Sulfation is an essential modification of many proteins, carbohydrates, and lipids, and it is necessary for normal growth and development (Weinshilboum and Otterness,1994; Falany,1997). In higher organisms, all sulfation reactions are mediated through the universal sulfate donor 3′-phosphoadenosinse 5-phosphosulfate (PAPS; Schwartz et al.,1998). The synthesis of PAPS in higher animals, including mammals is mediated by the bifunctional enzyme PAPS synthase (PAPSS). In the first step, the ATP sulfurylase combines ATP and inorganic sulfate to form APS. The subsequent step is catalyzed by APS kinase, which adds an ATP to APS to form PAPS (Geller et al.,1987; Lyle et al.,1994; Rosenthal and Leustek,1995; Klaassen and Boles,1997). Subsequently, sulfotransferases (SULT) transfer sulfate, donated by PAPS, to an acceptor substrate (Weinshilboum and Otterness,1994; Falany,1997). Under physiological conditions, sulfation is regulated, in part, by the supply of the donor molecule PAPS and the transport mechanisms by which sulfated conjugates enter and leave cells (Kauffman,2004).
In human and mouse tissues, two different genes encoding the isoforms PAPSS1 and PAPSS2 have been identified (Kurima et al.,1998; ul Haque et al.,1998; Venkatachalam et al.,1998; Xu et al.,2000). The two murine Papss genes are structurally similar. Sequence comparison between the coding regions of the APS-kinase and the ATP-sulfurylase domains of the murine Papss1 and Papss2 cDNA showed 73% homology within the cDNA (unpublished data), and 76% similarity within the protein sequences (Kurima et al.,1998).
Several studies indicate that PAPSS2 activity is important for normal skeletal development. A mutation in the PAPSS2 gene was found to cause an autosomal recessive skeletal disorder in humans, spondyloepimetaphyseal dysplasia (SEMD), Pakistani type. The clinical findings include disproportionate short stature evident at birth (ul Haque et al.,1998).
A missense mutation in the Papss2 gene causes brachymorphism in the mouse. Brachymorphic (bm) mice have abnormal hepatic detoxification, bleeding times, and postnatal growth (Kurima et al.,1998). The skeletal phenotype of bm mice can be attributed to reduced sulfation of the proteoglycans of the cartilage extracellular matrix (Orkin et al.,1976,1977; Schwartz and Domowicz,2002). It is intriguing that the Papss2 mutation specifically affects postnatal skeletal development. The skeleton of newborn bm mice appears to be normal, but postnatally they show a progressive reduction in size of the columnar and hypertrophic zones in the epiphyseal growth plates concomitant with the alteration in growth.
Despite the known PAPSS2-related skeletal abnormalities and functional studies in vitro, only limited information is available on the expression of the molecule in cartilage (Fuda et al.,2002). Therefore, our aim was to determine the precise spatial and temporal expression pattern of Papss2 during murine cartilage/bone development to gain insight into the pathogenesis of skeletal dysplasias.
RESULTS AND DISCUSSION
Reverse transcriptase-polymerase chain reaction (RT-PCR), which was used to perform a comparative analysis, revealed that Papss1 appears to be the dominant Paps synthase isoform in whole embryos of developmental stage embryonic day (E) 11.5, E12.5, and E13.5 (Fig. 1). Expression of the Papss2 is more variable, showing lower expression at E12.5 compared with E11.5 and E13.5, respectively.
To elucidate the participation and particular role of PAPSS2 during skeletogenesis, we have performed in situ hybridization of whole-mounts from 11.5–12.5 dpc and of histological sections from 13.5 dpc and 16.5 dpc mouse embryos, as well as from the appendicular skeleton of newborn mice.
Due to the high homology between the coding region of the Papss1 and Papss2 mRNA, false-positive expression patterns may be detected as a result of cross-hybridization in in situ experiments. Therefore, we generated probes that were specific for each of the Papss1 and Papss2 mRNA derived from the 3′-untranslated region and a second Papss2 probe covering the translated and 3′-untranslated sequence.
