Circular RNA circANAPC2 mediates the impairment of endochondral ossification by miR‐874‐3p/SMAD3 signalling pathway in idiopathic short stature

Abstract Idiopathic short stature (ISS) is a main reason for low height among children. Its exact aetiology remains unclear. Recent findings have suggested that the aberrant expression of circRNAs in peripheral blood samples is associated with many diseases. However, to date, the role of aberrant circRNA expression in mediating ISS pathogenesis remains largely unknown. The up‐regulated circANAPC2 was identified by circRNA microarray analysis and RT‐qPCR. Overexpression of circANAPC2 inhibited the proliferation of human chondrocytes, and cell cycle was arrested in G1 phase. The expressions of collagen type X, RUNX2, OCN and OPN were significantly down‐regulated following circANAPC2 overexpression. Moreover, Von Kossa staining intensity and alkaline phosphatase activity were also decreased. Luciferase reporter assay results showed that circANAPC2 could be targeted by miR‐874‐3p. CircANAPC2 overexpression in human chondrocytes inhibits the expression of miR‐874‐3p. The co‐localization of circANAPC2 and miR‐874‐3p was confirmed in both human chondrocytes and murine femoral growth plates via in situ hybridization. The rescue experiment demonstrated that the high expression of miR‐874‐3p overexpression antagonized the suppression of endochondral ossification, hypertrophy and chondrocyte growth caused by circANAPC2 overexpression. A high‐throughput screening of mRNA expression and RT‐qPCR verified SMAD3 demonstrated the highest different expressions following overcircANAPC2. Luciferase reporter assay results indicated that miR‐874‐3p could be targeted by Smad3, thus down‐regulating the expression of Smad3. Subsequent rescue experiments of SMAD3 further confirmed that circANAPC2 suppresses endochondral ossification, hypertrophy and chondrocyte growth through miR‐874‐3p/Smad3 axis. The present study provides evidence that circANAPC2 can serve as a promising target for ISS treatment.


| INTRODUC TI ON
Idiopathic short stature (ISS) is diagnosed by a height of less than two standard deviation scores in children after the exclusion of identifiable diseases, including systemic diseases, small for gestational infant, psychological disorders, overt hormone deficiency, nutritional imbalance and chromosomal abnormalities. 1 The patients with ISS often present some problems in behaviour, self-perception, social adjustment, school competence and attention. 2,3 Although the pathogenesis of ISS has been studied over past decades, its underlying cause may be complicated and remains largely unclarified. 1 At present, recombinant human growth hormone (rhGH) has been used to treat ISS patients owing to the unknown aetiology. Silvers et al 4 reported there are beyond 500 000 ISS children who need to be treated by rhGH and the expense on ISS treatment would exceed $10 billion per year in the United States. Unlike growth hormone deficiency-related short stature, ISS patients do not demonstrate the absence of growth hormones. Therefore, the treatment effect of recombinant human growth hormones on ISS varies considerably between studies. [5][6][7][8] Moreover, high-dose rhGH treatment seems to be related to some side effects, such as bone and cartilage cancer, cerebrovascular accidents, hyperglycaemia and hyperinsulinaemia. [8][9][10] Ying et al 8 found that the growth velocity of ISS patients (9.55 ± 0.13 cm) in the first year clearly increased after rhGH treatment compared to that before treatment. Kim  Circular RNAs (circRNAs) are a group of non-coding RNAs that are highly conserved and commonly found in mammalian cells. [12][13][14] Their stability is high compared to other non-coding lncRNAs and miRNAs owing to their closed circular structures. CircRNAs could suppress the functions of their downstream gene by acting as miRNA sponges. 15  Down-regulation B many diseases such as cardiovascular diseases, diabetes mellitus, osteoarthritis, cancers, pre-eclampsia and neurological disorders. [16][17][18] CircRNAs have demonstrated potential applications as disease biomarkers and novel therapeutic targets. 19 Li et al 20 observed that circ_0136474 inhibited cell proliferation and accelerated cell apoptosis via acting as a sponge with miR-127-5p in osteoarthritis.

