Human dental pulp tissues were isolated from healthy and non-functional third molars donated by patients aged 18–26 years old. In brief, healthy dental pulp tissues were collected and cut into tiny pieces in a sterile environment, and then digested with 3 mg/mL collagenase type I (#SCR103, Sigma-Aldrich, Saint Louis, MO, USA) for 1 h at 37 °C. The cells were cultured in minimal essential medium, alpha modification (#SH30625.01, Hyclone, Marlborough, MA, USA), which contains 10% fetal bovine serum (#10099141, Gibco, Carlsbad, CA, USA), 100 U/mL penicillin, and 100 mg/mL streptomycin (#15240-062, Gibco, Carlsbad, CA, USA). The surface stemness markers of hDPSCs were determined by flow cytometry. In brief, hDPSCs (3rd passage) were digested with trypsin and rinsed two times with phosphate-buffered saline (PBS) (#SH30256.01, Hyclone, Marlborough, MA, USA). Then, cells were resuspended with flow cytometry staining buffer (#S1001, MultiSciences, Hangzhou, China), followed by incubation with the following fluorescent dye-conjugated antibodies for 30 minutes: isotype control-anti-CD34 (#400207); anti-CD34 (#343603); isotype control-anti-CD45 (#400109); anti-CD45 (#368507); isotype control-anti-CD29, CD90, and CD146 (#400111); anti-CD29 (#303003); anti-CD90 (#328109); and anti-CD146 (#361005) (all antibodies were obtained from Biolegend, San Diego, CA, USA). The stained cells were analyzed by a CytoFLEX Flow Cytometer (#C09748, Beckman, Indianapolis, IN, USA). CD34 and CD45 are hematopoietic stem cells and leukocyte markers, whereas CD29, CD90, and CD146, are mesenchymal stem cell (MSC) markers. Analysis of data was performed with CytExpert software (Beckman, Indianapolis, IN, USA).

Mice were raised in a specific pathogen-free facility with a 12:12-h day/night cycle. Primary mDPCs were isolated from the first mandibular molars of C57BL/6 mice at postnatal day 0.5 (PN 0.5) and digested by 0.25% trypsin (#SH30042.02, Hyclone, Marlborough, MA, USA) for 10 min, and primary cells were fusiform or triangular under phase contrast microscope (Supplementary Fig. 1). CAL 27 cells, HEK 293 cells and HEK 293T cells were authenticated using short tandem repeat (STR) profiling analysis. Mycoplasma testing has been done for the cell lines used. The mDPCs, CAL 27 cells, HEK 293 cells, and HEK 293T cells, were grown in Dulbecco’s modified Eagle medium (DMEM) (#SH30022.01, Hyclone, Marlborough, MA, USA), which contains 10% fetal bovine serum.

To induce differentiation, hDPSCs and mDPCs were cultured in differentiation medium which comprised basal medium, 10 nM dexamethasone (#T5648, Sigma-Aldrich, Saint Louis, MO, USA), 50 µg/mL ascorbic acid (#A4544, Sigma-Aldrich, Saint Louis, MO, USA) and 10 mM $\beta$-glycerophosphate (#G9422, Sigma-Aldrich, Saint Louis, MO, USA). All cells were incubated at 37 °C in 5% CO${}_2$ humidified air.

T7-tagged human hnRNP A1 overexpression plasmid was made for our previous publication [21]. The mouse hnRNP A1 gene was amplified using primers 5${}^{\prime }$-ATGTCTAAGTCCGAGTCTCCCA-3${}^{\prime }$ and 5${}^{\prime }$-GAACCTCCTGCCACTGCCATA-3${}^{\prime }$. Then, the product was amplified with 5${}^{\prime }$-TTCCGAAGCTTGCCACCATGTCTAAGTCCGAGTCTCCCA-3${}^{\prime }$ and 5${}^{\prime }$-TTCCGGAATTCCCGAACCTCCTGCCACTGCCATA-3${}^{\prime }$, and cloned into the p3xFlag-CMV-14 at HindIII and EcoRI sites. The FLAG-fusion fragment was amplified with 5${}^{\prime }$-TTCCGACTAGTGCCACCATGTCTAAGTCCGAGTCTCCCA-3${}^{\prime }$ and 5${}^{\prime }$-TTCGCGGATCCCTACTTGTCATCGTCATCCTTG-3${}^{\prime }$, and cloned into the pLVX-IRES-Puro at SpeI and BamHI sites to obtain recombinant expression plasmids 3xFLAG-tagged hnRNP A1.