According to our study, Papss2 mRNA expression begins at precartilaginous stages at sites where the skeleton will be formed. At 11.5 dpc, hybridization with Papss2-specific riboprobes yielded signals in the developing limb buds. In the proximal part of the forelimb, faint expression was seen in the developing scapula, while stronger expression was detected in humerus, radius, and ulna. Weaker signals were visible in the distal part of the forelimb. A similar but weaker expression pattern was seen in the hindlimb. An additional expression domain was detected in the developing nasal cartilage (Fig. 2A). The Papss2 expression domains expanded during embryogenesis, concomitant with the condensations of the cartilaginous precursors of the skeleton, as seen in the developing appendicular skeleton of 12.5 dpc embryos (Fig. 2B). Papss2 mRNA-specific signals in the forelimb of 12.5 dpc embryos showed the ongoing formation of the scapula, humerus, ulna, radius, and the fingers (Fig. 2B). Other expression domains in precartilaginous tissues were present in the somites of the developing vertebral column and the otic vesicle (Fig. 2B). At 13.5 dpc, when the major switch from condensed mesenchymal cells toward fully differentiated chondrocytes occurs in the mouse, hybridization of tissue sections confirmed Papss2 mRNA to be present in all cartilage (Fig. 2C,F,G). Noncartilaginous Papss2 expression was visible in the brain, and also in cells of endocardial origin in the heart and in the connective tissue of the tongue (Fig. 2C–E).
Papss2 mRNA continued to be expressed at 16.5 dpc in all cartilaginous sites tested and is shown here in the axial and appendicular skeleton as well as the tracheal, nasal, and Meckel's cartilage (Fig. 2H–M). It appears that Papss2 mRNA is absent in hypertrophic chondrocytes, which is demonstrated in the cartilage primordium of the basioccipital bone in Figure 2K. The most prominent nonskeletal expression domains of Papss2 mRNA at 16.5 dpc, were detected in dorsal root ganglia and in the cortex of developing kidneys (Fig. 2N,O).
We tested the same tissues for expression of Papss1 mRNA. In contrast to Papss2 mRNA, Papss1 mRNA was not detectable in embryonic stages E11.5, E12.5, and E13.5 when performing whole-mount in situ hybridization. Only in cartilage of the 13.5 dpc embryo was faint Papss1 mRNA expression seen, overlapping with the Papss2 mRNA expression. Papss1 mRNA was not detectable in cartilage at later developmental stages (data not shown). These results confirm the observations of Fuda et al. (2002), showing less Papss1 than Papss2 mRNA in immature cartilage of the guinea pig by RT-PCR analysis.
Investigation of skeletal tissues from 13.5 dpc and 16.5 dpc embryos as well as from newborn mice indicates that the expression of Papss2 mRNA appears mainly in the proliferating and differentiating chondrocytes and is strongly down-regulated in hypertrophic chondrocytes (Fig. 2K). To show this finding in more detail, in situ hybridization on the tibiae of newborn mice was performed. At the newborn stage, the epiphyses of the mouse skeleton are still completely cartilaginous and secondary ossification is not yet initiated.
The localization of Papss2 signal was determined by comparison with several well-delineated cartilage markers. Sox9 is a marker for proliferating and resting chondrocytes and can be detected in the entire nonhypertrophic region of the epiphysis, whereas Col10a1 mRNA is restricted to the hypertrophic chondrocytes. Indian hedgehog is a marker for prehypertrophic chondrocytes, located at the transition zone between cartilage proliferation and hypertrophy (Fig. 3). When comparing the Papss2 mRNA patterns to these marker genes, it clearly shows a dominant expression in the resting, proliferating, and prehypertrophic chondrocytes (Fig. 3). The Papss2 expression in cartilage is very similar to that of Sox9. Therefore, the Papss2 gene might be a candidate for transactivation by Sox9 as a regulator of its expression in cartilage or vice versa.
