key: cord-0891880-qrg1rtzi
authors: nan
title: Isolation, characterization, and expression of cDNAs encoding murine alpha-mannosidase II, a Golgi enzyme that controls conversion of high mannose to complex N-glycans
date: 1991-12-02
journal: J Cell Biol
DOI: nan
sha: eb136cf532bef626a3f5d4d32fef88e7b43ef3fb
doc_id: 891880
cord_uid: qrg1rtzi

Golgi alpha-mannosidase II (GlcNAc transferase I-dependent alpha 1,3[alpha 1,6] mannosidase, EC 3.2.1.114) catalyzes the final hydrolytic step in the N-glycan maturation pathway acting as the committed step in the conversion of high mannose to complex type structures. We have isolated overlapping clones from a murine cDNA library encoding the full length alpha-mannosidase II open reading frame and most of the 5' and 3' untranslated region. The coding sequence predicts a type II transmembrane protein with a short cytoplasmic tail (five amino acids), a single transmembrane domain (21 amino acids), and a large COOH-terminal catalytic domain (1,124 amino acids). This domain organization which is shared with the Golgi glycosyl-transferases suggests that the common structural motifs may have a functional role in Golgi enzyme function or localization. Three sets of polyadenylated clones were isolated extending 3' beyond the open reading frame by as much as 2,543 bp. Northern blots suggest that these polyadenylated clones totaling 6.1 kb in length correspond to minor message species smaller than the full length message. The largest and predominant message on Northern blots (7.5 kb) presumably extends another approximately 1.4-kb downstream beyond the longest of the isolated clones. Transient expression of the alpha-mannosidase II cDNA in COS cells resulted in 8-12-fold overexpression of enzyme activity, and the appearance of cross-reactive material in a perinuclear membrane array consistent with a Golgi localization. A region within the catalytic domain of the alpha-mannosidase II open reading frame bears a strong similarity to a corresponding sequence in the rat liver endoplasmic reticulum alpha-mannosidase and the vacuolar alpha- mannosidase of Saccharomyces cerevisiae. Partial human alpha- mannosidase II cDNA clones were also isolated and the gene was localized to human chromosome 5.

Abstract. Golgi a-mannosidase II (G1cNAc transferase 1-dependent a1,3[a1,6] mannosidase, EC 3.2.1.114) catalyzes the final hydrolytic step in the N-glycan maturation pathway acting as the committed step in the conversion of high mannose to complex type structures . We have isolated overlapping clones from a murine cDNA library encoding the full length a-mannosidase II open reading frame and most of the 5' and 3' untranslated region . The coding sequence predicts a type II transmembrane protein with a short cytoplasmic tail (five amino acids), a single transmembrane domain (21 amino acids), and a large COOH-terminal catalytic domain (1,124 amino acids) . This domain organization which is shared with the Golgi glycosyltransferases suggests that the common structural motifs may have a functional role in Golgi enzyme function or localization .

Three sets of polyadenylated clones were isolated extending 3' beyond the open reading frame by as N AND O-GLYCAN structures are increasingly being found to contribute to biological recognition events during development, oncogenic transformation, and cell adhesion (3, 10, 11, 37, 38, 40) . The enzymes involved in the maturation of cell surface and intracellular N-glycans are found in the endoplasmic reticulum and Golgi complex where they act upon newly synthesized glycoproteins to generate an array of different structures from a common oligosaccharide precursor (18) . The N-glycan processing pathway consists of three stages : (a) the initial synthesis of a dolichol-linked precursor oligosaccharide and the en bloc transfer of the oligosaccharide to newly synthesized polypeptide Asn-X-Ser/Thr sequons on the lumenal face of the ER; (b) the trimming of the high mannose structures by a-glucosidases and a-mannosidases in the ER and Golgi complex; and (c) the elaboration of the branched oligosaccharide chains by Golgi glycosyltransferases . The trimming phase of the pathway is accomplished by a-glucosidases I and II as well as a collection of processing a1,2-mannosidases (29) in the ER and Golgi complex. The resulting Man5GlcNAc2 structure is then modified by the addition of © The Rockefeller University Press, 0021-9525/91/12/1521/14 $2 .00 The Journal of Cell Biology, Volume 115, Number 6, December 19911521-1534 1521 much as 2,543 bp . Northern blots suggest that these polyadenylated clones totaling 6.1 kb in length correspond to minor message species smaller than the full length message. The largest and predominant message on Northern blots (7.5 kb) presumably extends another N1.4-kb downstream beyond the longest of the isolated clones. Transient expression of the a-mannosidase II cDNA in COS cells resulted in 8-12-fold overexpression of enzyme activity, and the appearance of crossreactive material in a perinuclear membrane array consistent with a Golgi localization . A region within the catalytic domain of the a-mannosidase II open reading frame bears a strong similarity to a corresponding sequence in the rat liver endoplasmic reticulum a-mannosidase and the vacuolar a-mannosidase of Saccharomyces cerevisiae. Partial human a-mannosidase II cDNA clones were also isolated and the gene was localized to human chromosome 5.

