Expression of RMRP RNA is regulated in chondrocyte hypertrophy and determines chondrogenic differentiation

Mutations in the RMRP-gene, encoding the lncRNA component of the RNase MRP complex, are the origin of cartilage-hair hypoplasia. Cartilage-hair hypoplasia is associated with severe dwarfism caused by impaired skeletal development. However, it is not clear why mutations in RMRP RNA lead to skeletal dysplasia. Since chondrogenic differentiation of the growth plate is required for development of long bones, we hypothesized that RMRP RNA plays a pivotal role in chondrogenic differentiation. Expression of Rmrp RNA and RNase MRP protein subunits was detected in the murine growth plate and during the course of chondrogenic differentiation of ATDC5 cultures, where Rmrp RNA expression was found to be correlated with chondrocyte hypertrophy. Genetic interference with Rmrp RNA expression in ATDC5 cultures caused a deregulation of chondrogenic differentiation, with a prominent impact on hypertrophy and changes in pre-rRNA processing and rRNA levels. Promoter reporter studies showed that Rmrp RNA expression responds to chondrogenic morphogens. Chondrogenic trans-differentiation of cartilage-hair hypoplasia fibroblasts was impaired with a pronounced impact on hypertrophic differentiation. Together, our data show that RMRP RNA expression is regulated during different stages of chondrogenic differentiation and indicate that RMRP RNA may play a pivotal role in chondrocyte hypertrophy, with potential consequences for CHH pathobiology.


Materials and Methods
In situ hybridization (ISH). Decalcified knee epihyseal growth plates of 6-week-old C57BL/6 mice (use of mouse growth plates was approved by the Maastricht University Animal Ethics Committee, according to Dutch law; and methods utilized to obtain growth plates were carried out in accordance with Maastricht University Animal Ethics Committee and Dutch law) were embedded in paraffin and 5 μm sections were cut. Sections were deparaffinized in a xylene/ethanol series ending in PBS (136 mM NaCl (Merck Millipore, Darmstadt, Germany), 2.7 mM KCl (Merck Millipore), 9.0 mM Na 2 PO 4 .H 2 O (Merck Millipore), 1.8 mM KH 2 PO 4 (Merck Millipore)). Antigen retrieval was performed using 20 μg/ml proteinase K (Exiqon, Vedbaek, Denmark). Slides were washed extensively with demineralized water to remove proteinase K. Slides were pre-hybridized with microRNA ISH buffer (Exiqon; miRCURY LNA microRNA ISH optimization kit FFPE). Hybridization was performed for 1 hour at 50 °C on a heating plate using 80 nM of Rmrp RNA double-digoxigenin (DIG)-labeled miRCURY LNA probe (5′DIG-CTGACGGATGACGC-3′DIG; custom ordered at Exiqon) or a double-DIG labeled scrambled LNA miRNA probe (5′-GTGTAACACGTCTATACGCCCA-3′; Exiqon) in miRNA ISH buffer. Slides were washed three times with 5x SSC (0.75 M NaCl, 0.075 M sodium citrate) for 10 minutes at 50 °C. Subsequently slides were blocked with blocking solution (DIG Wash and Block buffer set; Roche, Basel, Switzerland) for 30 minutes and incubated with anti-DIG-Alkaline Phosphatase (AP) Fab fragments (1:500; Roche) in blocking solution for 1 hour. Endogenous AP was blocked with Levamisole solution (Vector Laboratories, Burlingame, CA, USA) and bound anti-DIG-AP Fab fragments were detected with AP substrate (NBT/BCIP; Roche) in demineralized water. Slides were washed in PBS-T (0.1% Tween 20; Sigma-Aldrich, St. Louis, MO, USA), counter-stained with 0.01% FastGreen (Sigma-Aldrich), dehydrated and mounted with Entellan (Merck Millipore).
RT-qPCR. RNA was isolated from TRIzol samples by collecting the aqueous phase after phase separation.
