The Ribosomal Protein QM Is Expressed DifferentiallyDuring Vertebrate Endochondral Bone Development

HELEN GREEN,1 ANN E. CANFIELD,1,2 M. CHANTAL HILLARBY, 1,3 MICHAEL E. GRANT,1

RAYMOND P. BOOT-HANDFORD,1 ANTHONY J. FREEMONT,3 and GILLIAN A. WALLIS 1,3

ABSTRACT

Endochondral ossification is a carefully coordinated developmental process that converts the cartilaginousmodel of the embryonic skeleton to bone with accompanying long bone growth. To identify genes that regulatethis process we performed a complementary DNA (cDNA) subtractive hybridization of fetal bovine prolifer-ative chondrocyte cDNA from epiphyseal cartilage cDNA. The subtracted product was used to screen a fetalbovine cartilage cDNA library. Ten percent of the clones identified encoded the bovine orthologue of thehuman ribosomal protein “QM.” Northern and western blot analysis confirmed that QM was highly expressedby cells isolated from epiphyseal cartilage as opposed to proliferative chondrocytes. In contrast, no detectabledifference in the expression of mRNA for the ribosomal protein S11 was detected. Immunohistochemicalanalysis of fetal bovine limb sections revealed that QM was not expressed by the majority of the epiphysealchondrocytes but only by chondrocytes in close proximity to capillaries that had invaded the epiphysealcartilage. Strongest QM expression was seen in osteoblasts in the diaphyseal region of the bone adjoining thegrowth plate, within the periosteum covering the growth plate and within secondary centers of ossification.Hypertrophic chondrocytes within the growth plate adjoining the periosteum also were positive for QM aswere chondrocytes in the perichondrium adjoining the periosteum. In vitro investigation of the expression ofQM revealed higher QM expression in nonmineralizing osteoblast and pericyte cultures as compared withmineralizing cultures. The in vivo and in vitro expression pattern of QM suggests that this protein may havea role in cell differentiation before mineralization. (J Bone Miner Res 2000;15:1066–1075)

Key words: subtractive hybridization, endochondral ossification, mineralization, QM, ribosomal proteins

INTRODUCTION

DURING VERTEBRATE development, the cartilaginousmodel of the axial and appendicular embryonic skele-

ton is replaced by bone by the process of endochondralossification.(1) This process is initiated early in embryogen-esis when chondrocytes within the center of the cartilagi-nous model are stimulated to undergo discrete stages ofdifferentiation that include proliferation, maturation, andhypertrophy. The matrix within the region of chondrocyte

hypertrophy becomes calcified and, after vascular invasion,the cartilage is replaced by bone. This process radiatesoutward with the development of the growth plates, whichseparate the cartilaginous epiphyses from the bony diaphy-sis. Within the growth plates, chondrocytes are organizedinto zones of differentiation, with the proliferative zone atthe epiphyseal margin and the hypertrophic zone at thediaphyseal margin. Later in development, secondary centersof ossification occur within the epiphyses. At puberty, theprimary and secondary centers of ossification fuse and the

1The Wellcome Trust Centre for Cell-Matrix Research, School of Biological Sciences, University of Manchester, Manchester, U.K.2Medical School, University of Manchester, Manchester, U.K.3Musculoskeletal Research Group, University of Manchester, Manchester, U.K.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 15, Number 6, 2000© 2000 American Society for Bone and Mineral Research

1066

endochondral ossification process and long bone growthstop.

Many studies have identified factors that regulate differ-ent aspects of the endochondral ossification process includ-ing chondrocyte differentiation within the growth plate,vascular invasion of the growth plate, osteoblast differenti-ation, and extracellular matrix calcification.(2–7) However,little is known about the factors that regulate the initialformation of the primary and secondary ossification centers.Previous work in our laboratory has utilized a polymerasechain reaction (PCR)–based subtractive hybridization tech-nique to identify genes that are switched on by chondrocytesas they enter and progress down the differentiation pathwaywithin the growth plate.(8,9) To identify genes that mayinhibit entry of chondrocytes into this differentiation path-way and that may be involved in the regulation of secondarycenter formation, we have now performed the reversesubtraction. For this purpose, two complementary DNA(cDNA) pools were generated from RNA isolated from fetalbovine cells extracted from epiphyseal cartilage and fromRNA extracted from proliferative chondrocytes isolatedfrom the growth plate. The PCR-based subtractive hybrid-ization technique was used to generate an epiphyseal minusproliferative cDNA pool after two rounds of subtraction.The subtractive product was used to screen a custom-madefetal bovine cartilage and growth plate cDNA library and 40positive clones were picked for further study. We found that10% of these clones represent the bovine orthologue of thehuman ribosomal protein “QM” and that this ribosomalprotein has a distinctive pattern of expression in the devel-oping bone. QM therefore may have a role in the differen-tiation of bone forming cells before mineralization in addi-tion to its putative “housekeeping” role in protein synthesis.