a single G1cNAc by G1cNAc transferase I (GnT 1),' before the final hydrolytic steps in the pathway are accomplished by a-mannosidase II (Man 11), catalyzing the removal of a1,3and a1,6-mannosyl residues (50) . The trimming and elongation phases ofthe pathway overlap at the GnT I/Man II steps, with each reaction being obligatory for further processing steps. GnT I is essential for processing to hybrid or complex type structures (34) , while the absence of Man II activity, either by inhibition with the alkaloid, swainsonine (52) , or in the human autosomal genetic disease hereditary erythroblastic multinuclearity associated with positive acidified serum (HEMPAS), characterized by the reduced expression of Man 11 (12) , results in the accumulation of Asn-linked hybrid oligosaccharides in lieu of the standard array of complex type structures . The cleavage of glycoprotein processing in-termediates by Man II also confers resistance to cleavage by endoglycosidase H, a commonly used marker for transit through the Golgi complex (42) . A number of the mammalian Golgi glycosyltransferases havebeen cloned recently (1, 7, 17, 21, 25, 32, 45, 46, 53, 56) allowing a comparison of the polypeptide structures and the ability to examine the regulation of transferase expression in relation to terminal oligosaccharide processing events. Among these enzymes a common domain motifhas been described (37) of a type II transmembrane structure with a small NHz-terminal cytoplasmic tail, a single transmembrane domain, a variable stem region, and a large 000Hterminal catalytic domain . Among the processing hydrolases, only an a-mannosidase presumably responsible for the removal of a single a1,2-mannosyl linkage in the ER has been fully cloned (2) . This enzyme bears no resemblance to the glycosyltransferase domain structure, but it does share extensive homology to the yeast vacuolar a-mannosidase. Both enzymes contain no classical signal sequence or membrane-spanning domain. Characterization of the primary structures and molecular aspects of the early processing steps therefore await the cloning and characterization of the remainder of the processing a-glucosidases and a-mannosidases .

The most well characterized of the processing a-mannosidases is Man lI. The enzyme has been purified and extensively characterized from rat liver (for review see reference 29) . It is a transmembrane glycoprotein with an apparent molecular mass of 124 kD by SDS-PAGE and a catalytic domain facing the lumen of the Golgi complex. Release of a 110AD catalytically active soluble form of the enzyme can be accomplished by a mild chymotrypsin digestion of permeabilized or solubilized Golgi membranes (28) . The chymotrypsin-cleaved form ofthe enzyme has been purified and is catalytically indistinguishable from the intact enzyme (30) , while it differs in NH,-terminal sequence and hydrophobic character. NHz-terminal sequence data from the 110-kD soluble form of the enzyme along with internal peptide sequences have allowed us to generate a Man II-specific cDNA probe by mixed oligonucleotide-primed amplification of cDNA (26) . A partial cDNA clone was isolated from an oligo(dT)-primed rat liver cDNA library which spanned -40% of the Man II open reading frame . We present here the isolation of cDNA clones which span the entire open reading frame and much of the 5' and 3'untranslated regions of Man II from a random-and oligo(dT)-primed murine cDNA library. The open reading frame confirms the map of peptide sequence data from the purified rat protein demonstrating that Man II conforms to the domain structure model common among the Golgi glycosyltransferases . Transient expression of a full length murine Man II cDNA clone in COS cells directs the overexpression of enzyme activity and the synthesis of immunoreactive material in a perinuclear membrane array consistent with the localization of the Man II polypeptide in the Golgi complex of the transfected cells. 

Man II was assayed using 4-methylumbelliferyl a-D-mannoside as substrate as described (27) . NaCl-washed microsomal membranes from COS cell monolayers were prepared from four 100-mm culture dishes (90% confluent) exactly as previously described for 3T3 cells (27) . Protein concentration was determined using the BCA protein assay reagent as described by Pierce Chemical Co. (Rockford, IL) using BSA as standard.