There is some diffuse staining for Papss2 mRNA throughout the region of hypertrophic cartilage, which might be either a residual weak signal or just a diffusion of the dye in the staining reaction, because there is no staining within the nuclei of the cells, like for Ihh. However, specific staining for Papss2 mRNA is present in bone (Fig. 3).
In summary, our analysis is an in situ documentation of Papss2 mRNA expression in cartilage. These data confirm previous studies that demonstrated PAPSS2 and not PAPSS1 is the dominant isoform in developing cartilage (Fuda et al.,2002) and support in vitro studies of Cho et al. (2004) that PAPSS2 is a major player in cartilage proteoglycan sulfation. Surprisingly, the bm mutation shows no obvious phenotype involving the embryonic mouse skeleton, suggesting that the weak expression of Papss1 can partially rescue the PAPSS2 inactivation in embryonic development. However, later in life, Papss1 expression might not be sufficient to completely compensate for PAPSS2 inactivation, causing postnatal growth delay in bm mice. In humans, PAPSS1 obviously cannot complement PAPSS2 activity in the developing embryo, resulting in the short stature of SEMD Pakistani type, which is already evident at birth. Our data also suggest that this skeletal dysplasia might be caused by a proliferation defect rather than a delay in maturation or differentiation. Furthermore, the PAPSS family might likely grow, as there are additional uncharacterized expressed sequence tags with a high amino acid identity to PAPSS1 and PAPSS2 (Kurima et al.,1998), possibly substituting for deficiency of known PAPSS activity. This finding may also be significant for the nonskeletal Papss2-expressing tissues shown in our study. Finally, the understanding of PAPSS2 expression and function might help to elucidate the pathogenesis of degenerative joint diseases such as arthritis (Ford-Hutchinson et al.,2005) or other skeletal conditions.
Isolation of total RNA, removal of contaminating genomic DNA, and cDNA synthesis were performed as previously described (Spangenberg et al.,1998). PCR amplification was done on 1 μl of the resulting cDNA as follows: Papss1 (94°C/1 min, 60°C/1 min, 72°C/1 min; 27 cycles), Papps2 (94°C/1 min, 60°C/1 min, 72°C/1 min; 27 cycles), pyruvate dehydrogenase (PDH; 94°C/1 min, 60°C/1 min, 72°C/1 min; 25 cycles). All programs additionally included an initial denaturation step (94°C/3 min) and a final elongation step (72°C/7 min). Forward (F) and reverse (R) primers used in this study were as follows: Papss1 F: 5′-aaacctccagagggcttcat-3′; Papss1 R: 5′-gacagaccaaaaccaatgcac-3′; PDH-F, 5′-gtgggccactgcctagaag-3′; PDH-R, 5′-gggcatcaaggaagttgaat-3′.
Generation of In Situ Hybridization Probes
Total RNA was isolated from differentiated ATDC5 cells (RIKEN cell bank) using Trizol (Invitrogen) according to the manufacturer's instructions. RT-PCR was performed with SuperScript II reverse transcriptase (Invitrogen) and gene-specific primers under standard conditions. Primers used were as follows. Papss2 cDNA fragment 1 (coding region and 3′ untranslated region, NM_011864: nt2480 to nt3110): forward 5′-CATGCCCTCATGATGCAGG-3′; reverse 5′-GAGCACTTGAGGAAGAAGCC-3′. Papss2 cDNA fragment 2 (3′ untranslated region, AK128962: nt2006 to nt2522): forward 5′-CTGGCTCTGGCTTCTTCCT-3′; reverse 5′-TTGGGCAGTGATGGTATGTGTATG-3′. Sox9 cDNA (NM_011448: nt969 to nt1620): forward 5′-TCTTCAAGGCGCTGCAAGCCG-3′; reverse 5′-GGGCTGTAGGAGATCTGTTGC-3′. Col10a1 cDNA (NM_009925: nt1611 to nt2035): forward 5′-GCCGCTTGTCAGTGCTAACC-3′; reverse 5′-GAGCCACTAGGAATCCTGAG-3′.