MATERIALS AND METHODS

Chondrocyte isolation

Growth plate and epiphyseal chondrocytes were isolatedfrom second and third trimester bovine fetuses(10) and thegrowth plate chondrocytes were fractionated on Percoll(Pharmacia, Uppsala, Sweden) density gradients as previ-ously described.(11) Previous studies have shown that thegrowth plate fractions 3–6 are enriched for chondrocytesfrom the hypertrophic zone, fraction 7 for maturing chon-drocytes, fractions 8 and 9 for proliferative chondrocytes,and fraction 10 represents a mixture of proliferative andepiphyseal chondrocytes.(8,10)

RNA extraction and poly(A) cDNA subtractivehybridization

RNA was extracted from the fractionated growth platechondrocytes and epiphyseal cells using the acid phenolmethod.(12) The cDNA and poly(A) cDNA products wereprepared from the extracted RNA as previously de-scribed.(8,13,14)For the subtraction, the poly(A) cDNA prod-ucts generated from Percoll gradient fractions 8 and 9 werepooled. A PCR-based poly(A) cDNA subtractive hybridiza-tion method(8,13,15) was used to carry out two rounds of

subtraction of epiphyseal minus proliferative poly(A)cDNA products.

cDNA probes

The bovine cDNA probes for type X collagen, type IIcollagen, matrix GLA protein (MGP) ribosomal proteinS11, and QM were isolated from alZAP Express/EcoRI/Xho I custom-made cDNA library prepared from fetal bo-vine epiphyseal cartilage and growth plate messenger RNA(mRNA; Stratagene, LaJolla, CA, U.S.A.).

Random oligo labeling of cDNA probes

For random prime labeling of probes, 50 ng of DNA waslabeled witha-[32P]-deoxycytosine triphosphate (a-[32P]-dCTP) (50mCi; Amersham Int Plc, Little Chalfont, Buck-inghamshire, U.K.) using an oligo labeling kit (Pharmacia,Uppsala, Sweden). Probes were purified on Sephadex G-50Nick purification columns (Pharmacia).

Library screening

The product of the subtractive hybridization was used toscreen the bovine chondrocyte cDNA library. Approxi-mately 200,000 plaques were screened using standardplaque hybridization techniques. Positive plaques wereidentified by primary screening and purified by secondaryscreening.

Southern blot hybridization

Poly(A) cDNA products separated on 2% (wt/vol) aga-rose gels were transferred to Hybond N nylon membranes(Amersham) following standard procedures.(14) Prehybrid-ization and hybridization of the filters with radiolabeledprobes were carried out at 37°C in a solution of 50%(vol/vol) formamide, 43 SSC, 1 mg/ml Herring spermDNA (Sigma Chemicals Co., Ltd. Poole, Dorset, U.K.), 13Denhardts, and 0.1% (wt/vol) sodium dodecyl sulfate(SDS). After hybridization, the filters were washed at in-creasing stringency to 0.23 SSC/0.l% (vol/vol) SDS at68°C and exposed to X-ray film (Kodak X-Omat AR, Roch-ester, NY, U.S.A.) at270°C with intensifying screens.

Northern blot hybridization

Total RNA was electrophoresed on a 0.8% agarose gelcontaining 2.2 M formaldehyde and transferred onto PRO-TRAN nitrocellulose (Schleicher & Schuell, Dassel, Ger-many). The RNA was fixed to the membrane by baking ina vacuum oven for 2 h at 70°C. Prehybridization and hy-bridization of the filters with radiolabeled probes were car-ried out at 68°C as described previously.(16)

Sequence analysis

ThelZAP vector system allows excision and propagationof cDNA inserts in a pBK-CMV phagemid (Stratagene,LaJolla, CA, U.S.A.). The excision procedure provided by

1067QM EXPRESSION DURING MINERALIZATION

the manufacturers of the library was followed in the gener-ation of pBK-CMV phagemids from the positive plaques.Phagemid DNA was purified using the QIAGEN miniprepkit (Qiagen, Inc, Dorking, Surrey, U.K.). Sequencing reac-tions were performed using the ABI PRISM Dye Termina-tor Cycle Sequencing Ready Reaction kit (Perkin-Elmer,Foster City, CA, U.S.A.). The DNA was resolved on an ABI377 automated sequencer (Applied Biosystems; Perkin-Elmer).

SDS-polyacrylamide gel electrophoresis andimmunoblot analysis

Cell pellets were resuspended in sample buffer (10 mMTris-HCl, pH 6.8, 10% [vol/vol] glycerol, 1% [wt/vol] SDS,0.05% [wt/vol] bromophenol blue, and 1% [vol/vol]b-mercaptoethanol) at a concentration of 1000 cells/ml andboiled for 3 minutes. Samples were separated by polyacryl-amide gel electrophoresis (PAGE) with SDS using 10%polyacylamide gels.(17) Protein was detected by stainingwith Coomassie blue (0.2% [vol/vol] Coomassie blueG-250 in 50% [vol/vol] methanol and 10% [vol/vol] aceticacid).(18) For immunoblot analysis, gels were transferred toPROTRAN nitrocellulose (Schleicher & Schuell). The blotswere probed with the C-17 anti-QM (Santa Cruz Biotech-nology, Inc., Santa Cruz, CA, U.S.A.) primary antibody ata final concentration of 0.1mg/ml and a goat anti-rabbitimmunoglobulin G (IgG)–horseradish peroxidase conjugatesecondary antibody and developed using the enhancedchemiluminescence ECL chemiluminescent system (Amer-sham).

Preparation of bovine tissue sections

Femurs were obtained from first trimester bovine fetalcalves and cut perpendicular to the longitudinal axis. Theepiphysis and the growth plate were sliced longitudinallyinto sections 3 mm in thickness. The sections were imme-diately fixed in formalin for 2 h and decalcified in formalinand 125 mM EDTA, pH 8.5, for 48 h. The sections wereembedded in paraffin wax, cut into sections of 5mm inthickness, and placed on 3-aminopropyltriethoxysilane(Sigma)-coated microscope slides.