An unamplified BALB/c 3T3 cDNA library primed with a mixture of oligo(dT) and random hexamers and packaged into a XZAP II cloning vector (Stratagene) was obtained from D. J . G. Rees (Massachusetts Institute of Technology) (41) . An amplified version of the same library was used in the second round of screening. A similarly prepared amplified HepG2 cDNA library was obtained from E . Marcantonio (Columbia University, New York, N .Y.) . The packaged libraries were plated on XLl-Blue host cells and screened by plaque hybridization using standard procedures. The probe that was used for library screening and blot hybridizations (see below) was a 1,170 by Man II PCR amplification fragment (PCR-1) generated from a rat liver cDNA preparation by amplification with Man II-specific degenerate and inosine-substituted oligonucleotide primers designed using protein sequence data from the purified Man II polypeptide (26) .

Selected XZAP II clones were subcloned into M13 and sequenced by the dideoxy" chain termination method (43) using deoxyinosine triphosphate in place of dGTP and Sequenase (United States Biochemical) as described by the manufacturer. Sequence data were obtained from successive deletions in M13 using T4 polymerase (Cyclone Biosystem, International Biotechnologies Inc., New Haven, CT) or using synthetic oligonucleotide primers. Oligonucleotides were synthesized on an Applied Biosystems (model 380B) DNA synthesizer.

RNA was prepared from adult male rat tissues as described (2) . Indicated quantities of poly(A+) RNA were resolved on a 1% formaldehyde/agarose gel (26) and transferred by capillary blotting to a Zetaprobe membrane (Bio-Rad Laboratories) . Filters were prehybridized, hybridized, and washed as described (6) using the rat PCRl cDNA amplification product (26) as the radiolabeled probe .

Clone MII-8 containing the entire Man II open reading frame inserted into the EcoRl site of pBluescript II (see Fig . 1 ) was linearized by digestion with BamHI and transcribed in vitro using the T7 polymerase promoter of pBluescript II and the in vitro transcription kit from Stratagene. RNA synthesis was carried out for 2 h at 37°C in a 40 pl reaction volume containing 40 mM Tris HCl (pH Z5), 50 mM NaCl, 8 mM M9Clz, 2 mM spermidine, 0.3 mM M~G(5)ppp(5)G, 30 mM l7rT, 80 U RNasin, 0.5 mM each ATP, UTP, and CTP, 0.17 mM GTP, 10 U of T7 polymerase, and 5,ug of the linearized clone MII-8 template. After the addition of GTP to 0.5 mM the reaction was continued for another 16 h at 37°C . The reaction mixture was extracted twice with phenol/chloroform/isoamyl alcohol followed by ethanol precipitation . Control reactions contained plasmid without insert or plasmid with a noncoding insert.

Translation in vitro was carried out for 1.5 h at 30°C in a 25,ul reaction volume containing all amino acids except methionine (1 mM each), 40 U RNasin, 25 pCi of a mixture of [35S]L-cysteine and [35 S]L-methionine (Tran 35 S-label, ICN Radiochemicals, 1,024 Ci/mmol), the RNA synthesized above, and a rabbit reticulocyte lysate preparation from Promega. Incubations containing dog pancreas microsomal membranes were carried out by addition of 1 .8 icl of microsomal membranes (Promega)/25-p1 reaction either before or following the translation reaction . Posttranslational addition of microsomal membranes was followed by a 45-min incubation at 30°C . Samples were either denatured directly in SDS sample buffer or processed for proteolysis by the addition of trypsin (Sigma Chemical Co.) to 100 jig/ml, t 0.1% Triton X-100, followed by incubation at 0°C for 1 h. Proteolysis was terminated by addition of 30 KIU of aprotinin and boiling in SDS sample buffer. SDS-PAGE and autoradiography were carried out as described (26) .

