cDNA Cloning and Sequencing of Mouse Mastocytoma Glucosaminyl ...

Vol. 269,No. 14,Issue of April 8,PP. 10436-10443, 1994 Printed in U.S.A.

THEJOURNAL OF BIOIDGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

cDNA Cloning and Sequencingof Mouse Mastocytoma Glucosaminyl N-DeacetylaselN-Sulfotransferase,an Enzyme Involvedin the Biosynthesis of Heparin* (Received for publication, November 8, 1993, and in revised form, December 30, 1993)

Inger Eriksson,Dagmar SandbackS, Bo Ekl, U l f LindahlS, and Lenayjellhn From the Department of Veterinary Medical Chemistryand the $Department of Cell Research, Swedish University of Agricultural Sciences, Box 7055, S-750Uppsala, 07 Sweden and the Wepartment of Medical and Physiological Chemistry, University of Uppsala, The Biomedical Centel; Box 575,S-75123Uppsala, Sweden

A 110-kDa protein involved in heparin biosynthesis in The subsequentmodifications, involving 0-sulfation at various mouse mastocytoma cells was previously shown to ex- positions and C5-epimerization of D-glucuronic acid to L-iduronic acid, all occur in thevicinity of N-sulfate groups. Hence, pressbothglucosaminylN-deacetylaseandN-sulfotransferase activity. In this study, the complete nucle- N-deacetylationlN-sulfationhas a key role in determining the to this protein is extent of modification of the polysaccharide chain (Lindahl, otide sequence corresponding reported. ThemF?NA,estimated to contain3.9 kilobases 1989). Heparin, produced by connective tissue-type mast cells, M rof 101,092. The predicted is extensively N-sulfated (generally >80% of the GlcN units), encodes a protein with an domain structure of the protein resembles those of pre-whereasheparansulfates, derivedfrom virtually all other viously characterized Golgi proteins with an N-terminal types of cells, usually show approximately equal amounts of cytoplasmic tail, a single membrane-spanning domain, N-acetylated andN-sulfated GlcN residues. Accordingly, hepaand a large catalytic domain linked to the transmemrin contains more L-iduronic acid and 0-sulfate than heparan brane domain through a “stem region.” Comparison of sulfate, as predicted from the substratespecificities of the corthe deduced amino acid sequence of the mouse masto- responding enzymes (Lindahl, 1989). However, why is cytoma protein anda previously cloned similar enzymeN-deacetylationlN-sulfationmore extensive in the mast cells from ratliver demonstrated that while large portions than of in othercells? the proteins, corresponding essentially to the putative Proteins withglucosaminyl N-sulfotransferase activity have catalytic domains, were closely related, other portions, been purifiedfrom (heparan sulfate-producing) rat liver (Branin particular in the N-terminal parts, were markedly different. The divergence was not due to species differ- danand Hirschberg,1988) as well as (heparin-producing) mouse mastocytoma (Petterssonetal., 1991). The cDNAences since two separate mouse transcripts could be identified that hybridized with probes specific for the sequence and thededuced amino acid sequenceof the rat liver protein havebeen published (Hashimotoet al., 1992). The mastwo proteins. Also, functional differences were noted protein was shown to catalyze also N-deacetylation, since the mastocytoma enzyme, contrary to the livertocytoma enbut only in the presence of a n unidentified protein cofactor zyme, requires a polycation cofactor for expression of 1991). While these resultssuggested that the N-deacetylase activity.The results are discussed in re- (Pettersson et al., lation to the structural properties of heparin and hepa- two enzyme activities reside in the sameprotein, they did not exclude that the unidentified “cofactor” would actually harbor ran sulfate. the active site for N-deacetylation (Pettersson et al., 1991). Recently, Wei et al. (1993) demonstrated thata purified soluble The biosynthesis of heparin and heparan sulfate is initiatedfusion protein containing the Golgi lumenal portion of the rat by glycosylation reactions that generate saccharide sequences liver N-sulfotransferase expressed both N-deacetylase and Nsulfotransferase activity (see also Ishihara et ul. (1993)), thus composed of alternating D-glucuronic and N-acetylglucosamine units (Lidholt and Lindahl, 1992). The resulting (GlcAp1,4- suggesting that this proteincontained both active sites. This report describes the cDNA sequence and the deduced GlcNAcal,4-),’ disaccharide repeats are modified through a series of reactions that is initiated by N-deacetylation of N- amino acid sequence for the mouse mastocytoma protein. Remastothe acetylglucosamine residues. The generated free amino groups sults are presented that indicate that the liver and N-deacetylaselN-sulfotransferases are closely related cytoma are then sulfated through the action of a n N-sulfotransferase. but distinct proteins. MATERIALS AND METHODS * This work was supported by Grants 2309 and 6525 from the Swedish Medical Research Council,Grant BMH1-CT92-1766 from the EuroPeptide Purification and Sequencing-The 110-kDa protein (-20 pg), pean Economic Community, and grants from Konung Gustaf Vs 80- purified from a detergent extract of mouse mastocytoma by affinity Arsfond; the Faculty of Veterinary Medicine, Swedish University of chromatography on wheat germ agglutinin-Sepharose, blue Sepharose, Agricultural Sciences; Italfarmaco S.p.A., Milan, Italy; and Polysacka- and 3’,5‘-ADP-agaroseas previously described(Petterssonet al.,1991) ridforskning AB, Uppsala, Sweden. The costs of publication of this was cleaved with a lysine-specific protease from Acromobacterlyticus article were defrayed in part by the payment of page charges. This (Waco) in the presence of 2 M guanidine HCI. The generated peptides article must therefore be hereby marked ”advertisement” in accordance were separated on a reverse phase C4 column (Brown-Lee) eluted at a with 18 U.S.C.Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted flow rate of 100 pVmin with a 6-ml2-60%acetonitrile gradient in 0.1% detected with a 990 Waters diode-array detector. to the GenBankTMIEMBL Data Bank with accession number(s) X75885. trifluoroacetic acid and ll To whom correspondence should be addressed. Tel.: 46-18-174217; Selected peptides were then analyzed with a model 470A protein sequenator (Applied Biosystems) equipped with an on-line 120 phenylFax: 46-18-550762. thiohydantoin Analyzer. The abbreviations used are: GlcA, D-glucuronicacid;MES,2-(NProbes for Screening-Single-stranded cDNA was synthesized using morpho1ino)ethanesulfonic acid; kb, kilobaseb); PCR, polymerase chain 1 pg of total mouse mastocytoma RNA in a 20-pl reaction mixture reaction; bp, base pairb); Mops, 4-morpholinepropanesulfonicacid.