A minigene of human RUNX2 with genomic DNA sequence from exon 4 to exon 6 (including exon 4, exon 5, exon 6, and partial intron) was amplified from the CAL 27 genome by the following primers: 5${}^{\prime }$-CACCTTGACCATAACCGTCTTCAC-3${}^{\prime }$ and 5${}^{\prime }$-TACTAGAGCATGACCAAACAACAGTG-3${}^{\prime }$, used to amplified exon 4 and intron 4 (535 bp downstream of exon 4); 5${}^{\prime }$-AAAGAAAAGTCACTATGGTCATCGTG-3${}^{\prime }$ and 5${}^{\prime }$-AGCCACACATTGACCTTGCCT-3${}^{\prime }$, used to amplified intron 4 (494 bp upstream of exon 5), exon 5 and intron 5 (481 bp downstream of exon 5); 5${}^{\prime }$-CCCTCCATTCTCATTCCTGC-3${}^{\prime }$ and 5${}^{\prime }$-ATCGGTGATGGCAGGAAGC-3${}^{\prime }$, used to amplified intron 5 (617 bp upstream of exon 6). Then, the three fragments were cloned into the pEGFP-N1 vector using In-Fusion Cloning at EcoRI and BamHI sites. HEK 293 cells were transfected with plasmids using Lipofectamine 2000 (#11668019, Invitrogen, Carlsbad, CA, USA). HEK 293T cells were cotransfected with lentiviral backbone plasmids (T7-tagged hnRNP A1, 3xFlag-tagged hnRNP A1, or control vector pLVX-IRES-puro), packaging plasmid psPAX2, and envelope plasmids pMD2.G. The supernatants containing lentiviral particles were collected 48 h after transfection and clarified by a 0.45 µm filter. The hDPSCs or mDPCs were then transfected with supernatants.

The sequences of hnRNP A1 siRNAs are as follows: 5${}^{\prime }$-GUGGUAAUGAGAGAUCCAATT-3${}^{\prime }$ (siA1-1) and 5${}^{\prime }$-GCUCUUCAUUGGAGGGUUGTT-3${}^{\prime }$ (siA1-2) for human, and 5${}^{\prime }$-GUAUCCAUUAUCAUGUGUATT-3${}^{\prime }$ (siA1-3) for mouse. Nonspecific siRNA (NS) was obtained from Sangon Biotech (Shanghai, China). Cells were transfected with 20 nM siRNA by using Lipofectamine 3000 (#L3000015, Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. After 48 h, cells were transfected again.

Total cellular protein samples were lysed with 2 $\times$ SDS sample buffer, then denatured at 95 °C for 3 min. Protein samples were loaded onto 4–12% SDS-PAGE gel (#M00929, GenScript, Nanjing, China) and transferred to nitrocellulose membranes. After blocking in 5% non-fat milk for 1 h, the membranes were incubated with the following antibodies: mouse monoclonal anti-hnRNP A1 (#sc-32301, Santa Cruz Biotechnology, Dallas, TX, USA); mouse monoclonal anti-$\beta$-actin (#66009-1-lg, Proteintech, Wuhan, China); and rabbit polyclonal anti-FLAG (#20543-1-AP, Proteintech, Wuhan, China).