Immunohistochemistry

Deparaffinized sections were incubated in methanol/HClcontaining 0.3% (vol/vol) hydrogen peroxide for 30 minutesto block any endogenous peroxidase activity. The sectionswere rinsed with 0.05 M Tris-buffered saline (TBS) pH 7.6before pretreatment with trypsin (1 mg/ml; Sigma) for 12minutes at 37°C in a humidity chamber. All subsequentsteps were performed in the humidity chamber at roomtemperature. After rinsing with TBS, the sections wereincubated with 20% (vol/vol) normal swine serum for 20minutes to block nonspecific staining. A 0.1-mg/ml dilutionin buffer (1% [wt/vol] bovine serum albumin [Sigma] in0.05 M TBS containing 0.01% [wt/vol] sodium azide) of theQM primary antibody was applied and incubated for 1 h.After 3 washes with TBS the sections were incubated with

the biotinylated swine anti-rabbit secondary antibody di-luted 1/300 in buffer. The sections were rinsed as before andincubated with horseradish peroxidase streptavidin (1:500dilution in TBS) for 30 minutes. The 3,39-diaminobenzidine/H2O2 system (Sigma) was used to detect the antibody com-plexes. The sections were counterstained in Mayers’ hemo-toxylin. Incubations replacing the primary antibody withdilution buffer and normal rabbit serum were used as negativecontrols.

Pericyte and MC3T3-E1 cell culture

Pericytes were isolated from bovine adult microvessels,cultured, and characterized as previously described.(19–21)

RNA was isolated from the pericyte cultures at confluenceand from picked mineralized nodules.(22–24) The pickedmineralized nodules were removed individually from thebase of the culture flask using a sterile needle.(24) Poly(A)cDNA products were prepared from the RNA as describedabove.

MC3T3-E1 cells (kindly provided by Dr. R. Francheschi,University of Michigan, U.S.A.), were cultured ina-minimumessential medium (a-MEM; Gibco BRL, Gaithersburg, MD,U.S.A.) supplemented with 10% fetal calf serum (FCS; GibcoBRL), 100 U/ml of penicillin, and 100mg/ml of streptomy-cin.(25,26) Cells were seeded at an initial density of 23 104

cells/cm2 and were grown in 10% FCS-MEM containing 50mg/ml of ascorbic acid in the presence or absence of 10 mMb-glycerophosphate for 3 weeks. The medium was changedevery 3 days. To detect calcium deposits, cells were rinsedtwice in phosphate-buffered saline (PBS) and incubated for 1 hin 0.1% alizarin red S solution at room temperature. As pre-viously shown, the presence ofb-glycerophosphate results inthe deposition of a mineral by these cells.(27) RNA and proteinwas prepared from the cells as described above.

RESULTS

Generation of an enriched epiphyseal cartilagecDNA pool

RNA, cDNA, and poly(A) cDNA products were gener-ated from cells extracted from fetal bovine epiphyseal car-tilage and from bovine growth plate proliferative chondro-cytes. The proliferative chondrocytes were pooled fromPercoll density gradient fractions 8 and 9(8). Two rounds ofsubtractive hybridization were performed using a ratio ofepiphyseal poly(A) cDNA to proliferative poly(A) cDNA of1:20 in each round of subtraction. Southern blot hybridiza-tion using the subtracted product as a probe showed that thesubtracted product was enriched for epiphyseal poly(A)cDNA species as opposed to proliferative chondrocytepoly(A) cDNA species (Fig 1). The subtracted product wasthen used to screen thelZAP Express/EcoRI/Xho I fetalbovine cartilage cDNA library. From an initial screen ofapproximately 200,000 plaques, 40 positive plaques werepurified and sequenced.

1068 GREEN ET AL.

Isolation of the bovine orthologue of human QM

BLAST search comparisons with sequences from theNational Center for Biotechnology Information (NCBI;http://ncbi.nlm.nih.gov/BLAST/) revealed that the se-quences of 4 of the 40 cDNA clones had 96% homology atthe cDNA level and 99% homology at the protein level tothat of the human QM gene and gene product, respective-ly.(28) More detailed comparisons of the translated bovinecDNA sequence(29) revealed significant homology with thehuman QM protein,(28) the mouse 60S ribosomal proteinL10,(30) the chick Jun-binding protein (Jif-1)(31) and theSaccharomyces cerevisiaeubiquinol-cytochrome C com-plex subunit VI requiring protein (QSR1)(32) (Fig 2). Thefour cDNAs isolated here therefore are likely to representthe bovine orthologue of human QM, as well as the chick,mouse, and yeast proteins that have been given alternatenames.

Southern, Northern, and Western analysis ofQM expression

To examine the expression of QM in the growth plate,Southern blots were prepared from poly(A) cDNA productsgenerated from RNA isolated from the Percoll density gra-dient growth plate fractions 3–10 and from epiphyseal car-tilage. The Southern blots were hybridized with bovine typeII and type X collagen cDNA probes and the bovine QMcDNA probe (Fig 3A). As expected, the type II collagenprobe hybridized to all fractions whereas the type X colla-gen probe hybridized to Percoll density fractions 3–7, whichrepresent the maturing zone (fraction 7) and the hypertro-phic zone (fractions 3–6) of the growth plate.(8) The bovineQM probe hybridized strongly to the epiphyseal cartilageproducts, less strongly to products generated from fraction10, and weakly to the products of the other growth platefractions.