10438

N-Deacetylase IN-Sulfotransferase cDNA Sequence

10439

TABLEI containing 10 n" Tris-HC1, pH 8.3, 50 m~ KCl, 5 n" MgCh, 1 n" of Peptide and primer sequences each deoxynucleotide, 1unit of RNase inhibitor (Perkin-Elmer Corp.), 2.5 p~ oligo(dT), and 1.25 units of murine leukemia virus reverse tranA. Peptide scriptase (Perkin-Elmer Corp.). The reaction mixture was incubated for 42 "C for 15 min, a t 99 "C for 5 min, and finally a t 5 "C for 5 min. The 1s 1A 1. KFYHTGTEEEDAGDDM sense and antisensedegenerate oligonucleotides 1sand lA correspond2. KYFELFPQERSPL ing to peptide 1(see Table I) were synthesized using a Pharmacia LKB 3. KIIXVLINPADRAY Gene Assembler Plus. A 100-pl PCR mixture containing 10 pl of the 4. KGFWCQGLEGG single-stranded cDNA, 100 pmol of the oligonucleotides 1s and lA, 10 n" Tris-HC1,pH8.3, 50 m~ KCl,2.5 m~ MgCl,, 1 m~ of each deB. Rimer" Degeneracy oxynucleotide, and 1.25 units of Taq polymerase was incubated for 35 1s. 5"AARTTYTAYCAxACNGGNAC-3' x256 cycles in a thermal cycler (Gene ATAQ controller; Kabi-Pharmacia AB). Each cycle included denaturation at 95 "C for 1min,annealing at 46 "C 1A. St-CATRTCRTz?CCNGCRTc-3' x128 for 30 s, and extension at 70 "C for 1 min. The reaction products were "R,AorGx,TorC;N,AorGorTorC. separated on 12% polyacrylamide gels and a 48-bp band was cut out from the gel. After incubation of the gel slice in sterile H,O a t 4 "C RNAPurification--TotalRNA was isolated frommousemastocyovernight, the eluted 48-bp product was reampliiied using the same conditions. After an additional polyacrylamide gel electrophoresis, the toma, rat liver, and mouse liver by the LiCVuredSDS method (Sam48-bp product was isolated and 5'-end 32P-labeledusing [Y-~~PIATPbrook et al., 1989). Polyadenylated RNA was purified using the polyATtract kit from Promega. (DuPont NEN) and polynucleotide kinase (New England Biolabs). PCR of Rat Liver cDNA-A sense and an antisense primer correInsert DNAof the 1.7-kb clone (see below) was 32P-labeled using [GI-~'P]~CTP (DuPont NEN) in the RPN.1601Y multiprime DNA label- sponding to bp 3195-3216 and 3682-3661, respectively, in the rat liver cDNAsequence(Hashimotoet al., 1992) weresynthesized. -To the 5'-end ing system, obtained from Amersham International. cDNA Libraries-Library 1 is a previously described A gtll mouse of the primers was added a dCCGAAlTC (where d is deoxy) extension to create an EcoRI site in the PCR product. For synthesis of singlemastocytoma cDNA library (&ell& et al., 1989) comprising 0.7 x 10' independent recombinants. Library 2 was constructed using mouse stranded rat liver cDNA, 300 ng of total rat liver RNA in 10 n" Trismastocytoma polyadenylated RNA as template for cDNA synthesis in HCl, pH 8.3,90 m~ KC1 was incubated at 70 "C for 5 min together with the cDNA synthesis kit from Pharmacia LKB Biotechnology Inc. The 0.5 p~ "downstream" antisense primer. After addition ofMnC1, (final , 4 deoxynucleotides (final concentration 0.2 resulting cDNA was ligated to EcoRI-digested alkaline phosphatase- concentration 1 m ~ ) the treated bacteriophage A g t l l DNA obtained from Pharmacia LKB Bio- m)and 5 units of thermostable rTth reverse transcriptaselDNA potechnology Inc.Bacteriophage DNA was packagedusing the Packagene lymerase (Perkin-Elmer Corp.), the 2 0 4 reaction mixture was incubated at 60 "C for 15 min. The cDNA was amplified using 35 cycles of system from Promega. Library 2 contains 4 x lo6 independent clones. Screening of the cDNA Libraries-Nitrocellulose replicas of plaques incubation aRer addition of 80 pl of 5%glycerol, 10n"Tris-HC1, pH8.3, from the bacteriophage A g t l l mouse mastocytomacDNAlibraries were 0.1 M KCl, 1.5 m~ MgCl,, 0.75 m~ EGTA, 0.05% Tween 20, and 0.1 p~ prehybridized in solution containing 6 x SSC (1x SSC is 0.15 M NaCl, "upstream" sense primer. Each cycle included denaturation at 95 "C for 1 min, annealing at 64 "C for 1min, and extension at 72 "C for 1 min. 0.015 M sodium citrate buffer, pH 7.0), 3 x Denhardt's solution, 2 m~ EDTA, 0.5% SDS, and 0.1 mg of denatured salmon D N N d for 3 h at After a final 10-min incubation at 72 "C, the 504-bp product was recov65 "C before hybridization for 12-15 h in the same buffer containing ered by 2% agarose electrophoresis,cleaved with EcoRI, and subcloned 32P-labeled probe. The filters were washed in 2 x SSC, 0.5% SDS at into pUC 119 plasmid for nucleotide sequence analysis. Northern Blot Hybridization-Polyadenylated RNA from mouse mas42 "C and subsequently at 65 "C. Whenthe 1.7-kbfragment was used as a probe, the filters were finally washed with 0.2 x SSC, 0.5% SDS at tocytoma, mouse