Total RNA was extracted using the AxyPrep miniprep kit (#AP-MN-MS-RNA-250, Axygen, Union City, CA, USA). RNA was reversely transcribed to cDNA by using Maxima H Minus Transcriptase (#EP0752, Thermo Fisher Scientific, Carlsbad, CA, USA), and amplified with Green Taq Mix (#P131, Vazyme, Nanjing, China) according to the manufacturer’s protocols. Sequences of primers were as follows: 5${}^{\prime }$-GACGAGGCAAGAGTTTCACC-3${}^{\prime }$ and 5${}^{\prime }$-GGTGGTAGAGTGGATGGACG-3${}^{\prime }$ for endogenous RUNX2 exon 5 alternative splicing; 5${}^{\prime }$-CACCTTGACCATAACCGTCTTCAC-3${}^{\prime }$ and 5${}^{\prime }$-GCTCCTCGCCCTTGCTCACCA-3${}^{\prime }$ for exogenous RUNX2 exon 5 alternative splicing; 5${}^{\prime }$-AAGCAATTTTGGAGGTGGTG-3${}^{\prime }$ and 5${}^{\prime }$-ATAGCCACCTTGGTTTCGTG-3${}^{\prime }$ for hnRNP A1; 5${}^{\prime }$-GACAAGAAGCCCTTCACTGC-3${}^{\prime }$ and 5${}^{\prime }$-AGACTGCGCCTGGTAGTTGT-3${}^{\prime }$ for alkaline phosphatase (ALP); 5${}^{\prime }$-CTACCTGTATCAATGGCTGGGAG-3${}^{\prime }$ and 5${}^{\prime }$-TGGTCAGCCAACTCGTCACAG-3${}^{\prime }$ for osteocalcin (OCN); 5${}^{\prime }$-GAAGGTGAAGGTCGGAGTC-3${}^{\prime }$ and 5${}^{\prime }$-GAAGATGGTGATGGGATTTC-3${}^{\prime }$ for GAPDH; 5${}^{\prime }$-GACGAGGCAAGAGTTTCACC-3${}^{\prime }$ and 5${}^{\prime }$-CGTGGTGGAGTGGATGGATG-3${}^{\prime }$ for mouse Runx2 exon 5 alternative splicing; 5${}^{\prime }$-CCAACTCTTTTGTGCCAGAGA-3${}^{\prime }$ and 5${}^{\prime }$-TGGCTACATTGGTGTTGAGCT-3${}^{\prime }$ for mouse Alp; 5${}^{\prime }$- CACCATGAGGACCCTCTCTC-3${}^{\prime }$ and 5${}^{\prime }$-TGGACATGAAGGCTTTGTCA-3${}^{\prime }$ for mouse Ocn; 5${}^{\prime }$-AAGGTCATCCCAGAGCTGAA-3${}^{\prime }$ and 5${}^{\prime }$- CTGCTTCACCACCTTCTTGA-3${}^{\prime }$ for mouse Gapdh.

After being cultured in differential medium for 7 days, the cells were rinsed three times with PBS and fixed with 4% paraformaldehyde for 15 min. Then, cells were washed with PBS for three times and alkaline phosphatase staining was performed with a BCIP/NBT ALP Color Development Kit (#C3206, Beyotime, Shanghai, China) according to the manufacturer’s instructions. The ALP activity of the harvested proteins was measured with an ALP activity kit (#A059-2-2, Jiancheng, Nanjing, China), and the absorbance values at 520 nm were measured. ALP activity relative to that of the control group was calculated after normalization with the total protein level.

After hDPSCs were cultured in differentiation medium for 14 days or mDPCs were cultured in differentiation medium for 7 days, cells were rinsed three times with PBS and fixed with 4% paraformaldehyde for 15 min. Then, cells were washed with PBS for three times. Calcium deposition was stained with ARS (#ALIR10001, Cyagen, Suzhou, China). Furthermore, mineralized nodules were dissolved in 10% cetylpyridinium chloride (#C129534, aladdin, Shanghai, China). The concentration of calcium was calculated by measuring the absorbance value at 562 nm.

All statistical analyses were performed using GraphPad Prism 8.0 (San Diego, CA, USA). Student’s t tests were used to analyze the statistical differences between two groups. One-way analysis of variance was performed for the comparison among three groups.

We explored the expression profile of hnRNP A1 during the mineralization induction of hDPSCs. Primary hDPSCs were isolated from extracted third molars, and the surface markers were analyzed. The results showed that these cells were positive for the MSC markers, including CD29, CD90, and CD146, although they were negative for the hematopoietic stem cells and leukocyte markers like CD34 and CD45 (Fig. 1A). In the current study, hnRNP A1 expression showed reduced at RNA (Fig. 1B) and protein (Fig. 1C) levels after hDPSC mineralization induction for 7 days. The expression levels of ALP and OCN were applied to represent the mineralized state of hDPSCs (Fig. 1B).

Fig. 1.