| Study patient characteristics
Sixty-eight pairs of ISS patients and age-/gender-matched healthy individuals were collected at the Second Affiliated Hospital of Nanchang University (China). The ISS patients comprised 32 males and 36 females ranging from 4 to 12 years with an average age of 8.92 ± 0.34 years and with height ranging from 88.71 to 142.2 cm (mean height of 120.2 ± 1.72 cm). The matched control individuals comprised 35 females and 33 males, with a mean age of 8.39 ± 0.28 years (ranging from 4 to 12 years) and a mean height of 132.1 ± 1.57 cm (ranging from 103.9 to 150.9 cm, Table 1). Patients exhibiting the following conditions were excluded: (a) abnormal hormone levels (a growth hormone peak of less than 6 ng/mL) or disturbance of thyroid function and puberty; (b) small for gestational age (size and weight at birth were below the 10th percentile); (c) chronic exposure to environmental contaminants and health conditions that can affect human growth; (d) cytogenetically detected chromosomal aberrations; (e) skeletal anomalies or dysmorphic features; and (f) whole-exome sequencing. Sixty-eight pairs of blood specimens were withdrawn from the study patients from October 2016 to March 2019, snap-frozen in liquid nitrogen and then kept at −80°C for long-term preservation. Four specimen pairs were used for circRNA microarray analysis, while all 136 specimens were used to validate circRNAs expression by quantitative real-time polymerase chain reaction (qRT-PCR). Ethical approval was obtained from the Ethics Committee of the Second Affiliated Hospital of Nanchang University. The parent and/or guardian signed a written consent form prior to the child's participation.

| Differential expression analysis of circRNAs
Microarray data processing was carried out using the Agilent's Feature Extraction version 11.0.1.1 software. The expression levels of circRNAs were measured by quantile normalization method using the Quantile algorithm and limma package version 3.11 (https:// www.bioco nduct or.org/packa ges/relea se/bioc/html/limma.html) in R software version 3.6.0. The expression profiles of circRNAs were then classified into control and ISS groups. The differently expressed circRNAs between the control and ISS groups were determined through volcano plot filtering. Statistical significance level was set at a fold-change value of greater than 2.0 and a false discovery rate (FDR)-corrected P-value of less than 0.05. Hierarchical cluster analysis of the differential expression patterns of circRNAs was carried out using the R program.
F I G U R E 2 CircANAPC2 suppressed chondrocyte proliferation. A, The up-regulated circANAPC2 in ISS patients was verified by RT-qPCR. B,C, No significant difference was shown in the circAXIN1 and circUBE2J2 between the ISS patients and normal control individuals. D, No significant difference in the circANAPC2 was observed between the RNase R and normal control groups. E, The linear ANAPC2 was significantly reduced after the RNase R treatment. Meanwhile, the expression of linear ANAPC2 did not present a significant difference between the ISS sample and the normal control. This indicated that the circANAPC2, but not linear ANAPC2, was up-regulated in ISS. F, CircANAPC2 was successfully overexpressed via overcircANAPC2 vectors. G, CCK8 revealed that human chondrocyte proliferation was inhibited following overcircANAPC2. H, Flow cytometry assays indicated that the cell cycle was arrested in the G1 phase. The data are presented as the mean ± SD. n = 3. t Test, ** P <.01, *** P <.001 vs control. ISS, idiopathic short stature; ANAPC2, anaphase promoting complex subunit 2; ISS, idiopathic short stature; and CCK-8, Cell Counting Kit-8 | 3413 LIU et aL.
A p-value of < 0.05 and the enrich factor were adopted to identify and rank the corresponding signalling pathways.
Gene ontology (GO) enrichment analysis (http://www.geneo ngolo ty.org/) was used to assess the biological processes (BP), molecular functions (MF) and cellular components (CC) of the identified circRNA targets according to their target mRNAs. Meanwhile, KEGG analysis was carried out using DAVID in order to identify the molecular pathways. The enrichment scores of -log10 (P-value) was used as cut-off value in both GO and KEGG analysis.

| Validation of differentially expressed circRNAs by qRT-PCR
Total RNAs were isolated from the frozen tissues using Eastep® Super RNA Extraction Kit (Promega (Beijing) Biotech Co., Ltd.) and then reversed-transcribed using PrimeScript™ RT reagent Kit with gDNA Eraser (TaKaRa). QRT-PCR was conducted on an Applied Biosystems QuantStudio 6 RT-PCR system (Thermo Fisher Scientific, Inc) using the TB Green® Premix Ex Taq™ II (TaKaRa).
The primers used are presented in Table 2. The qRT-PCR conditions for circANAPC2 amplification were as follows: 10 minutes at 95°C, followed by 40 cycles of 10 seconds at 95°C and 34 seconds at 60°C. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was employed as the internal control. Each assay was performed in triplicate. The 2 -ΔΔCt method was used to determine the relative mRNA expression levels of target genes.