liver, and rat liver was denatured in 20 n" Mops buffer, pH 7.0, containing 50%(v/v) formamideand 2.2 M formaldehyde 65 "C. Subcloning and Sequencingof cDNA Inserts-DNA inserts, isolated at 70 "C for 3 min before being loaded on 1.2% (vh) formaldehydeby preparative agarose gel electrophoresis (Sambrook et al., 1989)after containing agarose gels. ARer electrophoresis,the gel was soakedin 50 m~ NaOH for 30min and rinsed with water and 20 x SSPE ( 1x SSPE EcoRI restriction cleavage of recombinant bacteriophage DNA,were subcloned into pUC 119 plasmid. The complete nucleotide sequence was is 0.15 M NaCl, 10 n" NaH,PO,~H,O, 1 m~ EDTA,pH 7.4) before determined independently onboth strands using the dideoxy chain transfer to the nylon membrane, Hybond N' (AmershamCorp.),in 20 x SSPE. The membranes were hybridized at 50 "C in a solution containtermination reaction with [ G I - ~ S I ~ Aand T P the modified T7 DNA polymerase (Sequenase, U. S. Biochemical Corp.) (Taborand Richardson, ing 5 x SSPE, 5 x Denhardt's solution, 0.5% SDS, 100pg of denatured 1987). The sequencing primers used were universal oligonucleotide salmon DNA/ml and washed in 2 x SSC, 0.5% SDS at 50 "C for 2 x 20 primers, which bind close to the polylinker ofpUC119, and specific min then at 65 "C for 2 x 20 min. The probes used were: (a) the 1.7-kb clone,(b) a 356-bp cDNAfragprimers synthesized on the Gene AssemblerPlus. Generation of a Fusion Protein and Peptide Conjugates for ment from the 3'-end of the mouse mastocytomacDNA (mouse mastoImmunization-The insert from the 1.7-kb cDNA clone was ligated to cytoma probe), generated from pUC 19 containing the 5B clone a h r the bacterial expression vector pGEX-3X (Pharmacia LKB Biotechnol- cleavage with EcoRI (cleavagesite in the polylinker)and BamHI (cleavage site a t bp 2941,see Fig. 3), and ( c )a 284-bpcDNAfragmentfrom the ogy), and transformed into JM 83bacteria. The resulting fusion protein contains the C terminus of a 26-kDa glutathione S-transferase encoded 3'-noncoding part of the rat liver cDNA (rat liver probe) corresponding by the parasitic helminth Schistosomajaponicum. The fusion protein is to bp 3396-3679 (see Hashimoto et al. (1992)).This fragment was genunder control of the tac promoter, which is inducible by isopropyl-l- erated by cleavage with Hinff of the 504-bprat liver cDNAPCR product obtained as describedabove. The three cDNA fragments were32Pthio-0-D-galactopyranoside. After induction, the bacteria were lysed, and the solubilized proteins were purified by affinity chromatography labeled with the multiprime DNA labeling system described above. on a glutathione-agarose column (sulphur linkage, Sigma). The fusion N-Deacetylase Assay-Purified mastocytomaN-deacetylase(110-kDa "C with 10,000 cpm of substrate protein was eluted by competition with 5 n"free glutathione (Sigma)in protein) was incubated at 37 50 m~ Tris-HCI, pH 8.0 (Smith and Johnson, 1988).Synthetic peptides ([3Hlacetyl-labeledEscherichia coli K5 capsular polysaccharide) in a corresponding to peptides 1 and 2 (see Table I) were conjugated to total volume of 200 pl of 50 m~ MES, pH 6.3, 10 m~ MnCI,, 1%Triton bovine serum albumin with glutaraldehyde (Harlow and Lane, 1988). X-100. Polybrene (Janssen Chimica, Belgium;synthetic polycation)was The fusion protein and the two peptide conjugates were homogenizedin added as indicated. After 1h, reactions were terminated by addition of 200 plof 1M monochloroacetic acid,0.5 M NaOH, 2 M NaCl. The released Freund's complete adjuvant before injection into rabbits. SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting[3Hlacetatewas determined by scintillation counting in a biphasic sysPurified 110-kDa protein and a crude mastocytoma microsomalfraction tem (for a more detailed description see Naviaet al. (1983)and Petterswere subjected to polyacrylamide gelelectrophoresisin SDS on 10-15% son et al. (1991)). gradient gels according to the method of Blobel and Dobberstein (1975). Separated proteins were transferred to nitrocellulose filters using semiRESULTS dry electrophoretic transfer according to Kyhse-Andersen (1984). The Generation of a Probe and Screening of Library 1"TO obtain filters were then blocked in 5%nonfat dry milk and incubated with 1/50 amino acid sequence data for the 110-kDa protein, highly pudilutions of the different sera. After washing, the specifically bound antibodies were allowed to react with 1261-labeledProtein A. The anti- rified enzyme was digested with lysine-specific protease, and gens recognized were detected by autoradiography. the generated peptides were separated on a reverse phase col"