HnRNP A1 expression decreased during the mineralization induction in hDPSCs. (A) The expression of cell surface markers on hDPSCs was analyzed by flow cytometric. (B) RNA expression levels of ALP and OCN increased while the levels of hnRNP A1 decreased after seven days of mineralization induction. Histogram showed quantification for the relative RNA expression levels of hnRNP A1, ALP, and OCN, n = 3. GAPDH served as a loading control. (C) The corresponding protein expression levels of hnRNP A1 decreased after seven days of mineralization in hDPSCs. Histogram showed quantification for the relative protein expression levels of hnRNP A1, n = 3. $\beta$-actin served as a loading control. *: p 0.05; **: p 0.01. ALP, alkaline phosphatase; OCN, osteocalcin.

Our previous study demonstrated that the inclusion of RUNX2 exon 5 is enhanced during the odontogenic differentiation of hDPSCs [10]. To explore whether hnRNP A1 participates in RUNX2 exon 5 alternative splicing, we applied an online RNA binding sequence database, SpliceAid. We found that RUNX2 exon 5 has four potential binding sites for hnRNP A1 (Fig. 2A). We further explored the role of hnRNP A1 in the alternative splicing of RUNX2 exon 5 by gain- and loss-of-function experiments. We found that overexpression of hnRNP A1 significantly suppressed RUNX2 exon 5 inclusion in hDPSCs (Fig. 2B,C). In contrast, in the hnRNP A1 knockdown, there was a markedly increased RUNX2 exon 5 inclusion (Fig. 2D,E). To confirm the inhibition of hnRNP A1 on the inclusion of RUNX2 exon 5, we co-transfected HEK 293 cells with RUNX2 exon 5 minigene and hnRNP A1 expression plasmid. Consistently, hnRNP A1 overexpression notably suppressed RUNX2 exon 5 inclusion compared with the control group in the minigene system (Fig. 2F). These results suggested that hnRNP A1 suppressed the inclusion of RUNX2 exon 5.

Fig. 2.

HnRNP A1 suppressed the inclusion of RUNX2 exon 5. (A) HnRNP A1 binding sites in human RUNX2 exon 5 RNA sequence were predicted by SpliceAid online program. Negative score represents suppressing exon inclusion. (B) Overexpression of hnRNP A1 suppressed RUNX2 exon 5 inclusion in hDPSCs. The ratio of inclusion versus exclusion of RUNX2 exon 5 in hDPSCs was represented by histogram, n = 3. GAPDH served as a loading control. (C) Overexpression of exogenous T7-tagged hnRNP A1 in hDPSCs was confirmed by Western blot. $\beta$-actin served as a loading control. (D) Knockdown of hnRNP A1 promoted RUNX2 exon 5 inclusion in hDPSCs. The ratio of inclusion versus exclusion of RUNX2 exon 5 in hDPSCs was represented by histogram, n = 4. GAPDH served as a loading control. (E) Knockdown efficiency of hnRNP A1 was analyzed by Western blot. (F) Overexpression of hnRNP A1 suppressed the inclusion of RUNX2 exon 5 in a minigene in HEK 293 cells. RT-PCR was performed to analyze the alternative splicing of exon 5 of RUNX2 minigene with or without hnRNP A1 overexpression. Schematic diagram of RUNX2 exon 5 minigene was presented. The histogram showed the statistical analysis of the ratios of the inclusion versus exclusion of RUNX2 exon 5 in minigene, n = 3. GAPDH served as a loading control. (G) Overexpression of exogenous T7-tagged hnRNP A1 in HEK 293 cells was confirmed by Western blot. $\beta$-actin served as a loading control. *: p 0.05; ***: p 0.001; ****: p 0.0001.

Next, the function of hnRNP A1 in the odontogenic differentiation of hDPSCs was investigated. The expression levels of ALP and OCN were reduced after overexpression of hnRNP A1 in the mineralization induced hDPSCs (Fig. 3A). Meanwhile, ALP activity (Fig. 3B) and the formation of mineralized nodules were reduced in hDPSCs with overexpressed hnRNP A1 (Fig. 3C). Conversely, knockdown of hnRNP A1 led to a great increase in the expression of ALP and OCN in mineralized hDPSCs (Fig. 3D). ALP activity (Fig. 3E) and the formation of mineralized nodules (Fig. 3F) were also promoted in hDPSCs with hnRNP A1 knockdown. These results demonstrated that hnRNP A1 inhibited the mineralization of hDPSCs.