COS cells were grown in 100-mm culture dishes in a humidified incubator at 37°C in 5% COz with DME media containing 0.1 fcg/ml penicillin and streptomycin and 10% FCS (DME/10% FCS) . Confluent cell monolayers were trypsinized and split 1 :8 two days before transfection . The large EcoRI fragment from clone MII-8 containing the entire Man II open reading frame was excised and ligated into the EcoRI site of the pXM COS cell expression vector (58) . Recombinant plasmids were checked by restriction mapping to confirm the correct orientation of the insert. COS cells (80% confluent) were transfected either with the pXM without an insert, pXM containing the Man II insert in the correct orientation (MII-pXM), or pXM containing the Man II insert in the antisense orientation (asMII-pXM) by liposomemediated transfection (Lipofectin ; Bethesda Research Labs, Gaithersburg, MD) as described by the manufacturer. Briefly, 20 kg of the relevant plasmid were mixed with 50 A,1 of Lipofectin reagent in 3 ml of serum-free medium (Opti-MEM ; Bethesda Research Labs) and incubated at room temperature for 15 min. Monolayer cultures were rinsed twice with Opti-MEM and incubated with the 3 ml media containing the plasmid/lipofectin mixture tbr 6-8 h at 37°C. The cultures were rinsed twice with DME/10 % FCS and incubated with DME/10% FCS at 37°C for 24-72 h . Cells prepared for inrmunofluorescence studies were plated on coverslips placed in 24-well rnicrotiter plates, grown to 50 % confluency, and transfected at the same plasmid/Lipofectin/Opti-MEM ratio as in the larger scale transfections (total volume of 200 pl/well) .

Biosynthetic Labeling of NIH-373 Cells and COS Cells NIH-3T3 cells were grown in 100-mm culture dishes in DME containing 10% calf serum. For biosynthetic labeling 3T3 or COS cells in 100-mm plates were washed twice with Met-free DME and incubated in Met-free DME for 30 min . Cells were then labeled for 60 min at 37°C with 4 ml Met-free DME containing 120 ttCi/nil [35 S]-labeled mixture of methionine and cysteine (Trap 35 S-label, 1,024 Ci/mmol ; ICN Radiochemicals) . Cell monolayers were washed with DME/10% FCS and chased for 60 min with Moremen and Robbins cDNA Cloning and Expression of Golgi a-Mannosidase 111523 the unlabeled medium . Cells were collected by trypsinization, washed twice in PBS, and vigorously resuspended in lysis buffer containing 1% Triton X-100, 0.5 M NaCl, 20 mM Tris HCl, pH 7.5 . The extract was clarified by centrifugation at 200,000 g for 30 min in a centrifuge (TL-100; Beckman Instruments, Inc., Palo Alto, CA) using a TLA 100.3 rotor at 2°C. The clarified extracts were preadsorbed with 100 Wl of a 50% (vol/vol) slurry of protein ASepharose beads (Phamtacia Fine Chemicals, Piscataway, NJ) for 1 h at 4°C. After the removal ofthe beads by centrifugation, 5 p1 of anti-Man II antiserum (27) was added and incubated at 4°C for 4 h with constant mixing. Protein ASepharose was added (50 Wl of the 50% slurry) and the mixture was incubated for an additional 2 h at 4°C with mixing. The immunoprecipitates were washed, eluted, and resolved by SDS-PAGE as previously described (27) . Samples digested with N-glycanase (Genzyme) were processed as described (27) .

Cell monolayers were fixed at the indicated time posttransfection by washing three times with PBS and incubating for 15 min at 37'C in a 3.7 % solution of formaldehyde in PBS. Coverslips were removed from the wells, washed with PBS, and permeabilized by incubation for 1 min in a solution of 100% methanol at -20°C . After washing with PBS the cells were incubated in a solution containing 1% FCS in PBS for 30 min at room temperature followed by a 30-min incubation at 30°C with a 1:2,000 dilution of anti-Man II antibody (27) in PBS. The coverslips were washed again in PBS and incubated for 30 min at 30°C with a 1 :1,000 dilution of a FITC-conjugated goat anti-rabbit antibody (Cappel Laboratories) in PBS containing 1% FCS. Cells were observed and photographed using an Axioplan microscope (Carl Zeiss, Oberkochen, Germany) .