The pattern of expression of the QM was further inves-tigated by Northern analysis (Fig 3B). For this purpose, theQM cDNA was used as a probe against equal quantities of

total RNA isolated from epiphyseal cartilage and fromgrowth plate chondrocytes (Fig 3B). Northern blot analysisshowed that there was a higher proportion of QM mRNA inepiphyseal RNA samples as opposed to growth plate RNAsamples despite the levels of 28S and 18S ribosomal RNAbeing similar in both populations. In contrast, Northern blotanalysis using the cDNA representing the ribosomal proteinS11, for which there is no published evidence of differentialexpression, dectected no difference in the level of S11mRNA in the two RNA samples. These findings show thatin this instance, the differences in the level of QM expres-sion cannot be accounted for by differences in the ribosomalcontent of the representative cell populations. Western blotanalysis using the QM antibody revealed expression of aprotein of the predicted size for QM (25 kDa) in cellextracts from epiphyseal cartilage and chondrocytes in frac-tion 10. No QM protein could be detected in protein extractsfrom chondrocytes fractions 8 and 9 (Fig. 3C).

Immunohistochemical analysis of QM expression

Immunohistochemical analyses of fetal bovine limb sec-tions revealed that QM was not expressed by the majority ofthe chondrocytes within the epiphysis (Fig. 4, Bi). However,

FIG. 1. Specificity of the subtracted probe for poly(A)cDNA products from epiphyseal cells. (A) Poly(A) cDNAproducts from epiphyseal cells (i), proliferative chondro-cytes (ii), and the second round of subtractive hybridization(iii) separated on a 2% agarose gel and stained withethidium bromide. (B) The agarose gel in (A) was Southernblotted and hybridized to the radiolabeled poly(A) cDNAproducts from the second round of subtraction.

FIG. 2. Alignment of the translated sequence of the bo-vine QM cDNA with the QM proteins of other species. Thesequence comparison was carried out by MULTALIN ver-sion 5.3.3. Lines indicate regions of identical sequence. Thesequences have the accession numbers P27635 (human:Homo sapiens, QM), P45634 (mouse:Mus musculus, 60SRibosomal L10), Q08200 (chick:Gallus gallus, JIF-1), andP41805 (yeast:Saccharomyces cerevisiae, QSR1). TheGenbank accession number for the bovine QM cDNA isAF143815.

1069QM EXPRESSION DURING MINERALIZATION

QM was clearly expressed by epiphyseal chondrocytes inclose proximity to capillaries that had invaded the epiphy-seal cartilage and by cells within these capillaries (Fig. 4,Bii). QM staining was observed by cells within and close tothe junction between the periosteum and the perichondriumwith progressively less staining in the perichondrium in thedirection of the epiphysis (Fig. 4, Biii). The periosteumbecame progressively more positive for QM in the directionof the diaphysis of the bone. The strongest expression ofQM was exhibited in osteoblasts in the diaphyseal region ofthe bone adjoining the growth plate (Fig. 4, Biv). The cellsof the periosteum stained strongly for QM with the highestexpression being in cells close to the growth plate (Fig. 4,Bv). Hypertrophic chondrocytes within the growth plateadjacent to the periosteum also were positive for QM,whereas the remaining hypertrophic chondrocytes werelargely negative for QM as were the maturing and prolifer-ative chondrocytes of the growth plate (Fig. 4, Biv). Osteo-blasts deeper in the diaphyseal region of the bone wereprogressively less positive for QM and osteocytes werenegative for QM (Fig. 4, Bvi).

QM expression in MC3T3-E1 and pericyte cultures

The pattern of expression of QM in fetal bovine limbsections (Fig. 4B) suggested that the QM protein may havea role in the differentiation of cells before their mineraliza-tion. We investigated this observation further using theosteoblast cell line MC3T3-E1 and a pericyte cell culturesystem. Both cell culture systems were investigated be-cause, in the MC3T3-E1 cell culture system, the mineral-ization is not spontaneous, requires the addition ofb-glycerophosphate, and often is diffuse and patchywhereas pericytes are known to differentiate into boneforming cells and to form spontaneously mineralized nod-ules in culture.(33) Northern blot analysis with the QMcDNA against equal amounts of RNA from both nonmin-eralized and mineralized MC3T3-E1 cultures that had beencultured for the same length of time revealed that QMexpression was higher in nonmineralizing osteoblast cul-tures as opposed to the mineralizing osteoblast cultures (Fig.5A). Western blot analysis showed QM protein was presentin the osteoblast MC3T3-E1 cell line (Fig. 5B). To inves-tigate the expression of QM by nonmineralizing and min-eralizing pericytes, poly(A) cDNA products were preparedfrom confluent pericyte cultures and from pericytes withinpicked mineralized nodules. The poly(A) cDNA productswere Southern blotted and hybridized with the QM cDNA,MGP cDNA, and ribosomal S11 cDNA probes (Fig. 5C).QM hybridized strongly to the poly(A) cDNA from theconfluent pericyte cultures but not from the picked miner-alized nodules. MGP, as previously described,(33) hybrid-ized strongly to the poly(A) cDNA products from the pickedmineralized nodules but not to the poly(A) cDNA productsfrom the confluent pericyte cultures. In contrast, the ribo-somal S11 cDNA probe hybridized to both the pickedmineralized nodule poly(A) cDNA and the confluent culturepoly(A) cDNA.