RNA was precipitated with isopropanol (30 minutes, −80 °C) and centrifuged for 30 minutes at 20,000 × g, 4 °C. RNA pellets were washed with 80% ethanol and potential DNA contamination was removed by DNase I (Roche) treatment (1 hour, 37 °C). After subsequent ethanol precipitation, RNA was dissolved in 15 μL DNase/RNase free water (Eurogentec). RNA quantity and purity were determined spectrophotometrically (Biodrop, Isogen Life Sciences, Utrecht, the Netherlands). DNA-free total RNA was reverse transcribed using standard procedures and random hexamer priming as described previously 33 . Real time quantitative PCR (RT-qPCR) was performed in 96-well optical plates. For each cDNA sample a mix was prepared consisting of Mesagreen qPCR Mastermix Plus for SYBR Green (Eurogentec) and 300 nM forward and reverse oligonucleotides. Serially diluted standard curves were utilized to quantify gene expression in the samples. A Biorad CFX96 Real-Time PCR Detection System was used for amplification using the following protocol: denaturation at 95 °C for 10 minutes, followed by 50 cycles of amplification (15 seconds 95 °C and 1 minute 60 °C) followed by a dissociation curve. Data were analyzed using Biorad CFX Manager Software version 3.1, based on the relative quantification of mRNA expression of the target gene normalized to a housekeeping gene (ATDC5/MCT: β-actin, human fibroblasts: GAPDH). Primer sequences are depicted in Table 2.

Results
Rmrp RNA and RNase MRP protein subunits are differentially expressed in the developing growth plate. To investigate RMRP RNA expression during chondrogenic differentiation of the growth plate, 5 μm tissue sections were prepared from growth plates of 6 weeks-old mice. Expression of Rmrp RNA was determined by in situ hybridization (Fig. 1A). Resting zone chondrocytes express Rmrp RNA and weak expression was observed in the chondrocytes of the proliferative zone. The highest expression levels of Rmrp RNA were detected in the hypertrophic zone. Cells in the remodeling zone of the growth plate (osteoclasts and osteoblasts) were also positive for the expression of Rmrp RNA. Subsequently, we evaluated the spatiotemporal expression of a number of RNase MRP protein subunits RPP25, RPP30, RPP38, RPP40 and POP1 (Fig. 1B). Expression of COL10A1 was immunohistochemically detected to visualize the location of the hypertrophic zone (Fig. 1B). Without exception, all RNase MRP protein subunits tested displayed a spatiotemporal growth plate distribution pattern highly similar to Rmrp RNA; resting zone chondrocytes express RNase MRP subunits, whereas weak expression was observed in chondrocytes of the proliferative zone. Highest RNase MRP protein subunit expression levels were detected in the hypertrophic zone of the developing growth plate.

Rmrp RNA expression is induced during ATDC5 chondrogenic differentiation. To study RMRP
RNA expression in a model for chondrogenic differentiation we used the ATDC5 cell line 31 . ATDC5 cells follow a defined chondrogenic differentiation program after stimulation with insulin, transferrin and sodium selenite. The induction of Col2a1 expression at day 7 during differentiation marks chondrogenic differentiation. At 14-21 days during differentiation cells have acquired a predominant hypertrophic/mineralizing phenotype as indicated by co-expression of Col10a1, and at day 21 mineralization becomes evident by expression of osteocalcin ( Fig. 2A). Rmrp expression is positively correlated with the hypertrophic phenotype of the chondrocyte. Above data indicate that RMRP RNA expression is associated with late phase/hypertrophic chondrogenic differentiation. Therefore we next asked whether RMRP RNA expression adapts to a changing chondrocyte hypertrophic phenotype. To drive differentiating ATDC5 cells towards a hypertrophic phenotype we reduced the expression of Sox9 or Bapx1 by targeting the corresponding mRNAs by siRNA-mediated knockdown. In line with previous reports and work from our group 33, 35-37 , knockdown of Sox9 expression (Fig. 3A) resulted in reduced expression of chondrogenic markers Bapx1 (Fig. 3B), Col2a1 (Fig. 3C) and Aggrecan (Fig. 3D) and concomitantly increased expression of hypertrophic markers Runx2 (Fig. 3E), Mef2c (Fig. 3F) and Col10a1 (Fig. 3G), which is consistent with a prominent hypertrophic chondrocyte phenotype. As a result of Sox9 knockdown, the expression of Rmrp RNA (Fig. 3H) was