Human Man II-specific primers were synthesized to assay for the presence of the human Man II gene in a human/hamster somatic cell hybrid panel . DNA isolated from a panel of 25 human/hamster hybrids (50 pg/ml, BIOS Corp .) was screened by PCR using primers with the following sequence : sense primer-GCTCGGATGCTACTAGA (3/17 mismatch with murine Man 11) ; antisense primer-TCTTAACTTTAAACTTGGA (7/19 mismatch with murine Man II) bracketing a 178 by of the human Man II gene which was not interrupted by introns (see Fig . 1 ) . Amplification reactions (25 pl) containing 50 mM KCI, 10 mM Tris HCl (pH 8.3), 1.5 mM MgC12, 200 pM each dNTP, 50 ng genomic DNA, 0.5 AM each primer, and 0.6 U Taq polymerase. A temperature cycle of 92°C (1 min), 55°C (1 min), and 72°C (2 min) was repeated for 35 cycles followed by an extension of 4 min at 72°C. Amplification products were resolved on a 2 % agarose gel containing ethidium bromide and scored for the presence of the 178-bp human Man II amplification fragment . A faint 178-pb band as well as several characteristic nonspecific bands were found in hamster controls and in human/hamster hybrids lacking chromosome 5, but the human amplification product could be unambiguously detected above this background by a greater yield, the presence of unique HindUI and EcoRl sites, and the DNA sequence of the human Man II cDNA (data not shown) . 

R

ce'Pr:~:C"sWeeftec~:"c.Wr"r~~ar ""sr; Computer Methods DNA sequence data was assembled into a contiguous sequence database by the method of Staden (49) . Sequence comparisons against the GenBank or GenPept sequence databases were performed using the FASTA program (39) . Statistical analysis of sequence similarity between two protein sequences was determined using the Bestfit program of the University of Wisconsin Genetics Computer Group (version 6.2) or the sequence similarity investigation program (DIAGON) of Staden (49) .

We have previously isolated a partial cDNA clone encoding -40% of the rat Man II open reading frame (26) using a probe isolated by mixed oligonucleotide-primed amplification of cDNA (PCR-1) . Since Northern blots demonstrated a message size of -7.5 kb (adjusted downward from the previously published -8 kb; 26) we decided to perform subsequent rounds of screening using a library primed with both oligo(dT) and random hexamers (41) . Approximately 1 x 106 independent recombinants from the unamplified 3T3 Moremen We have also screened a human HepG2 cDNA library with the rat PCR amplification product (PCR-1) as a probe to isolate human cDNA clones for sequencing and for use as hybridization probes . These probes have been used to examine the expression of Man II in HEMPAS disease (12) , a heterogeneous disease characterized in some individuals by a deficiency in Man II . The sequence has also been used to design human Man II specific primers for chromosome mapping by PCR (see below) . The HepG2 library was screened (4 X 105 recombinants) and three independent clones were isolated . The longest of the clones (ti1.9 kb) contained -55 ofthe human Man II open reading frame and aligned to position -5 to +1,928 on the 3T3 sequence in Fig . 2 (Fig . 1 ) .

The murine Man II coding sequence predicts a polypeptide of 1,150 amino acids (M, 131,000) with a single type II The Joumal of Cell Biology, Volume 115, 1991 Figure 3 In vitro transcription/ translation of Man II clone MII-8 and comparisonwithbiosynthetically labeled Man II. Man II clone MII-8 was transcribed in vitro using T7 polymerase as described in Materials and Methods . In vitro translation was performed with or without the addition of microsomal membranes (co-or posttranslational addition of membranes) as indicated at the bottom of the figure . Aliquots of the translation were thendigested or mock digested with trypsin in the presence or absence of Triton X-100 as indicated . Samples were resolved by SDS-PAGE and subjected to autoradiography. Lanes A and B represent cell extracts from biosynthetically labeled 3T3 cells which were immunoprecipitated with the anti-Man II antibody and either digested (B) or mock digested (A) with N-glycanase . Samples were resolvedon the same gel as the in vitro translation samples . Lanes A and B were exposed to x-ray film for 3 d while the in vitro translation samples were exposed for 12 h . Identical results for the in vi immunoprecipitated with anti-Man II antibody following the synthesis and pro-kD) are indicated at the left of the figure . transmembrane domain from amino acid residue 6-26. Several lines of evidence suggest that the coding region shown in Fig . 2 contains the correct initiation site for the Man II open reading frame. The Met codon in position 1 is the first ATG in the sequence and conforms to the consensus eucaryotic translation sequence with a purine at position -3, the most critical residue in the initiation sequence (19) . The 5' untranslated region is G/C rich (68%), a characteristic common to several of the Golgi glycosyltransferases (46, 53) and also a common feature among housekeeping genes (9) . Finally, the NHZ-terminal peptide sequence of the purified intact rat liver enzyme starts at residue 6 of the predicted open reading frame (30) . The purified enzyme is indistinguishable in size from the biosynthetically labeled 3T3 enzyme (27) and the enzyme detected by Western blots from freshly prepared Golgi membrane extracts (27), but the cleavage of five residues would be too small a change to detect on SDS gels during purification . Although the cleavage of the five residue cytoplasmic tail most likely reflects in vitro proteolysis during purification, it could also possibly represent the product of cleavage in vivo .