FIG. 3. Expression of QM by cells isolated from epi-physeal cartilage and growth plate cells. (A) Poly(A) cDNAsamples were prepared from growth plate chondrocytes inPercoll density fractions 3–10 and from cells extracted fromepiphyseal cartilage (Epi). Poly(A) cDNA products sepa-rated on a 2% agarose gel stained with ethidum bromide (i).The gel in i was Southern blotted and hybridized to radio-labeled probes for type X collagen (ii), type II collagen (iii),and QM (iv). (B) Total RNA was prepared from bovineepiphyseal and growth plate chondrocytes (i). The RNA wasthen blotted onto nitrocellulose and probed with ribosomalS11 (ii) and QM (iii). (C) Western blot of QM againstprotein extracted from fractionated proliferative chondro-cytes and cells extracted from epiphyseal cartilage. Proteinfrom Percoll fractionated proliferative chondrocytes (frac-tions 8–10), epiphyseal cartilage (Epi). The relative positionof the 22-kDa and 30-kDa standards are indicated.

1070 GREEN ET AL.

FIG. 4. Immunohistochemical analysis of QM expression pattern within the developing bone. (A) On the left is aschematic diagram of the epiphyseal region of the bone indicating the regions from where the sections in Fig. 4B arederived. On the right is an expanded schematic diagram of the populations of chondrocytes within the growth plate and,in brackets, the Percoll gradient fraction that contains that population of chondrocytes (see Fig. 3A). (B) Negative stainedepiphyseal chondrocytes (i), capillary within the epiphyseal region (ii), epiphyseal perichondrial junction (iii), growth plateplus diaphyseal junction (iv), growth plate periosteal junction (v), and region of bone within the diaphysis (vi). Brownstained cells are positive for QM. Black arrows indicate QM positive cells; red arrows indicate QM negative cells. B, bone;Epi, epiphyseal region; H, hypertrophic zone; M, maturing zone; Ob, osteoblasts; Oc, osteocytes; P, proliferative zone;P.ch, perichondrium; P.ost, periosteum

1071QM EXPRESSION DURING MINERALIZATION

DISCUSSION

We have used a PCR-based subtractive hybridizationtechnique to isolate genes expressed by fetal bovine cellsextracted from epiphyseal cartilage but not by proliferativechondrocytes extracted from the epiphyseal growth plate.This subtractive hybridization experiment led to the identi-fication of differentially expressed clones of which 10%represented the bovine orthologue of the human ribosomalprotein QM. The increased expression of QM within epiph-yseal cartilage as opposed to the growth plate was con-firmed by Southern, Northern and Western blotting. North-ern blot analysis with the cDNA for the consitutivelyexpressed ribosomal protein S11 showed that the differen-tial expression of QM was not a consequence of the alteredribosomal content of the cell populations.

To more closely examine the expression of QM duringendochondral bone formation, we performed immunohisto-chemistry on fetal bovine limb sections using a QM-specificantibody. As expected, based on the subtraction results, wefound that QM was expressed by cells within the epiphysealregion but, surprisingly, that this expression was confined tocells within the capillaries, as well as to chondrocytes inclose proximity to these capillaries. It is known that thevascular invasion of the epiphyses signals the formation ofsecondary centers of ossification and so it is possible thatchanges in the microenvironment of chondrocytes in closeproximity to the capillaries induces the expression of QMby the chondrocytes. However, it is not known whether theQM expressing chondrocytes actually participate in second-ary center formation. Further, the exact nature of the QMexpressing cells within the capillaries is not clear, becausethese cells were not stained with an antibody raised to vonWillebrand factor (not shown). In addition to the epiphysealregion, QM also was found to be expressed differentiallywithin the growth plate. There was no detectable expressionof QM in the proliferative chondrocytes (consistent with itsidentification by subtractive hybridization), the maturingchondrocytes, or the hypertrophic chondrocytes within thecentral region of the growth plate, whereas the hypertrophicchondrocytes adjoining the perichondrium clearly expressedQM. The hypertrophic chondrocytes located at the lateraledges of the growth plates have been previously termed“borderline chondrocytes”(34) to reflect their physical loca-tion and their potentially ambiguous differentiation status. Ithas been proposed that these lateral hypertrophic chondro-cytes are in a different endocrine/paracrine environment asopposed to the more centrally located hypertrophic chon-drocytes and that they have the potential to differentiate toosteoblast-like cells, which contribute to the formation ofthe periosteum. The expression of QM by these borderlinechondrocytes clearly distinguishes them from the more cen-trally located hypertrophic chondrocytes, but to date, thereis still no clear in vivo evidence that transdifferentiation ofthese chondrocytes to osteoblasts occurs within the mam-malian growth plate. Our finding that QM is expressed onlyby borderline chondrocytes within the growth plate con-trasts with the findings in a recent report concluding thatQM is expressed by transition zone chondrocytes and not byhypertrophic or proliferative chondrocytes in mouse em-

FIG. 5. Expression of QM by MC3T3-E1 cells and peri-cytes in culture. (A) Total RNA was prepared from culturedmineralized (M) and confluent (C) MC3T3-E1 cells (i). ThisRNA was blotted and probed with QM (ii). (B) Protein fromMC3T3-E1 cells were separated by 10% SDS-PAGE andstained with Coomassie blue (i). A duplicates of the gel ini was western blotted with the anti-QM antibody (ii). Therelative position of the 22-kDa and 30-kDa standards areindicated. (C) Poly(A) cDNA products were generated frompericyte cultures that were confluent(1) and from pickedmineralized nodules(2) and electrophoresed on a 2% agarosegel and stained with ethidium bromide (i). The gel in i wasSouthern blotted and probed with radiolabeled QM (ii), withMGP (iii), and with ribosomal protein S11 (iv).