increased. We previously reported 35 that knockdown of the key hypertrophic repressor Bapx1 (Fig. 3B) does not influence the expression of chondrogenic factors Sox9, Col2a1 and Aggrecan (Fig. 3A,C,D). Instead, a selective hypertrophic chondrocyte phenotype was provoked, as evidenced by increased expression of hypertrophic markers Runx2, Mef2c and Col10a1 (Fig. 3E-G). Similar to what was observed for Sox9 knockdown, preferential hypertrophic differentiation by knockdown of Bapx1 expression led to increased expression of Rmrp RNA (Fig. 3H). Increased hypertrophy was also observed when ATDC5 cells were exposed to the hypertrophic inducer BMP-2 36 , as evidenced by induced expression of hypertrophic markers. Under these conditions, the expression of Rmrp RNA was also induced (Fig. 3H). To further investigate Rmrp induction by chondrocyte hypertrophy, ATDC5 cell were differentiated for 14 days to induce a hypertrophic phenotype. To counteract hypertrophy, cultures were exposed to PTHrP from day 10 in differentiation onwards 23,38 . Addition of PTHrP reduced the induction of the hypertrophic markers Runx2 (Fig. 4A) and Col10a1 (Fig. 4B). As a result of PTHrP-mediated reduction of hypertrophy, Rmrp RNA expression was downregulated as well (Fig. 4C). To confirm the PTHrP-mediated reduction of Rmrp RNA expression in another hypertrophic chondrocyte model, we used MCT cells. MCT cells are mouse chondrocytes, immortalized with a temperature-sensitive simian virus 40 large tumor antigen. These cells proliferate at 32 °C, but terminally differentiate and become hypertrophic within 24 hours at 37 °C 32 . As expected, culturing MCT cells at 37 °C induced hypertrophic differentiation as indicated by upregulation of Runx2 and Col10a1 expression (Fig. 4D,E). Rmrp RNA expression was also upregulated in this hypertrophic cell model (Fig. 4F). When the hypertrophic MCT phenotype was suppressed by PTHrP (Fig. 4D,E), Rmrp RNA expression responded similarly (Fig. 4F). Overall, our data indicate a positive correlation between Rmrp RNA expression levels and chondrocyte hypertrophy.
Chondrogenic morphogens alter Rmrp promoter activity. Our data described above suggested that RMRP RNA expression may be responsive to common chondrogenic morphogens, for example via modulation of proximal promoter activity. To acquire a broader understanding of chondrogenic pathways that may influence RMRP RNA expression, ATDC5 cells were transfected with a Gaussia luciferase reporter, driven by the 1500 nucleotide sequence upstream of the mouse Rmrp transcription start site (pGluc-Rmrp-prom plasmid) and cells were exposed to different chondrogenic mediators (Fig. 5). PTHrP keeps chondrocytes in a proliferative state and counteracts hypertrophic differentiation 23,38 . PTHrP decreased Rmrp promoter activity by 18%. bFGF (or FGF2), a negative regulator of chondrocyte hypertrophic maturation 39 , reduced Rmrp promoter activity by 35%. TGFβ isoforms activate SMAD signaling via SMAD2/3 40 . Of all TGFβ isoforms, TGFβ3 was described to be able to induce matrix mineralization 41 . Indeed, TGFβ3 increased Rmrp promoter activity by 40%. In agreement with the TGFβ3-induced increase of Rmrp promoter activity, exposure to (5Z)-7-Oxozeaenol, an inhibitor of TGF-β activated kinase-1 (TAK1) 42 , decreased Rmrp promoter activity by 22%. Exposure to BMP-2 (pro-hypertrophic, see Fig. 3) increased Rmrp promoter activity by 105%, whereas dorsomorphin, an inhibitor of BMP-mediated SMAD1/5/8 phosphorylation 43 , decreased Rmrp promoter activity by 76%. GDF-5 (a BMP family member described to induce both chondrogenic differentiation and hypertrophy 44 did not significantly change Rmrp promoter activity. WNT-3A and WNT-5A, two prominent ligands of the Wnt signaling pathway and crucial in chondrocyte differentiation 45 , increased Rmrp promoter activity by 45% and 26%, respectively. Finally, Sonic Hedge Hog (SHH), the main ligand that activates Patched-1-mediated GLI-signaling and which is crucial in limb bud patterning during embryogenesis 46 , did not significantly alter promoter activity. We thus concluded that the 1500 nucleotide sequence upstream of the Rmrp RNA transcription start site is responsive to mediators known to alter the chondrocyte phenotype. This indicates that, in chondrocytic cells, Rmrp RNA levels may be controlled by transcription factors involved in chondrocyte differentiation and hypertrophy.