| Cell culture
We purchased the human chondrocytes from the Procell life science technology Co., Ltd. and then cultured in DMEM (Gibco, Thermo Fisher Scientific, Inc) containing 10% FBS (Gibco) before being maintained at 37°C in a humidified incubator with 5% CO 2 .
Cell Counting Kit 8 (CCK-8; TransGen Biotech Co., Ltd) was used to detect chondrocytes proliferation rate in compliance with the manufacturer's protocol. The detailed steps of CCK8 assay were referring to our previous report. 22

| mRNA high-throughput sequencing
We employed the TruSeq Stranded mRNA LT Sample Prep Kit sequences were filtered by Trimmomatic after removing the lowquality reads with adapter and ploy-N. We used the HISAT2 to compare the high-quality reads and human genome. Cufflink quantitatively evaluated the FPKM value of each gene. HTSeq-count calculated the read counts of each gene. Differential expression analysis was conducted using the DESeq (2012) R package. A P-value less than .05 and a fold change of greater than 2 were considered as the cut-off value for GO and KEGG analysis.

| Luciferase reporter assay
The circANAPC2 plasmid or its mutation fragments ( Figure S3

| Western blotting
Total protein was isolated using RIPA lysis buffer (Applygen Technologies Inc). BCA assay (Thermo Fisher Scientific, Inc) was used to determine the protein levels. The protein lysate was then

Kossa staining
The detailed steps of Von Kossa staining and alkaline phosphatase assay were referring to our previous report. 22 First, we washed the chondrocytes three times using PBS. Then, the chondrocytes were fixed using 4% paraformaldehyde for 15 min at room temperature.
Afterwards, the fixed chondrocyte cells were exposed to a BCIP/

| Statistical analysis
Statistical differences between groups were compared with the unpaired t test or one-way analysis of variance (ANOVA) followed by  Thus, the circANAPC2 was considered as a candidate gene to reveal ISS pathogenesis. Our outcome of RT-qPCR showed excellent reliability with the circRNA microarray ( Figure 2A). No obvious difference was found between the normal control and RNase R groups ( Figure 2D). In contrast, the linear ANAPC2 was significantly reduced after the RNase R treatment ( Figure 2E). Moreover, the expression levels of linear ANAPC2 were relatively similar between ISS and control groups. This indicated that the circAN-APC2, but not linear ANAPC2, was up-regulated in ISS. The outcome of a Western blot analysis and RT-qPCR confirmed that the expression levels of RUNX2 and collagen type X were significantly down-regulated following circANAPC2 overexpression. This suggests that chondrocyte hypertrophy was suppressed ( Figure 3A).