Fig. 3.

HnRNP A1 inhibited the differentiation of hDPSCs. (A–C) During the process of mineralization induction, overexpression of hnRNP A1 decreased the expression of ALP and OCN (A), ALP activity (B), and the formation of mineralized nodules (C). Alizarin red staining (ARS) was applied to stain mineralized nodules after hDPSCs were cultured in differentiation medium for 14 days. Mineralized nodules were collected with 10% cetylpyridinium chloride and the absorbance values at 562 nm were measured for quantification of mineralized nodules. Histogram respectively showed quantification for the expression of ALP and OCN, ALP activity or the formation of mineralized nodules, n = 3. GAPDH served as a loading control. Scale bar = 500 µm. (D–E) HnRNP A1 knockdown increased the expression of ALP and OCN (D), ALP activity (E) and the formation of mineralized nodules (F). ARS was applied to stain mineralized nodules after hDPSCs cultured in differentiation medium for 14 days. Histogram respectively showed quantification for the expression of ALP and OCN (n = 3), ALP activity (n = 3) or the formation of mineralized nodules (n = 4). GAPDH served as a loading control. Scale bar = 500 µm. *: p 0.05; **: p 0.01; ***: p 0.001; ****: p 0.0001.

According to our previous study, RUNX2 exon 5 is highly conserved in human and mouse except for one nucleotide replacement [10]. Therefore, we speculated that hnRNP A1 may also suppress mouse Runx2 exon 5 inclusion and odontogenic differentiation in mDPCs. As we speculated, hnRNP A1 plays an identical role both in hDPSCs and mDPCs in Runx2 exon 5 inclusion and odontogenic differentiation. Overexpression of hnRNP A1 significantly suppressed the inclusion of Runx2 exon 5 (Fig. 4A), and hnRNP A1 knockdown markedly promoted the inclusion of Runx2 exon 5 (Fig. 4E). We also found that overexpression of hnRNP A1 inhibited the expression of Alp and Ocn in the differentiation of mDPCs (Fig. 4C). In addition, the formation of mineralized nodules was reduced in hnRNP A1 overexpressed mDPCs compared with the control group (Fig. 4D). On the contrary, knockdown of hnRNP A1 led to increased expression of Alp and Ocn and increased mineralized nodules (Fig. 4G,H). These findings demonstrated that hnRNP A1 contributed to the skipping of RUNX2 exon 5 and repressed odontogenic differentiation of mDPCs.

Fig. 4.

HnRNP A1 suppresses the inclusion of Runx2 exon 5 and differentiation in mDPCs. (A) Overexpression of hnRNP A1 suppressed the inclusion of Runx2 exon 5. Bar graph showed the summarized ratio of inclusion versus exclusion of Runx2 exon 5, n = 3. Gapdh served as a loading control. (B) Overexpression of exogenous FLAG-tagged hnRNP A1 was confirmed by Western blot. Actin served as a loading control. (C,D) Overexpression of hnRNP A1 reduced the expression of Alp and Ocn (C) and the formation of mineralized nodules (D). ARS was applied to stain mineralized nodules after mDPCs cultured in differentiation medium for 7 days, Scale bar = 500 µm. Histogram showed quantification for the expression of Alp and Ocn, or the formation of mineralized nodules, n = 3. Gapdh served as a loading control. (E) Knockdown of hnRNP A1 promoted the inclusion of Runx2 exon 5. Bar graph showed the summarized ratio of inclusion versus exclusion of Runx2 exon 5, n = 3. Gapdh served as a loading control. (F) Knockdown efficiency of hnRNP A1 was analyzed by Western blot. $\beta$-actin served as a loading control. (G,H) Knockdown of hnRNP A1 increased the expression of Alp and Ocn (G) and the formation of mineralized nodules (H). ARS was applied to stain mineralized nodules after mDPCs cultured in differentiation medium for 7 days, Scale bar = 500 µm. Histogram showed quantification for the expression of Alp and Ocn, or the formation of mineralized nodules, n = 3. Gapdh served as a loading control. *: p 0.05; **: p 0.01.