Rmrp RNA knockdown deregulates chondrogenic differentiation of ATDC5 cells. To investi-
gate whether RMRP RNA regulates chondrogenic differentiation, Rmrp RNA expression was targeted by RNAi. ATDC5 cells were transfected with an siRNA duplex at day -1, 2 and 5 during chondrogenic differentiation and gene expression was determined at day 0, 7 and 10 in differentiation. The siRNA duplex reduced the expression of Rmrp RNA significantly (Fig. 6A). To confirm that this led to a functional reduction of RNase MRP activity, the expression of substrate RNAs for RNase MRP, i.e. Clb2 mRNA, Viperin mRNA and ITS1 pre-rRNA, were measured 4-6 . Reduced Rmrp RNA levels indeed resulted in elevated levels of RNase MRP substrates Clb2 and Viperin mRNAs, as well as an accumulation of an ITS1 pre-rRNA processing intermediate. (Fig. 6B-D). Following Rmrp knock-down, expression of Sox9 (Fig. 6E), Col2a1 (Fig. 6F), Runx2 (Fig. 6G), Col10a1 (Fig. 6H) and Alpl (Fig. 6I) was reduced at both 7 and 10 days in differentiation. Overall, expression of Runx2, Col10a1 and Alpl seemed to be more heavily affected than Col2a1. In accordance, Bapx1 mRNA was strongly induced (Fig. 6J), which is indicative of deregulated hypertrophic differentiation 35 . These data show that reduction of Rmrp RNA levels impacts the course of chondrogenic differentiation, with a prominent impact on the hypertrophic differentiation program. In concordance with the observed deregulated chondrogenic differentiation, expression of Pthrp was increased after Rmrp RNA knockdown (Fig. 6K). Since RNase MRP has been shown to be implicated in pre-rRNA processing 3, 4 , we assessed the levels of 18 S, 5.8 S and 28 S following Rmrp RNA knockdown and we observed reduced levels of 18 S and 5.8 S, but not of 28 S rRNAs at day 7 in ATDC5 chondrogenic differentiation (Fig. 6L).

Rmrp RNA knockdown affects cell proliferation, glycosaminoglycan content and mineralization of ATDC5 cells. We next investigated whether deregulation of chondrogenic differentiation by Rmrp
RNA knockdown is accompanied by functional changes of the chondrogenic differentiation program at different  (Fig. 7A). Rmrp RNA knockdown resulted in reduced cell numbers, inferring reduced proliferation (quantified in Fig. 7B and visually depicted in Fig. 7C,D). Rmrp RNA knockdown led to a significant reduction of GAG content at day 7 and this was still detectable at day 14 in chondrogenic differentiation (Fig. 7E). GAG content at day 14 before extraction of the Alcian Blue dye is visualized in Fig. 7F. Mineralization, quantified by Alizarin Red staining was reduced at day 7 and day 14 in chondrogenic differentiation (Fig. 7G) and visualized in Fig. 7H,I. In concert with the Alizarin Red data, ALP enzyme activity was almost absent at day 7 in chondrogenic differentiation as a result of Rmrp knockdown and was still significantly reduced at day 14 in chondrogenic differentiation (Fig. 7J).