| CircANAPC2 suppressed chondrocyte proliferation, hypertrophy and endochondral ossification
As shown in Figure 3B, the expressions of the osteogenic genes (OCN and OPN) were down-regulated by RT-qPCR and Western blot analysis after circANAPC2 overexpression. Meanwhile, ALP activity was also decreased ( Figure 3C). Additionally, Von Kossa staining demonstrated reduced mineralization ( Figure 3D). These F I G U R E 5 CircANAPC2 suppressed chondrocyte proliferation, hypertrophy and endochondral ossification by regulating miR-874-3p. A, MiR-874-3p mimics up-regulated the expression of miR-874-3p. B,C, Although the chondrocyte proliferation was suppressed in the overcircANAPC2 + NC mimic group, it was reversed in the overcircANAPC2 + miR-874-3p mimic group. D, Von Kossa staining demonstrated reduced mineralization in overcircANAPC2 + NC mimic group. However, Von Kossa staining was recovered in the overcircANAPC2 + miR-874-3p mimic group. E, Like the Von Kossa staining, ALP activity was decreased in overcircANAPC2 + NC mimic group, but it also was recovered in the overcircANAPC2 + miR-874-3p mimic group. F, Although the mRNA and protein expressions of collagen type X and RUNX2 were down-regulated in overcircANAPC2 + NC mimic group, no significant difference was observed in the overcircANAPC2 + miR-874-3p mimic group compared to the NC group. G, The mRNA and protein expressions of OPN and OCN were down-regulated in overcircANAPC2 NC mimic group. However, no significant difference was observed in the overcircANAPC2 + miR-874-3p mimic group compared to the NC group. The data are presented as the mean ± SD. n = 3. Two groups were compared using t test or three groups were compared using ANOVA followed by Tukey's test. ** P <.01, *** P <.001 vs control. ANAPC2, anaphase promoting complex subunit 2; miR, microRNA; NC, negative control; ALP, alkaline phosphatase; OPN, osteopontin; OCN, osteocalcin; and RUNX2, runt-related transcription factor 2 | 3419 LIU et aL.
results indicated that overexpression of circANAPC2 can markedly suppress endochondral ossification, hypertrophy and chondrocyte proliferation ( Figure 3C and 3D).
Subsequently, we observed that luciferase reporter activities were For example, although the chondrocyte proliferation was suppressed in overcircANAPC2 + NC mimic group, it was reversed in overcircANAPC2 + miR-874-3p mimic group ( Figure 5B and C).
Von Kossa staining revealed that mineralization was reduced in overcircANAPC2 + NC mimic group. However, Von Kossa staining was recovered in the overcircANAPC2 + miR-874-3p mimic group ( Figure 5D). Similar to Von Kossa staining, ALP activity was decreased in overcircANAPC2 + NC mimic group, but it also was recovered in overcircANAPC2 + miR-874-3p mimic group ( Figure 5E).
Although the mRNA and protein expression levels of RUNX2 and collagen type X were down-regulated in overcircANAPC2 + NC mimic group, no significant difference was observed in overcircAN-APC2 + miR-874-3p mimic group compared to NC group ( Figure 5F).
The mRNA and protein expression levels of OPN and OCN were down-regulated in overcircANAPC2 NC mimic group. However, no significant difference was observed in overcircANAPC2 + miR-874-3p mimic group compared to NC group ( Figure 5G). Taken together, these results indicated that circANAPC2 suppressed endochondral ossification, hypertrophy and chondrocyte proliferation via regulation of miR-874-3p.

| CircANAPC2 significantly increases SMAD3 expression, but not SMAD2 expression, by inhibiting miR-874-3p
To identify the target genes of miR-874-3p, the differentially ex- were associated with chondrocyte development and chondral ossification ( Figure 6E). Interestingly, SMAD3 is involved in three of these pathways (cell cycle, TGFβ and Wnt). Moreover, SMAD3 and miR-874-3p were estimated to have four binding regions with higher scores ( Figure 6F). Thus, SMAD3 was selected as the key gene for ISS pathogenesis.
Luciferase reporter assay showed the reduction in HEK293 cells cotransfected with SMAD3-WT/miR-874-3p mimics, and it was restored after the SMAD3-Mut and miR-874-3p mimics were cotransfected ( Figure 6G and H). As expected, SMAD3 exhibited upregulation and the expression of IHH, a SMAD3 target gene, was down-regulated in the overcircANAPC2 group ( Figure 7A). There were no remarkable differences in the expression levels of SMAD2 and IHH between the overcircANAPC2 + miR-874-3p group and normal control group ( Figure 7A). The rescue experiment of miR-874-3p indicated that the up-regulated SMAD3 and the down-regulated IHH following overcircANAPC2 were reversed in the overcircAN-APC2 + miR-874-3p mimic group ( Figure 7B). This further confirmed the circANAPC2 significantly increases SMAD3 expression, but not SMAD2, by suppressing miR-874-3p.

| CircANAPC2 suppressed endochondral ossification, hypertrophy and chondrocyte proliferation by regulating miR-874-3p/SMAD3
RT-qPCR and Western blot verified the effectiveness of SMAD3 siRNA in the human chondrocyte ( Figure 7C). SMAD3 presented up-regulation and IHH showed down-regulation in the overcircAN-APC2 compared to normal control group. Meanwhile, Von Kossa staining also demonstrated mineralization was attenuated in the overcircANAPC2 + siNC group ( Figure 7E). However, it did not demonstrate significant difference in SMAD3 and IHH between the overcircANAPC2 + siSMAD3 and normal control group ( Figure 7D).