PCR products were cloned into pCRR-BluntII-TOPOR vector (Invitrogen), sequenced, and linearized with appropriate restriction enzymes for generation of the templates for in vitro RNA synthesis. Antisense and sense riboprobes were transcribed with SP6/T7 RNA-polymerase under the incorporation of digoxigenin (DIG) -11-UTP using a DIG RNA labeling kit (Roche) according to the manufacturer's protocol (Roche).
In Situ Hybridization of Whole-Mounts and Sections
For in situ hybridization embryos of specific gestation stages (E11.5, E12.5, E13.5, E16.5) were generated. Embryos and hind legs of newborn mice were fixed overnight in 4% paraformaldehyde in PBS at 4°C. For whole-mount in situ hybridization, embryos were washed twice in PBS, dehydrated through a series of increasing methanol concentrations, and stored in methanol at −20°C. Whole-mount in situ hybridization was performed as described by Rosen and Beddington (1993) with slight modifications. Briefly, after rehydration, the embryos were bleached with 6% H2O2, followed by digestion with 20 μg/ml proteinase K in 10 mM Tris/HCl, pH 7.0, 1 mM ethylenediaminetetraacitic acid for 3–5 min, and then refixed in 4% paraformaldehyde/PBT/0.2% GA for 20 min. The embryos were hybridized with DIG-11-UTP–labeled sense and antisense probes overnight at 65°C as described above. After the first two posthybridization washes in hybridization solution (65°C), embryos were treated with 100 μg/ml RNase A in 0.5 M NaCl, 10 mM Tris/HCl, pH7.0, 0.1% Tween20. Afterward, embryos were washed for 3 hr in 4× standard saline citrate (SSC)/50% formamide, 0.1% Tween20. All further steps were performed according to Spörle and Schughart (1998). Embryos were analyzed using an Olympus binocular microscope and documented with a high-resolution CCD color camera (Color View 8, Olympus).
For in situ hybridization of sections, embryos and hind legs of newborn mice were fixed as described above and dehydrated through a series of increasing ethanol concentration. The specimens were then transferred to xylene followed by xylene/paraffin and finally embedded in paraffin. Embedded tissues were cut at 7 μm and mounted on microscope slides. In situ hybridization of tissue sections was performed as described by Dietz et al. (1999) with a slightly modified protocol for the posthybridization washes. In brief, specimens were washed after hybridization twice in 2× SSC at room temperature, followed by three washes with 1× SSC at 60°C. After RNase treatment a second series of three washes was performed at 60°C.
Hybridization products were detected by alkaline phosphatase reaction using BM-purple (Roche) as a substrate for staining. Stained sections were mounted with cover slips in Kaiser's glycerol gelatin (Merck) and photographed under a Zeiss Axioplan 2 microscope.
We thank Melanie Finger and Sarah Kujawski (Aventis Pharma Frankfurt) for excellent performance of in situ hybridization on tissue sections, Dr. Nicole Gerwin for reading of the manuscript, and Drs. Eckart Bartnik (Aventis Pharma Frankfurt) and Thomas Aigner (Department of Pathology, Friedrich-Alexander-University Erlangen Nürnberg) for the Coordination of the BMBF funded Leitprojekt “Diagnose und Therapie der Osteoarthrose mit den Mitteln der molekularen Medizin,” which made the presented work possible. The cDNA probe for in situ hybridization of indian hedgehog was kindly provided by Prof. Dr. Andrea Vortkamp (University Essen-Duisburg). We also thank Dr. Bernd Kirschbaum (Vice President, DG Thrombosis/Degenerative Joint Diseases) for supporting the research in Osteoarthritis at Aventis Pharma. Dr. Christiane Stelzer (Children's Hospital, University of Mainz) was funded by a BMBF fellowship within the above-mentioned project (Subproject Prof. Zabel: 01GG9825).