Chondrogenic trans-differentiation is impaired in CHH fibroblasts. CHH is caused by mutations
in RMRP RNA 14,17 , but it is unknown whether these mutations influence chondrogenic differentiation in CHH patients. To test whether the chondrogenic capacity of CHH cells is impaired, we employed a trans-differentiation protocol that drives dermal fibroblasts towards a chondrocyte-like phenotype 34 . High density plating of fibroblasts on an Aggrecan-coated surface in the presence of TGFβ3 induced the formation of dense aggregates within 24 hours post-plating, resembling chondrogenic nodules. To focus on functional chondrogenic read-out markers we measured induction of COL10A1, COL2A1 and ALPL (alkaline phosphatase) gene expression in cultures from healthy fibroblasts, substantiating trans-differentiation into the chondrogenic lineage ( Fig. 8A-C). Induction of COL10A1 expression was more robust than COL2A1, which indicates that these cultures display a predominant hypertrophic phenotype (Fig. 8A,B). As a result of the chondrogenic trans-differentiation RMRP RNA expression levels increased (Fig. 8D). Compared to healthy control cultures we found that CHH fibroblasts (4 CHH patients, carrying different CHH pathogenic mutations in the RMRP gene (i.e. 127 G > A and 261 C > G; 4 C > T and 77 C > T; 70 A > G and 70 A > G; 4 C > T and -21_-9dup CTCTGTGAAGCTG)) displayed an impaired induction of COL10A1 and ALPL expression upon chondrogenic trans-differentiation (Fig. 8B,C). Induction of COL2A1 expression was not affected at day 3 in trans-differentiation. At days 5 and 7 during chondrogenic trans-differentiation, COL2A1 expression appeared lower in CHH fibroblasts as compared to healthy controls (Fig. 8A), although this difference was not significant. The induction of RMRP RNA expression observed in healthy control fibroblasts was absent in CHH cultures (Fig. 8D). Moreover, and in concert with the observed impaired hypertrophic differentiation (Fig. 8B,C), the expression of PTHrP was increased in CHH cultures as compared to healthy control cultures (Fig. 8E). Finally, we detected a significant increase in the accumulation of the ITS1 pre-rRNA processing intermediate 4 (Fig. 8F). Together, data show that in CHH fibroblasts chondrogenic trans-differentiation is impaired with a major impact on hypertrophic development of these cultures.

Discussion
The involvement of the RMRP lncRNA in the pathophysiology of CHH has previously been identified 17 ; however it is not clear why mutations in RMRP RNA lead to a severe skeletal dysplasia phenotype. Skeletal development depends on chondrogenic differentiation in the growth plate and therefore we hypothesized that RMRP RNA has a functional role during chondrogenic differentiation, explaining the dwarfism that characterizes CHH patients. We found that expression levels of RMRP RNA and RNase MRP protein subunits are induced and spatiotemporally regulated during the course of chondrogenic differentiation. RMRP expression levels are especially induced during the hypertrophic phase of chondrogenic differentiation, indicating an increased demand for RMRP RNA levels during this phase of the differentiation process. As long as the substrate(s) of RNase MRP that are involved in the molecular mechanisms of chondrogenic differentiation has/have not been identified, we can only speculate how induction of RMRP RNA expression is associated with chondrocyte hypertrophy. It is remarkable that in CHH tissues with fast-dividing cell types (growth plate, hair follicles, bone marrow etc.) seem to be the most affected ones 16 , suggesting a role for RMRP RNA in cell cycle regulation. Tight control over cell mitosis and differentiation is paramount to controlling growth plate development 47 . The proliferative zone of the growth plate presents a high mitotic activity, while terminally differentiated chondrocytes in the hypertrophic zone most likely lack a mitotic cycle. One of the identified activities of the RNase MRP complex is endoribonucleolytic cleavage of the mRNA of cyclin b2 (Clb2). This B-type cyclin activates cyclin-dependent kinase-1 (CDK1) during M-phase and anaphase-promoting complex (APC)-dependent proteolysis of Clb2 is essential for mitotic exit 48 . Low Clb2 levels also keep CDK1 activity to a minimum during G0 phase via similar APC-dependent proteolysis 49 . It is likely that the strictly regulated terminal differentiation of hypertrophic chondrocytes is orchestrated by mitotic exit or entrance into G0 phase. As RNase MRP activity has been shown to degrade the Clb2 message it is tempting to speculate that