1072 GREEN ET AL.

bryos at E 14.5.(35) The reason for this apparent discrepancymay be that QM is only expressed by transition zone chon-drocytes in the earliest stages of growth plate formation butthat once the growth plate is established, the expression ofQM by these chondrocytes is down-regulated. Alterna-tively, it may be that the QM expressing cells within thetransition zone in the developing growth plate of E 14.5mice represent the centrally located early hypertrophicchondrocytes that are known to express “osteoblastic”traits.(34) This latter explanation is in keeping with ourfinding that QM is expressed by borderline hypertrophicchondrocytes but not by chondrocytes within the transitionzone once the growth plate has formed.

The most striking expression of QM was observed byosteoblasts on the diaphyseal margin of the growth plate,within the secondary centers of ossification and in theperichondrium surrounding the growth plate. This highlevel of expression of QM progressively diminished deeperin the diaphyseal region and expression of QM was notdetected in osteocytes. A similar reduction of expression ofQM occurred within the periosteum further from the growthplate with no apparent expression in the adjoining fibroustissue or in the epiphyseal perichondrium.

The distinctive pattern of expression of QM within thedeveloping bone suggests that the synthesis of this protein ismarkedly up-regulated by cells that are poised to produce amineralized matrix but once that mineralization process isestablished, down-regulation of QM occurs. To test thishypothesis in vitro, two mineralizing cell culture systemswere examined, the MC3T3-E1 osteoblast culture systemand a pericyte culture system. In the MC3T3-E1 culturesystem, cells have been shown to exhibit a developmentalsequence similar to osteoblasts in bone tissue.(27) In theMC3T3-E1 cultures, there was a decrease in the level ofexpression of QM in the mineralized cultures comparedwith the nonmineralized cultures. The pericyte system wasexamined as pericytes that are embedded within the base-ment membrane of vessels have been shown to have osteo-genic potential in vivo(24) and to form spontaneously nod-ules that undergo mineralization in vitro.(21) The formationof mineralized nodules by pericytes in vitro is associatedwith the stage-specific expression of markers of the miner-alization process such as alkaline phosphatase, osteocalcin,osteonectin, osteopontin, bone sialoprotein, and MGP.(33)

We found that QM was expressed by pericytes in confluentcultures but that this expression was down-regulated bypericytes within picked mineralized nodules. In contrast, ashas been previously described, MGP was not expressed inthe confluent cultures but was expressed by pericytes withinpicked mineralized nodules.(33) Further, S11 was expressedequally in both cell populations. Thus QM expression invivo and in vitro appears to be down-regulated duringmineralization and we hypothesize that this protein mayhave a role in the differentiation of cells before their min-eralization or in the regulation of the mineralization process.

The QM protein sequence is highly conserved amongprokaryotic and eukaryotic species.(36) QM is known to bepart of the 60S ribosomal subunit and studies in yeast haveshown that it is added to the 60S subunit after the coreribosomal proteins have been assembled in the nucleus and

transported into the cytoplasm. QM is one of a class ofribosomal proteins that are able to cycle on and off theribosome in the cytoplasm.(37–40) As QM is a ribosomalprotein, its expected primary role would be to respond to thecell’s requirements for protein synthesis and its absence tohave a serious detrimental effect on cellular function. In-deed, theSaccharomyces cerevisiaeorthologue of QM,GRC5 (which has also been termed QSR1), when mutatedhas been shown to cause cell cycle arrest after one to threecell divisions and to prevent protein synthesis and mating ofthe arrested cells.(41) However, the finding that QM is dif-ferentially expressed during endochondral bone formationimplies that the regulation of the synthesis of this protein isnot just in response to a generalized increased need forprotein synthesis in tissues with a high proliferation activity.In particular, up-regulation of QM was not detected in theproliferative zone of the growth plate or in the majority ofchondrocytes within the epiphyseal cartilage where there isactive synthesis of a cartilaginous extracellular matrix. Fur-ther, the QM gene has previously been identified in otherdifferential screens and found to be expressed differentiallythroughout the mouse embryo.(35) The human QM cDNAwas found originally by subtractive hybridization: (i) to bedown-regulated in the development of Wilms’ tumor(28);(ii) to be down-regulated during the differentiation of mousepreadipocytes to adipocytes(42); and (iii) to be expresseddifferentially during heart development in the rat.(43) Al -though the function of QM in these processes is not known,Jif-1 (which is the chick orthologue of human QM) origi-nally was identified via its interaction with the transcriptionfactor Jun.(31) It has been shown that Jif-1 binds Jun directlyin a protein-to-protein zinc-dependent manner, which pre-vents the binding of Jun to target sequences.(44) It thereforehas been proposed that Jif-1 could alter the expression ofgenes that are controlled by Jun(31,44) but this assertionremains controversial.(35)

That some ribosomal proteins may have extraribosomalfunctions and that these functions may be involved in spe-cific differentiation processes recently has been convinc-ingly shown by the finding that mutations in the geneencoding the ribosomal protein S19 cause the human dis-order Diamond-Blackfan anemia (DBA).(45) In DBA, theclinical symptoms of the disease primarily are confined toerythropoiesis caused by the specific reduction of erythroidprecursor cells in the bone marrow, which point toward S19having a key regulatory role in this differentiation pathway.The finding that mutations in a ribosomal gene can cause aninherited human disorder coupled with the pattern of ex-pression of QM in bone development raises the possibilitythat QM could be a candidate gene for forms of osteochon-drodysplasia that map to the QM locus Xq28.(46)

ACKNOWLEDGMENTS

This work was funded by the Arthritis Research Cam-paign, UK, Faculty of Medicine Bequest Fund, Universityof Manchester, UK and The Wellcome Trust, UK.