"

N-DeacetylaselN-SulfotransferasecDNA Sequence umn. Sequence was obtained from four peptides (Table I). A cDNA probe was produced by PCR using mouse mastocytoma cDNA and degenerated sense and antisense primers based on the sequence of peptide 1(see Table I). The 48-bp product was purified by polyacrylamide gelelectrophoresis, reamplified using the same primers, and finally isolated after polyacrylamide gel electrophoresis. The 48-bp PCR product was labeled with 32Pand used for screening of a mouse mastocytoma A g t l l library (library 1).One hybridizing clone, containing a 1.7-kb insert was identified. Nucleotide sequence analysis indicated that the insert corresponded to a region entirely within the coding part of the transcript. Peptides 1,2, and 3 (see Table I) were all identified in the deduced amino acid sequence. The 1.7-kb fragment was subcloned into the procaryotic expression vector PGEX-~X,and the resulting glutathione S-transferase fusion protein was used to produce a rabbit polyclonal antiserum. Synthetic peptides with the structures of peptides 1 and 2 were also used for ~ m u n i z a t i o nof rabbits after conjugation to bovine serum albumin. All three antisera recognized the 110-kDa protein in immunoblotting (data not shown), demonstrating that the 1.7-kb cDNA is derived from the transcript encoding the 110-kDa protein. Characterization of cDNA Corresponding to the Entire Coding Region-Screening of a larger mouse mastocytoma A g t l l library (library 2) using the 1.7-kb fragment as a probe identified 13 additional clones. Based on restriction patterns using ApaI and PssI alone or in combination, two clones, 6B (1.3kb) and 5B (1.75 kb), extending from the known sequence into the 5' and 3' directions, respectively, were selected for subcloning and sequence determination. The cDNA sequence obtained comprises 3306 bp (Fig. 1).The first ATG, found a t position 158, is preceded by a guanosine at position -3 and thus conforms to the consensus initiation sequence (Kozak, 1989).The coding region contains 2646 bp encoding a protein of 882 amino acids. Northern blot analysis of mastocytoma mRNA revealed a strongly predominant -3.9-kb transcript (Fig. 2). Predicted Protein Structure-The mouse mastocytoma cDNA encodes an 882 amino acid protein with a predicted molecular mass of 101,092 daltons. The deduced primary structure contains regions that correspond to the four sequenced peptides (see Table I, Fig. 1). The predicted protein has typical features of a resident Golgi protein (see Machamer (1991)):a short Nterminal cytoplasmic tail (18 amino acids) followed by a membrane-spanning region (24 amino acids) and a large luminal domain. Seven potential N-glycosylation sites are present. At least some of these are likely to be glycosylated, since the mastocytoma enzymebinds to wheat germ agglutinin (Pettersson et al., 1991).In addition, the purified enzyme showedan apparent M,of 110,000 on SDS-polyacrylamide gelelectrophoresis (Pettersson et al., 1991), -9 kDa larger than the estimated mass (see above), hence allowing the for presence of oligosaccharides. Comparison with the Rat Liver N-SulfotransferaseRecently, the cDNA sequence of a rat liver N-sulfotransferase was published (Hashimoto et at., 1992). Comparisonof the predicted amino acid sequences of this protein and the mouse mastocytoma enzyme reveals that thetwo proteins are closely related with 71% amino acid sequence identity (72% at the nucleotide level) (Fig. 3). However, the N-terminal regions (amino acids 1-81) are quite different (only 30%identity at the amino acid level and 46% at thenucleotide level).In addition,

FIG.1. Nucleotidesequenceandpredicted amino acid mequence of the 110-kDamouse mastocytoma protein. The numbers on the right and the left indicate the nucleotide residue and the amino

acid residue in the respective sequence. The four sequenced peptides are underlined. Amino acids in the potential membrane-spanning domain are shown in italics. The asparagine residues that may be glycosylated are marked by asterisks. The two potential polyadenylation signals in the 3'-noncoding region are underlined.

N-Deacetylase I N-Sulfotransferase cDNA Sequence

kb

mouse rat

10441

1 MLQLWKVVRPARQLELHRLILLLIGFSLVSMGFLAYYVSTSPKAKEPLPL 50 1 PA ACLR LC H SPQAVLF FV C F VFVS LYGWNRGL .... 46 51 PLGDCSSSGAAGPGPARPP..VPPRPQRPPETTRTEPWLVFVESAYSQL 47 SA A E DCGD P VA SRLL IK VQAVAPS D LL

9.46.2-

98 96

99 G Q E I V A I L E S S R F R Y S T E L V F G R G D M P T L T D H T H G R W L V I 148 146 97 v I I FAKGR K RK IA A

K

3.92.8-

149 LDAWSRELLDRYCVEYGVGIIGFFRAREHSLLSAQLKGFPLFLHSNLGLR 198 147 N K N N K 196 199 DYQVNPSAPLLHLTRPSRLEFGPLFGDDWTIFQSNHSTYEPVLIASHRPA 248 197 CSI KS W L V EV EVK KT S S 246

....GPVLRRARLPTWQDLGLHDGIQRVLFGHGLSFWLHKLVFV

249 ELSMP 247 NNLN LHA HLGADAG HAI

.