The in vitro transcription/translation of the clone MR-8 cDNA also resulted in a band of identical mobility to the deglycosylated 3T3 enzyme (Fig . 3) . Translation in vitro in the presence of microsomal membranes resulted in the glycosylation of the polypeptide and a decrease in mobility to a doublet of identical size to that of the glycosylated 3T3 . Comparison of murine Man II protein sequence with the yeast vacuolar a-mannosidase and the rat ER a-mannosidase. Murine Man II was aligned with the yeast vacuolar a-mannosidase and the rat ER a-mannosidase using the FASTA program (54) . The sequences in the figure show the individual alignments of murine Man II (center sequence) with the yeast vacuolar a-mannosidase (59) (Yeast Man, upper sequence) and the rat ER a-mannosidase (24) (ER Man, lower sequence) . The vertical bar indicates amino acid identities between the two lines of sequence . The double dots indicate conservative amino acid substitutions between the two lines of sequence . The numbering below the aligned sequences refers to the polypeptide numbering of 3T3 Man II in Fig . 2 . When the indicated sequence pairs were subjected to statistical analysis using the Bestfit program from the University of Wisconsin Genetics Computer Group (comparison versus randomized sequences with the same amino acid content, gap weight = 3, gap length weight = 0.1) the sequence alignments were found to be 9 .7 and 6.0 SDs above random for the Man II/yeast vacuolar a-mannosidase pair and the Man II/ER a-mannosidase, respectively. biosynthetic product . The doublet likely results from the glycosylation of Asn residue 1,129 and a partial glycosylation at Asn residues 78 or 93, since the purified soluble form of the rat liver enzyme (residues 107-1,150) has been shown to be fully glycosylated at a single site (30) . The apparent molecular weight on SDS gels is slightly less than is predicted for the open reading frame of the cDNA (rv117 kD vs 131 kD for the deglycosylated forms, respectively) . This presumably results from an anomalous gel migration since the in vitro translation product and the product of biosynthetic labeling run at an identical position by SDS-PAGE . Translation in the presence of membranes also resulted in the apparent co-translational translocation of the polypeptide across the lipid bilayer. Trypsin cleavage of the in vitro translation products in the presence of microsomal membranes did not alter the amount or the mobility of the glycosylated form but the smaller in vitro-synthesized products were fully susceptible to digestion . The lack of a mobility shift following trypsin digestion also confirms the orientation in the membrane and the location of the membrane-spanning domain . These results are also consistent with the inability to detect any cytoplasmically oriented polypeptide in protease protection experiments in intact Golgi membranes (27) . Addition of membranes posttranslationally did not result in a significant production of glycosylated product suggesting a requirement for co-translational translocation of the enzyme .

In addition to the NHZ-terminal peptide sequence, the mouse equivalent of the six remaining rat Man II polypeptide sequences determined previously (30) were all found in the Man II open reading frame confirming the predicted peptide sequence map (Ref. 30, Fig . 8 ) . Chymotrypsin has been used to cleave Man II in vitro generating a soluble catalytically active form of the enzyme (28, 30) . The cleavage site is at residue 107 in the open reading frame and predicts that the cytoplasmic tail, transmembrane domain, and 80 residues of Moremen and Robbins cDNA Cloning and Expression of Golgi a-Mannosidase 11 1527 lumenally oriented polypeptide are not essential for enzyme activity.