1073QM EXPRESSION DURING MINERALIZATION

REFERENCES

1. Poole AR 1991 The growth plate : Cellular physiology, carti-lage assembly and mineralisation. In: Hall BK, Newman SA(eds.) Cartilage : Molecular Aspects. CRC Press, Boca Raton,FL, U.S.A., pp. 179–211.

2. Cancedda R, Cancedda FD, Castagnola P 1995 Chondrocytedifferentiation. Int Rev Cytol159:265–358.

3. Karsenty G 1998 Genetics of skeletogenesis. Dev Gen22:301–313.

4. Ducy P, Karsenty G 1998 Genetic control of cell differentia-tion in the skeleton. Curr Biol10:614–619.

5. Vu TH, Shipley LM, Bergers G, Berger JE, Helms JA, Hana-han D, Shapiro SD, Senior RM, Werb Z 1998 MMP-9/Gelatinase B is a key regulator of growth plate angiogenesisand apoptosis of hypertrophic chondrocytes. Cell93:411–422.

6. Wallis GA 1996 Bone growth: Coordinating chondrocyte dif-ferentiation. Curr Biol6:1577–1580.

7. Schinke T, McKee MD, Karsenty G 1999 Extracellular matrixcalcification: Where is the action? Nat Genet21:150–151.

8. Hillarby MC, King KE, Brady G, Grant ME, Wallis GA,Boot-Handford RP 1996 Localisation of gene expression dur-ing endochondral ossification. Ann N Y Acad Sci785:263–266.

9. Chapman KL 1998 Identification and analysis of the genesinvolved in the initiation of endochondral ossification. PhDthesis, University of Manchester, Manchester, U.K.

10. Alini M, Matsui Y, Dodge GR, Poole AR 1992 The extracel-lular matrix of cartilage in the growth plate before and duringcalcification: Changes in composition, and degradation of typeII collagen. Calcif Tissue Int50:327–335.

11. Lee ER, Matsui Y, Poole AR 1990 Immunochemical studies ofthe C-propeptide of type II procollagen in chondrocytes of thegrowth plate. J Histochem Cytochem38:659–673.

12. Chomczynski P, Sacchi N 1987 Single step method of RNAisolation by acid guanidinium thiocyanate-phenol-choloroformextraction. Anal Biochem162:156–159.

13. Culbert AA, Wallis GA, Kadler KE 1996 Tracing the pathwaybetween mutation and phenotype in osteogenesis imperfecta:Isolation of mineralisation-specific genes. Am J Med Genet63:167–74.

14. Brady G, Iscove NN 1993 Construction of cDNA librariesfrom single cells. Methods Enzymol225:611–623.

15. Brady G, Billia F, Knox J, Hoang T, Kirsch IR, Voura EB,Hawley RG, Cumming R, Buchwald M, Siminovitch K,Miyamoto N, Boehmelt G, Iscove NN 1995 Analysis of geneexpression in a complex differentiation hierarchy by globalamplification of cDNA from single cells. Curr Biol5:909–922.

16. Sambrook J, Fritsch EF, Maniatis T 1989 Molecular Cloning:A Laboratory Manual. Cold Spring Harbor, NY, U.S.A.

17. Laemmli UK 1970 Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature227:680.

18. Towbin HT, Staehelin T, Gordon J 1979 Electrophoretic trans-fer of proteins from polyacrylamide gels onto nitrocellulosesheets: Procedure and some applications. Proc Natl Acad Sci US A 76:4350–4354.

19. Schor AM, Allen TD, Canfield AE, Sloan P, Schor SL 1990Pericytes derived from the retinal microvasculature undergocalcification in vitro. J Cell Sci97:449–461.

20. Schor AM, Schor SL 1986 The isolation and culture of endo-thelial cells and pericytes from the bovine retinal microvascu-lature: A comparative study with large vessel vascular cells.Microvasc Res32:21–38.

21. Schor AM, Canfield AE 1998 Osteogenic potential of vascularpericytes. In: Beresford JN, Owen ME (eds.) Marrow StromalCell Culture. Cambridge University Press, Cambridge, U.K.,pp. 128–148.

22. Canfield AE, Sutton AB, Hoyland JA, Schor AM 1996 Asso-ciation of thrombospondin-1 with osteogenic differentiation ofretinal pericytes in vitro. J Cell Sci109:343–353.

23. Canfield AE, Allen TD, Grant ME, Schor AL, Schor AM 1990Modulation of extracellular matrix by bovine retinal pericytesin vitro: Effects of the substratum and cell density. J Cell Sci96:159–169.

24. Doherty MJ, Ashton BA, Walsh S, Beresford JN, Grant ME,Canfield AE 1998 Vascular pericytes express osteogenic po-tential in vitro and in vivo. J Bone Miner Res13:828–838.