294 295

295 DAVAYLTKRLCLDLDRYILVDIDDIFVGKEGTRMKVADVEALLTTQNKL 344 296 F S P E K F D E 345

1.8-

345 RTLVPNFTFNLGFSGKFYHTTEEEDAGDDMLLKHRREFWWFPHMWSHMQ 394 395 346 L HIDA F Y SWK 395 P.LFHNRSVLADQMRLNKQFALEHGIPTDLGYAVAPHHSGWP1HSQLYE 443 445 396H Q EK AV M v v

0.9K

444 AWKSVWGIQVTSTEEYPHLRPARYRRGFIHNGIMVLPRQTCGLFTHTIFY 493 495 R 446N Q 494 NEYPGGSRELDRSIRGGELFLTVLLNPISVFMTHLSNYGNDRLGLYTFES 543 496 KI N KH 545 544 546

LVRFLQCWTRLRLQTLPPVPLAQKYFELFPQERSPLWQNPCDDKRHKDIW HS Q

N

QI SE KD

D

E

593 595

594 SKEKTCDRLPKFLIVGPQKTTTAIHFFLSLHPAVTSSFPSPSTFEEIQF 643 645 596 F L I DLSGMLYL NY SE

FIG.2. Size of the mouse mastocytoma transcript. Polyadenylated RNA from the mouse mastocytoma was separated by agarose electrophoresis and hybridized with 32P-labeled 1.7-kb fragment a s described under “Materials and Methods.” Sizes of RNA markers (Life Technologies, Inc.) are indicated in kilobases.

the 5’- and 3’-untranslated regions seem unrelated (both 37% identity). To investigate whether these discrepancies could be due to species difference, probes were produced that would hybridize with the3’-ends of the rat liver and themouse mastocytoma transcripts, respectively. Northern blotting of rat liver, mouse liver, and mouse mastocytoma mRNA, demonstrated that the mouse mastocytomaprobe hybridized with the -3.9-kb band in mouse mastocytoma mRNA as expected (Fig. 4, compare with Fig. 3). Small amounts of this transcript, detected with the mouse mastocytoma probe after prolonged exposures, also occurred in rat liver and mouse liver mRNA (not evident in Fig. 4A). In contrast, therat liver cDNA probe recognized a n 8-kb transcript inboth rat and mouse liver mRNA, while no such band was apparent in the mouse mastocytoma mRNA preparation (Fig. 4). Thus, different transcripts appear to be utilized for the synthesis of the two proteins. Effect of Polybrene on N-Deacetylase Activity-Purification of the mastocytoma N-deacetylaselN-sulfotransferaseyielded a 110-kDa protein that expressed N-sulfotransferase activity, but lackedN-deacetylaseactivity unlesssupplementedwith a crude protein fraction that was separated from the 110-kDa protein by chromatography on immobilized wheat germagglutinin (Pettersson etal., 1991). Thus itcould not be established whether theactive site for N-deacetylation was actuallylocated in the 110-kDa component or in one of the added proteins. It was accidentally found that basic proteins, such as histones (data not shown) or, indeed, a synthetic polycation, Polybrene, could substitute for the crudemastocytoma protein fraction as a N-deacetylase “cofactor”(Fig. 5; see alsoKjellen et al. (1992)). While no N-deacetylation occurred in theabsence of Polybrene, enzyme activity was readily detected in the presence of this compound, with maximal effect at -25 pg/ml. At higher Polybrene concentrations thestimulatory effect decreased. The mechanism behind the stimulation remainsunclear. However, the findings clearly demonstrate that the catalytic sites for N-deacetylation and N-sulfation both reside in the 110-kDa

644 FNGPNYHKGIDWYMDFFPVPSNASTDFLFEKSATYFDSEWPRRGAALLP 693 695 646 H E I W S Y N A A 694 RAKIITVLINPADRAYSWYQHQRAHGDPIALNYTFYQVISASSQAPLLLR 743 745 696 K VL I DK V HE T GPD SSK 744 SLQNRCLVFGYYSTHLQRWLTYYPSGQLLIMDGQELRVNPAASMEIIQKF 793 795 746 WYA A IE SAFHAN VL IKL T E KV DTV 794 LGITPFLNYTRTLRFDEDKGFWCQGLEGGKTRCLGRSKGRRYPDMDMESR 843 845 796 V STVD AHK PK K K L K E L D 844 LFLTDFFRNHNLELSKLLSRLGQPAPLWLREELQHSSVG 846 AK YY D IYKM TL T D NTR..

