn on the following 4 days. For the comparison with ESC protocol published by Hannan et al. 4 (H), BMP4 (10 ng/ml) and bFGF (10 ng/ml) were added to the differentiation system on the first two days.

Culture and differentiation of ESCs
The embryonic stem cell line H9 was kindly provided by Dr Du Mingxia (Peking Union Medical College, Beijing, China). H9 cells were cultured on Matrigel (BD Biosciences, San Jose, CA) with a mTeSR1 Complete Kit (Stem Cell Technology, Canada) and passaged every 3-5 days using 0.05 mM EDTA (Cell APY, Beijing, China) according to the manufacturer's instruction.

Microarray analysis
To identify DE-related lncRNAs, hAMSCs were exposed to DE induction medium as described above, and total RNA was isolated using the Trizol reagent (Invitrogen, USA) on days 0, 3, and 5. hAMSCs from three donors were analysed in this study. Sample processing and hybridization were conducted by Cnkingbio Biotechnology (China) with Affymetrix mRNA + lncRNA microarray chips. Briefly, a fold change ≥ 2 (expression value ≥3 and p value < 0.05, day 3 and 5 versus day 0) were chosen as the cutoff criteria for differentially expressed genes. Overlapping DEGs in all three donors were used in GO enrichment analyses.
For ANCR-affected genes, RNA was extracted from hAMSCs transfected with si-NC (negative control), si-ANCR1, and si-ANCR2 for 48 h. hAMSCs from two donors were analysed in this study. A fold change ≥2 over the control, an expression value ≥4 and a p value < 0.05 were chosen as the cutoff criteria for differentially expressed genes.
siRNA and lentivirus infection siRNAs used to knockdown target lncRNAs or mRNAs were designed using online tools (BLOCK-iT™ RNAi Designer) and synthesized by the RIBOBIO company (Suzhou, China). For transfection of the siRNAs, Lipofectamine 2000 (Life Technology, USA) was used according to the manufacturer's recommendations.
For overexpression, full-length ANCR and ID2 CDS were inserted into the pEZ-LV225 lentivirus expression vector and packaged by GeneCopoeia TM . hAMSCs were infected with viral precipitates at an MOI of 10 and stable cell lines were established by puromycin treatment.

RNA extraction and qRT-PCR analysis
Total RNA was extracted using the Trizol Regent (Invitrogen, USA), and 2 μg of RNA were reverse transcribed with oligo (dT) primer and M-MLV Reverse Transcriptase (Takara, Japan). qRT-PCR was performed on a QuantStudio™ Design & Analysis system (ABI, USA) with SYBR-Green Mastermix (YEASEN, China). The relative RNA levels were normalized to GAPDH using the 2 −ΔΔCt method. The primer sequences are listed in Supplementary Table 1.

Western blot
Protein was extracted using RIPA buffer with PMSF (1:100, Beyotime, China) and quantified with a BCA Protein Assay kit (Beyotime, China). Proteins in lysates were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes (0.22μm, Millipore, Danvers, MA, USA). The membranes were blocked with 5% milk for 1h at room temperature, incubated with primary antibody overnight at 4°C, and then incubated with horseradish peroxidase-conjugated secondary antibodies (1:3000, YEASEN, China) at room temperature for 1h. The proteins were detected using an ECL reagent (Millipore, USA).

Immunofluorescent staining
Cells were fixed with 4% paraformaldehyde, permeabilized in 0.3% Tween-100, blocked in PBS + 0.5% Tween-100 + 5% BSA, and then incubated with primary antibody overnight at 4°C. After washing three times with PBS, the sample was incubated with the corresponding secondary antibody at room temperature for 1 h, washed with PBS, and then incubated with Hoechst for 5min to dye the nuclei. The antibodies used in this study are summarized as follows.
The OCT4 and FOXA2 antibodies were obtained from Abcam (Cambridge, MA, USA); SOX17 was bought from Cell Signaling Technology (CST, USA); Hoechst was purchased from Solarbio (Beijing, China); and Alexa Fluor 488 goat-anti-rabbit secondary antibody and Alexa Fluor 594 goat-anti-mouse secondary antibody were bought from Thermo Fisher Scientific (Waltham, MA, USA).

The flow cytometry assay
The differentiated cells were analysed for cell antigen expression by flow cytometry using an Accuri C6 (BD Biosciences, San Jose, CA). In total, 1 × 10 5 cells were fixed in 4% paraformaldehyde, permeabilized with PBS + 0.2% Triton X-100 for 10 min, and then incubated at 4°C for 1 h with the following primary antibodies: SOX17, FOXA2, or isotype antibodies, which served as negative controls. Then, the cells were incubated with the following antibodies: goat-anti-rabbit-488 or goat-anti-mouse-594 at room temperature for 30 min.
mRNA stability analysis hAMSCs were firstly transfected with si-NC and si-ANCR or si-PTBP1 for 48 h. Then, the hAMSCs were treated with 5 μg/ml actinomycin D (MedChemExpress, NJ, USA). At different time points (0, 2, and 4 h), total RNA was extracted using the Trizol regent and ANCR or ID2 mRNA was analysed by qRT-PCR and normalized to GAPDH. mRNA at 0 h served as a reference.

Subcellular fractionation
The separation of the nuclear and cytosolic fractions was performed using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific, USA) according to the manufacturer's instructions. RNA was extracted, and qRT-PCR was performed to assess the relative proportion in the nuclear and cytoplasmic fractions.

RNA-pull down and mass spectrometry
Full-length ANCR and ID2 were synthesized and subcloned into the pCI-neo vector. Biotin-labelled RNAs were transcribed in vitro with the Biotin RNA Labeling Mix (Roche, Basel, Switzerland) and T7 RNA Pol II (NEB, USA). Whole-cell extracts prepared from 1 × 10 7 hAMSC cells (empty vector control or ANCR overexpression) in 1 ml of RIP buffer (150 mM KCl, 25 mM Tris (pH 7.4), 5 mM EDTA, 0.5 mM DTT, and 0.5% NP-40) containing RNase and protease inhibitors was mixed with 3 µg of biotinylated RNA and incubated at RT for 2 h, followed by the addition of 50 µl washed Streptavidin Dynabeads to incubate for another 1 h. After magnetic separation, beads were washed four times with ice-cold buffer and resuspended in 1× SDS sample buffer. The precipitated components were separated using SDS-PAGE, followed by silver staining. Differential bands were cut for mass spectrometry (LTQ Orbitrap XL).

The RIP assay
The RIP assay was performed with the EZ-Magna RIP Kit (Millipore, USA) according to the manufacturer's instructions. Anti-PTBP1 antibody was purchased from Proteintech (Wu Han, China). The coprecipitated RNAs associated with PTBP1 were extracted with the Trizol reagent, and ANCR and ID2 enrichment was examined using RT-qPCR. Enrichment associated with normal rabbit IgG served as controls.

Statistical analysis
GraphPad Prism7 (GraphPad Prism, San Diego, CA) software was used for all statistical analysis and expressed as mean ± standard. Student's t-test was used for statistical comparison between two groups. One-way ANOVA was used for comparison between multiple groups. Differences were considered statistically significant at *P < 0.05, **P < 0.01, and ***P < 0.001.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.