the elevation of RNase MRP levels during chondrocyte hypertrophy serves to decrease Clb2 protein levels by increased turnover of Clb2 mRNA to inhibit activation of CDK1. Compelling evidence from recent work indeed shows the crucial involvement of CDK1 in controlling chondrocyte hypertrophic differentiation 50 . CDK1 expression was found to be highly expressed in proliferative chondrocytes and was greatly diminished in hypertrophic chondrocytes in differentiating ATDC5 cultures, as well as in mouse growth plates. In addition genetic interference with CDK1 expression caused absence of proliferative chondrocytes in the growth plate and a switch towards hypertrophic differentiation in ATDC5 cultures, strongly suggesting that the reduction of CDK1 levels or its decreased activity is a prerequisite for hypertrophic differentiation. We thus speculate that in a healthy growth plate the increased RMRP RNA levels in hypertrophic chondrocytes may lead to decreased Clb2 levels, thereby hampering CDK1 activity and inducing an overall mitotic arrest. Work from the same group 50 shows that the proliferation-promoting and hypertrophy-suppressive action of PTHrP is, at least in part, CDK1-dependent. In agreement with this notion we observed that Rmrp RNA levels are down regulated upon PTHrP exposure, presumably via decreased Rmrp promoter transcriptional activity. This may result in elevated Clb2 levels, thereby potentially aiding in PTHrP downstream CDK1 activity. In this respect, it is important to note that knockdown of Rmrp RNA levels during ATDC5 chondrogenic differentiation increases Clb2 and Pthrp mRNA levels and deregulates chondrogenic differentiation in an overall hypertrophy-suppressing fashion.
One of the earliest identified roles of RNase MRP is endoribonucleolytic maturation of 5.8 S rRNA by cleaving site A3 in the internal transcribed spacer 1 (ITS1) in yeast 3,4 . Even though there is presently no conclusive evidence showing that specifically hypertrophic chondrocytes in the growth plate display the highest protein synthetic capacity, an RMRP RNA/RNase MRP mediated contribution to the synthesis of the large ribosomal subunit in the growth plate may support the high protein synthetic activity of the growth plate to produce protein-rich cartilaginous extracellular matrix. For long an involvement of RNase MRP in the processing of ITS1 in human cells remained elusive 51 . However, recently it has been elegantly shown that human RMRP RNA in the RNase MRP complex indeed catalyzes the endoribonucleolytic cleavage of ITS1, thereby contributing to pre-rRNA maturation 4 . Indeed our RMRP RNA knockdown data in ATDC5 show accumulation of an ITS1 pre-rRNA processing intermediate as well as reduced levels of mature 5.8 S and 18 S rRNA. In concert with these findings we detected similar accumulation of an ITS1 processing intermediate in differentiating CHH cells. Chondrogenic differentiation of ATDC5 and the developing growth plate is associated with high proliferative capacity 31 and increased synthesis of protein-rich ECM. It is conceivable that this alters the cellular demand for mature rRNAs for ribosome biogenesis, potentially explaining the changes in Rmrp RNA expression during chondrogenic differentiation and the phenotype observed after Rmrp RNA knockdown. Finally, RMRP RNA has been described to associate with TERT 7 , the reverse transcriptase that is associated with the telomerase holoenzyme. This RMRP RNA-containing macromolecular complex was demonstrated to display RNA-dependent RNA polymerase activity. This activity generates a double-stranded RMRP RNA molecule that is converted by Dicer into an siRNA that targets RMRP RNA 7 , as well as specific mRNAs 52 . Other RMRP RNA-derived siRNAs have been identified as well, but it is currently unknown whether these are also generated via a similar mechanism. The latter siRNAs, termed RMRP-S1 and RMRP-S2 8 , seem to target genes relevant for skeletal development like SOX4, PTCH2 and BMPR2. It remains elusive whether this relates to increased RMRP RNA levels in hypertrophic chondrocytes, but it is conceivable that upregulation of RMRP RNA leads to higher production of RMRP-S1 and RMRP-S2, which in turn may change the magnitude by which RMRP-S1 and RMRP-S2 targets are influenced. Our observations that the expression of RNase MRP components is modulated in differentiating cells promoted us to investigate whether RMRP RNA expression levels can be controlled at the transcriptional level. RMRP RNA expression is driven by RNA polymerase III (RNAPIII) and the proximal RMRP RNA promoter has been