882 882

FIG.3. Comparison of the deduced amino acid sequences of the mouse mastocytoma and rat liver enzymes. Only the amino acid residues in the rat liver N-sulfotransferase (Hashimoto et al., 1991) that differ from those in the mouse enzyme are shown. The numbers indicate the amino acid residues of the proteins. Gaps (.) inserted to optimize the alignment of the sequences were identified by the command “BESTFIT”of the GCG program (Devrew et al., 1984).

protein. Similar conclusions have been put forward with regard to the rat liver enzyme (Wei et al., 1993). By contrast, the N-deacetylase activity of the latter enzyme did not depend on the presence of additional polycation (in support of the notion that themastocytoma and liver enzymesare not only structurally but alsofunctionally different). DISCUSSION

Current notions on the structural organization of Golgi proteins derive largely from studies on glycosyltransferases (Joziasse, 1992; Shaper andShaper, 1992). These enzymes allshow the orientationof type I1 membrane-bound proteins (see Wickner andLodish (1985)), with the N-terminal regions projecting into the cytoplasm. The short cytoplasmic tail is positively charged and is followed by a single membrane-spanning domain. The large catalytic domain, which includes the C terminus, is linked to the transmembranedomain through a peptide segment (“stem region”) composed of 40-60 amino acid residues. The stemregion is flexible and largely devoid of secondary structure. It contains a high proportion of proline and glycine residues and often consensus sites for N-linked carbohydrate side chains. Recent studies on the targeting of Golgi proteins (reviewed in Machamer (1991) and in Shaper and Shaper (1992)) suggestedthat, while sequences located within the transmembrane domains appear to be essential for Golgi retention, thecytoplasmic tail and the stem region may contain

N-Deacetylase IN-Sulfotransferase cDNA Sequence

10442

a b c d e f

first 40 amino acids C-terminal to the transmembrane domain), whereas the glycine content is moderate (10%). No potential N-glycosylation site is seen within this region. A comparison of the deduced amino acid sequence of the mouse mastocytoma N-deacetylaselN-sulfotransferase with that of the corresponding rat liver enzyme reveals close homology between large portions of the two proteins (Fig. 3). However, the N-terminal regions comprised by the 81 first amino acid residues (corresponding to the cytoplasmic tail), the transmembrane domain, and 39 amino acid residues of the stem region appear unrelated.Analogous variants, with closely similar catalytic domains but discrepant N-terminal regions, have been noted also-for certain glycosyltransferases (see Joziasse (1992)). In the case of a murine al,3-galactosyl transferase gene, it has been determined that the transitionfrom homologous to unrelated regions coincides with theboundary between et al., 1992). It was suggested that a portion two exons (Joziasse of a n ancestral gene, encoding much of the catalytic domain of the enzyme, had been reutilized in a number of genes (Joziasse, 1992). The relationship at the gene level between the mouse mastocytoma andthe rat liver N-deacetylaselN-sulfotransferases remains to be elucidated. However, Northern blots using probes based on selected divergent nucleotide sequences indicated that the two enzymes are encoded by transcripts of markedly different size (Fig. 4). Moreover, both of these transcripts could be demonstrated inmouse tissues, thusexcluding simple species difference. The two transcripts may represent the products of two different genes or differently spliced variFIG.4. Northern blot analysis with probes specific for the