| Down-regulation of circANAPC2 enhanced chondrocyte hypertrophy and endochondral ossification by regulating miR-874-3p/SMAD3
The expression of miR-874-3p was up-regulated after circANAPC2 down-regulation by siRNA ( Figure 8A and B). Meanwhile, SMAD3 was down-regulated and IHH was up-regulated following circAN-APC2 down-regulation ( Figure 8J and K). Though no obvious difference was found in the proliferation of human chondrocytes as determined by CCK-8 and flow cytometric assays between the cir-cANAPC2 down-regulation group and control group ( Figure 8C-E), the expressions of collagen type X, RUNX2, OCN and OPN were significantly up-regulated following circANAPC2 down-regulation ( Figure 8H and I). Moreover, ALP activity was also increased ( Figure 8G) and Von Kossa staining demonstrated increased mineralization ( Figure 8F). These results indicated that circANAPC2 down-regulation can markedly enhance chondrocyte hypertrophy and endochondral ossification.

| D ISCUSS I ON
CircRNAs have been increasingly recognized as diagnostic and therapeutic biomarkers for many diseases. This is primarily owing to their high stability than other non-coding RNA, such as miRNAs and lncR- Meanwhile, the expression of IHH, a SMAD3 target gene, was down-regulated compared to the control. No significant difference was observed in the SMAD2 between the overcircANAPC2 group and normal control group. B, The up-regulated SMAD3 and the down-regulated IHH were reversed in the overcircANAPC2 + miR-874-3p mimic group, and no significant difference was observed in the SMAD3 and IHH between the overcircANAPC2 + miR-874-3p mimic group and normal control group. This rescue experiment further confirmed the miR-874-3p negatively regulates SMAD3. C, The effectiveness of SMAD3 siRNA was detected in the human chondrocyte by RT-qPCR and Western blot. D, Although SMAD3 presented up-regulation and IHH showed down-regulation in the overcircANAPC2 + siNC, it did not demonstrate significant difference between the overcircANAPC2 + siSMAD3 and normal control group. E, Von Kossa staining demonstrated reduced mineralization in the overcircANAPC2 + siNC group. However, Von Kossa staining was recovered in the overcircANAPC2 + siSMAD3 group. This indicated that circANAPC2 suppressed chondrocyte proliferation, hypertrophy and endochondral ossification by regulating miR-874-3p/ SMAD3 axis. The data are presented as the mean ± SD. n = 3. Two groups were compared using t test or three groups were compared using ANOVA followed by Tukey's test. ** P <.01, *** P <.001 vs control. ANAPC2, anaphase promoting complex subunit 2; mRNA, messenger RNA; SMAD2, SMAD family member 2; SMAD3, SMAD family member 3; TGFβ, transforming growth factorβ; RT-qPCR, reverse transcription-quantitative PCR; siNC, negative control siRNA; siSMAD3, SMAD family member 3 siRNA; miR, microRNA; and IHH, Indian hedgehog homolog There still is limitation in our study. The authors observed that the expression of circANAPC2 was up-regulated in ISS patients and found that circANAPC2 can inhibit chondrocyte proliferation, hypertrophy and endochondral ossification via a novel axis circAN-APC2/ miR-874-3p/SMAD3 in vitro. However, these results need to be verified in vivo via knockout or overexpression animal models.
In conclusion, our study reveals that circANAPC2 can bind to miR-874-3p and regulate SMAD3 signalling pathway, thereby contributing to the pathogenesis of ISS.
F I G U R E 8 Down-regulation of circANAPC2 enhanced chondrocyte hypertrophy and endochondral ossification by regulating miR-874-3p/SMAD3. A and B, The expression of miR-874-3p was up-regulated after circANAPC2 down-regulation by siRNA. C-E, No significant difference was observed in the proliferation of human chondrocytes as measured by CCK-8 and flow cytometry assays between the group of circANAPC2 down-regulation and control group. F and G, ALP activity was increased and Von Kossa staining demonstrated increased mineralization. H and I, The expressions of collagen type X, RUNX2, OCN and OPN were significantly up-regulated following circANAPC2 down-regulation. J and K, SMAD3 was down-regulated and IHH was up-regulated following circANAPC2 down-regulation. The data are presented as the mean ± SD. n = 3. Two groups were compared using t test or three groups were compared using ANOVA followed by Tukey's test. ** P <.01, *** P <.001 vs control. ANAPC2, anaphase promoting complex subunit 2; siCircANAPC2, circANAPC2 siRNA; siNC, negative control siRNA; miR, microRNA; SMAD3, SMAD family member 3; IHH, Indian hedgehog homolog; ALP, alkaline phosphatase; OPN, osteopontin; OCN, osteocalcin; and RUNX2, runt-related transcription factor 2