studied in the past to some extent 53 , showing that the −84 bp promoter region is sufficient to drive RNAPIII-dependent transcription of RMRP RNA. However, to fine-tune promoter activity and to be able to respond to alternating demands of RMRP RNA (e.g. during chondrocyte hypertrophy), it is expected that additional transcription regulatory elements are present. Indeed, WNT-3A was recently found to be able to drive RMRP transcription in cancer via activation of β-catenin and YAP proteins 54 . Here, we showed that the 1500 base pair sequence upstream of the Rmrp RNA transcription start site is responsive to a series of chondrocyte morphogens or pathways. Several of the growth factors tested increased or decreased the transcriptional activity of the minus 1500 Rmrp promoter, indeed indicating that in chondrocytic cells transcription of Rmrp RNA may be under transcriptional control of chondrogenic cues. In agreement with their known role in chondrocyte hypertrophy 45 , WNT-5A, WNT-3A and BMP-2 induced Rmrp promoter activity, suggesting that increased Rmrp RNA expression in chondrocyte hypertrophy is, at least in part, transcriptionally controlled by one or more of these morphogens and their downstream pathways. Indeed, in silico prediction of putative transcription factor binding sites in the minus 1500 bp Rmrp promoter sequence suggests the presence of CREB, MEF2, TCF/LEF and SMAD binding sites (Genomatix; data not shown) which are transcription factors acting downstream of WNT-5A, WNT-3A and BMP-2 signaling. Dorsomorphin (a BMP/SMAD1/5/8 inhibitor) exposure greatly reduced Rmrp promoter activity, further supporting BMP-mediated control of RMRP transcription. In keeping with a potential feedback mechanism it is noteworthy that BMPR2 (the type II BMP receptor) is a potential target of RMRP-S2, the RMRP RNA-derived siRNA 8 .
Active control over Rmrp transcription in chondrocyte hypertrophy is further substantiated by the observation that PTHrP and bFGF are able to reduce Rmrp promoter activity. PTHrP and bFGF delay chondrocyte hypertrophy/terminal differentiation 23,38,39,55 . At this point we can only speculate whether reduced Rmrp promoter activity and Rmrp expression by PTHrP and bFGF is an indirect result of a morphogen-dependent changing chondrocyte hypertrophic phenotype (e.g. decreased WNT or BMP signaling), or that these morphogens directly control Rmrp abundance by downstream inhibition of Rmrp promoter activity. We conclude that Rmrp transcription in chondrocytes is controlled by well-known chondrogenic signaling pathways that are associated with hypertrophic differentiation. It remains to be determined whether such transcriptional control is also present in other cell types and how this communicates with the RNAPIII transcription machinery 56 .
The clinical presentation of CHH is dominated by short stature, caused by impaired skeletal development. Our data are consistent with a pivotal role for RMRP RNA during chondrogenic differentiation and is especially linked to chondrocyte hypertrophic differentiation. Genetic interference with Rmrp RNA expression in ATDC5 cells and chondrogenic trans-differentiation of CHH fibroblasts indicates delayed chondrocyte hypertrophy. Growth plates of CHH patients are characterized by delayed ossification and the presence of very few hypertrophic chondrocytes 25 . Whether and how delayed chondrocyte hypertrophy may influence the development of the growth plate in CHH remains to be determined. Finally, considering our observations that the expression of PTHrP is induced during knockdown of Rmrp RNA and in chondrogenic differentiating CHH fibroblasts and its inhibiting action on RMRP promoter activity and RMRP RNA expression, it is interesting to realize that disrupted PTH/PTHrP signaling is deleterious for skeletal development by endochondral ossification 57 . Considering this it is tempting to speculate whether there are molecular connections between RMRP RNA and PTH/PTHrP in skeletal dysplasias like cartilage-hair hypoplasia, metaphyseal chondrodysplasia Jansen type (OMIM #156400) 58 and Blomstrand type chondrodysplasia 59 , (OMIM #215045).
In conclusion, lncRNAs involved in mesenchymal cellular differentiation processes are scarcely known 60 and we found that the RMRP lncRNA is differentially expressed during different stages of chondrogenic differentiation and displaying a prominent association with chondrocyte hypertrophic development. Our findings shed new light on potential pathobiological mechanisms involved in the skeletal dysplasia phenotype associated with cartilage-hair hypoplasia.