mousemastocytomaandratlivertranscripts. Polyadenylated ants originating from the same gene. However, since differRNA from rat liver ( a and d ) , mouse liver ( b and e ) , and mouse masto- ences were found at two distantly located parts of the trancytoma tissue (c and f ) was separated by agarose electrophoresis and scripts, the latterpossibility is less likely. hvbridized with 32P-labeled mouse mastocvtoma urobe (a-c:)and rat The structuraldivergence between the two N-deacetylaselNl k e r probe (d-f), recognizing regions in thk 3’-noncoding parts cIf t he transcripts (see ‘Materials andMethods”). Sizes of RNA markers (Li fe sulfotransferases primarily involves regions of the proteins thathave beenimplicated inintracellulartargeting(see Technologies, Inc.) are indicated in kilobases. above), and it is, therefore, conceivable that the enzymes are differently located within the Golgi network. However, the present work also revealed a more clearcut, functional difference. The mastocytoma enzyme expresses N-deacetylaseactivity, but only in thepresence of a macromolecular cofactor (presumably a protein in the intactcell (Pettersson et al., 1991) that can be replaced by a synthetic polycation (Polybrene) in a purified system (Fig. 5; see also Kjell6n et al.(1992)). Experiments along the same line demonstrated, on the other hand, that the recombinant rat liver protein does not require any such cofactor in order to catalyze N-deacetylation and that the activity is insensitive to theaddition of Polybrene (Wei et al., 1993). It is recalled that the mast cell N-deacetylaselN-sulfotransferase initiates the seriesof polymer modification reactions that ultimately yields heparin, a heavily N - and 0-sulfatedpolysaccharide with few residual, usually isolated, N-acetyl groups. The 50 100 154 liver enzyme (and, presumably, analogous enzymes in most other cell types) is involved in the biosynthesis of heparan Polybrene concentration lg/ml sulfate, which generally contains extended regions of consecuFIG.5. N-Deacetylase activityas a function of Polybrene con- tive, N-acetylated disaccharide units. It is tempting to specucentration. Purified mouse mastocytoma 110-kDa protein (-40 ng) was assayed for N-deacetylase activity as described under “Materials late that the demonstrateddistinctive features of the isolated cell and Methods” in thepresence of increasing concentrationsof Polybrene. enzymes entail different functional propertiesin the intact The given values represent the mean of two determinations. that are ultimately reflected in the structural characteristics of heparin versus heparan sulfate. information of importance in this regard. The deduced amino acid sequence of the mouse mastocytoma REFERENCES N-deacetylaselN-sulfotransferaseimplies a proteindomain Blobel, G. & Dobberstein, B. (1975) J. Cell Biol. 67,835-851 structure similar to that ascribed to the glycosyltransferases. Brandan, E.& Hirschberg, C. B. (1988) J. Biol. Chem. 263,2417-2422 The putative transmembranedomain of 24 amino acid residues Devreux, J., Haeberli, P. & Smithies, 0. (1984)Nucleic ACI& Res. 12, 387395 Harlow, E.& Lane, D. ( 1988) in Antibodies: A Labomtory Manual, pp. 78-79, Cold (Fig. 1)is preceded by a n N-terminal region of 18 residues that Spring Harbor Laboratory, Cold Spring Harbor, NY contain 4 positively charged (1lysine and 3 arginine) amino Hashimoto, Y.,Orellana, A., Gil, G. & Hirschberg, C. B. (1992) J. B i d . Chem. 267, 15744-15750 acids and 1negatively charged (glutamic acid). The postulated Ishihara, M.,Guo, Y.,Wei, Z., Yang, 2..Swiedler, S. J., Orellana, A. & Hirschberg, stem region contains a large proportion of proline (30% of the C. B. (1993) J. Bid. Chem. 268,20091-20095