25. Kodama H, Amagai Y, Sudo H, Kasai S, Yamamoto S 1981Establishment of a clonal osteogenic cell line from newbornmouse calvaria. Jpn J Oral Biol23:899–901.

26. Sudo H, Kodama H, Amagai Y, Yamamoto S, Kasai S 1983 Invitro differentiation and calcification in a new clonal osteo-genic cell line derived from newborn mouse calvaria. J CellBiol 96:191–198.

27. Quarles ID, Yohay DA, Lever LW, Caton R, Wenstrup RJ1991 Distinct proliferative and differentiated stages of murineMC3T3–E1 cells in culture—an invitro model of osteoblastdevelopment. J Bone Miner Res7:683–692.

28. Dowdy SF, Lai KM, Weissman BE, Matsui Y, Hogan BLM,Stanbridge EJ 1991 The isolation and characterisation of anovel cDNA demonstrating an altered mRNA level in nontu-morigenic Wilms’ microcell hybrid cells. Nucleic Acids Res19:5763–5769.

29. Corpet F 1988 Multiple sequence alignment with hierarchicalclustering. Nucleic Acids Res16:10881–10890.

30. Chan YL, Diaz JJ, Denoroy L, Madjar JJ, Wool IG 1996 Theprimary structure of rat ribosomal L10: Relationship to aJun-binding protein and to a putative Wilms’ tumour suppres-sor. Biochem Biophys Res Commun225:952–956.

31. Monteclaro FS, Vogt PK 1993 A jun-binding protein related toa putative tumour suppressor. Proc Natl Acad Sci U S A90:6726–6730.

32. Tron T, Yang M, Dick FA, Schmitt ME, Trumpower BL 1995QRS1, an essential yeast gene with a genetic relationship to asubunit of the mitochondrial cytochrome bc1 complex, ishomologous to a gene in eukaryotic cell differentiation. J BiolChem270:9961–9970.

33. Doherty MJ, Canfield AE 1999 Gene expression during vas-cular pericyte differentiation. Crit Rev Eukaryot Gene Expr9:1–17.

34. Bianco P, Cancedda FD, Riminucci M, Cancedda R 1998Bone formation via cartilage models: The “borderline”chon-drocyte. Matrix Biol17:185–192.

35. Mills AA, Mills MJ, Gardiner DM, Bryant SV, Stanbridge EJ1999 Analysis of the pattern of QM expression during mousedevelopment. Differentiation64:161–171.

36. Farmer AA, Loftus TM, Mills AA, Sato KY, Neil JD, Tron T,Yang M, Trumpower BL, Stanbridge EJ 1994 Extreme evo-lutionary conservation of QM, a novel c-Jun associated tran-scription factor. Hum Mol Genet3:723–728.

37. Zinker S, Warner JR 1976 The ribosomal proteins of Saccha-romyces cerevisiae. J Biol Chem251:1779–1807.

38. Dick FA, Karamanou S, Trumpower BL 1997 QSR1, anessential yeast gene with a genetic relationship to a subunit ofthe mitochondrial cytochrome bc1 complex, codes for a 60sribosomal subunit protein. J Biol Chem272:13372–13379.

39. Nguyen YH, Mills AA, Stanbridge EJ 1998 Assembly of theQM protein onto the 60s ribosomal subunit occurs in thecytoplasm. J Cell Biochem68:281–285.

40. Loftus TM, Nguyen YH, Stanbridge EJ 1997 The QM proteinassociates with ribosomes in the rough endoplasmic reticulum.Biochemistry36:8554–8230.

41. Koller H, Klade T, Ellinger A, Breitenbach M 1996 The yeastgrowth control gene GRC5 is highly homologous to the mam-malian putative tumour suppressor gene QM. Yeast12:53–65.

1074 GREEN ET AL.

42. Eisinger DP, Jiang HP, Serrero G 1993 A novel mouse genehighly conserved throughout evolution: Regulation in adipo-cyte differentiation and in tumorigenic cell lines. BiochemBiophys Res Commun196:1227–1232.

43. Kirby Ml, Cheng G, Stadt H, Hunter G 1995 Differentialexpression of the L10 ribosomal protein during heart develop-ment. Biochem Biophys Res Commun212:461–465.

44. Inada I, Mukai J, Matsushima S, Tanaka T 1997 QM is a novelzinc-binding transcription regulatory protein: Its binding toc-Jun is regulated by zinc ions and phosphorylation by proteinkinase C. Biochem Biophys Res Commun230:331–334.

45. Draptchinskaia N, Gustavsson P, Andersson B, Pettersson M,Willig T-N, Dianzani I, Ball S, Tchernia G, Klar J, Matsson H,Tentler D, Mohandas N, Carlsson, Dahl N 1999 The geneencoding ribosomal protein S19 is mutated in Diamond-Blackfan anaemia. Nat Genet21:169–175.

46. Van den Ouweland AMV, Kioschis P, Verdjik M, Tamanini F,Toniolo D, Poustka A, van Oost BA 1992 Identification andcharacterisation of a new gene in the human Xq28 region.Hum Mol Genet1:269–273.

Address reprint requests to:Dr. Gillian A. Wallis

2.205 Stopford BuildingOxford Road

University of ManchesterManchester M13 9PT, U.K.

Received in original form May 4, 1999; in revised form November30, 1999; accepted January 6, 2000.

1075QM EXPRESSION DURING MINERALIZATION