9.46.2-

3.92.8

1.8

0.90.6-

N-Deacetylase IN-Sulfotl "ansferasecDNA Sequence Joziasse, D.H. (1992) Glycobwlogy 2,271-277 Joziasse, D. H., Shaper, N. L., Kim, D., Van den Eijnden, D. H. & Shaper, J. H. (1992) J. Biol. Chem. 267,5534-5541 Kjellbn, L., Pettersson, I., Lillhager, P., Steen, M.-L., Pettersson, U., Lehtonen, P., Karlsson, T.,Ruoslahti, E. & Hellman, L. (1989) Biochem. J. 283, 105-113 Kjellbn, L.,Pettersson, I., Unger, E. & Lindahl, U. (1992)Adu.Exp. Med. Biol. 313, 107-111 Kozak, M. (1989) J. Cell Bwl. 108,229-241 Kyhse-Andersen, J. (1984) J. Biochem. Biophys. Methods 10,203-209 Lidholt, K. & Lindahl, U. (1992) Biochem. J. 287,21-29 Lindahl, U. (1989) Heparin: Chemical and Biological Properties, Clinical Applications (Lane, D.A., and Lindahl, U. , eds) pp. 159-189, Edward Arnold, London Machamer, C. E.(1991) l h e n d s Cell Biol. 1, 141-144

10443

Navia, J. L., Riesenfeld, J., Vann, W. F., Lindahl, U. & Rodbn. L. (1983) Anal. Biochem. 136, 13P140 Pettersson, I., K w h e , M., Unger, E., Wad, H., Nylund, L., Lindahl, U. & Kjellbn, L. (1991) J. Bwl. Chem. 266,8044-8049 Sambmk, J., Fritsch, E. F. & Maniatis, T.(1989) in Molecular Cloning: A Laboratory M a n w l , Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Shaper, J. H. & Shaper, N.L. (1992) Cum Opin. Struct. Biol. 2,701-709 Smith, D. B. & Johnson, K S . (1988) Gene (Amst.)8 7 , 3 1 4 0 Tabor, S . & Richardson, C.C. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 47674771 Wei, Z., Swiedler, S . J., Ishihara,M., Orellana, A. & Hirschberg, C. B. (1993) Proc. Natl. Amd. Sci. U.S.A. 90, 388543888 Wickner, W. T. & Lodish, H. F.(1985) Science 230,400-407