Although Man II appears to have the domain structure common to the other Golgi glycosyltransferases no direct sequence homology was found with any of the glycosyltransferases. The previously described similarity between rat Man II and rabbit GnT 1 (45) was found to be marginally significant (4 SDs above random) . When the murine and human Man II equivalents of this sequence were compared to either the rabbit or human (15) GnT I sequences the similarities were even less significant (2 .1 and 1 .7 SDs above random for the murine and human Man II sequences, respectively) . Comparison of the Man II polypeptide to a translated form of the GenBank database (GenPept, version 64 .3) revealed that a 215 amino acid region of Man II bears a statistically significant similarity to a corresponding region in the rat ER a-mannosidase (2) and the vacuolar ci-mannosidase from Saccharomyces cerevisiae (59) (6 .0 and 9 .7 SDs above random, Fig . 4) . The rat ER a-mannosidase and the yeast vacuolar a-mannosidase share extensive homology across their entire length (2), whereas Man II shares a more limited similarity within a region of the soluble catalytic domain of the enzyme . All three enzymes recognize a-mannoside linkages and cleave the synthetic substrates p-nitrophenyl a-Dmannoside and 4-methylumbelliferyl a-D-mannoside, but differ in their specificity toward natural high mannose oligosaccharides, inhibition by alkaloid inhibitors, antibody crossreactivity, and subcellular-_Iocalization (2, 29, 59, 60) . The sequence similarity and the common activity toward synthetic substrates between the three enzymes would, however, suggest that this region may represent a portion of the active site involved in a-mannoside recognition .

Comparison of the sequence of the 3' untranslated region of murine Man II with the GenBank database (version 67.0) revealed that the extended 3'-untranslated region contains a Arrows indicate the positions of the major and minor message species in rat tissues. The blot in lane 1 was exposed to x-ray film for 8 d at -70°C. The gel in the remaining lanes was exposed for 2 d at 70°C. single interrupted copy of a mouse Bl repetitive sequence (data not shown), the mouse equivalent of the human Alu consensus sequence (20, 48) .

The isolation of several size classes of polyadenylated clones that differ in their degree of extension in the 3' untranslated region demonstrates the heterogeneous nature of the termination and polyadenylation of the Man 11 message. Each of Table I Figure 6 . pH profiles ofa-mannosidase activity in COS cells transfected with control (pXM, o) or Man II clone MII-8 (MII-pXM, o) constructs . Salt-washed membranes were prepared and assayed as described in Materials and Methods with the 4methylumbelliferyl a-n-mannoside substrate at the indicated pH . Membranes prepared from COS cells transfected with MII-pXM were also assayed at the indicated pH inthe presence of 10 AM swainsonine (0) . Residual a-mannosidase activity in the supernatant following immunoprecipitation of the salt-washed membrane extract with antirat Man II antibody is as indicated (m) . transcript of 7.5 kb was found in all tissues with the greatest enrichment being in adrenal and thymus . Most other tissues resulted in an autoradiographic band of -3-10-fold lesser intensity except skeletal muscle which was undetectable except with prolonged exposure. The ratio of the multiple Man II message species appeared constant across the range of rat tissues despite the >100-fold difference in the level ofexpression between adrenal and muscle tissues. Comparison with the rat ER a-mannosidase (2), reveals that while the Man II transcript is most abundant in adrenal and thymus, the ER enzyme has a highest abundance level in adrenal and testis and minor differences in other tissues except spleen, intestinal epithelia, and muscle tissues which have a slightly reduced level ofmessage. The differences with Man II message expression in the thymus, testis, and muscle suggests that there is a differential tissue expression of the two a-mannosidases at least in these tissues . A low but detectable level of Man II message was also found in rat brain tissues consistent with a report of reduced levels of Man II activity in rat brain and the presence of a novel brain a-mannosidase (51) which is not sensitive to inhibitionby swainsonine and is able to directly cleave high mannose structures to Man3GlcNAc2 without the prior addition of a G1cNAc residue by GnT I. Whether the message levels adequately reflect the levels of the Man II enzyme activity in the brain or inthe other tissues remains to be determined. Although there is a >100-fold difference in Man II autoradiographic intensity on Northern blots between thymus and muscle, it is also not clear if this merely reflects the biosynthetic level and the degree of elaboration of the Golgi complex and the secretory pathway/glycosylation machinery in the respective tissue types .

The large EcoRI fragment of clone MII-8 containing the en- Moremen tire Man II open reading frame was subcloned into the COS cell expression vector pXM . This is an SV-40-based expression vector driven by the adenovirus major late promoter (58) . COS cells were transfected and assayed for Man II activity either in cell homogenates or in membrane fractions . Crude homogenates showed approximately threefold overexpression of 4-methylumbelliferyl a-mannoside activity (Table I), but Man II activity in crude homogenates is commonly masked by the lysosomal a-mannosidase or the ER a-mannosidase when using synthetic substrates . A salt washed membrane fraction was prepared which has been shown to result in the enrichment ofMan II activity (27) . Assays of this membrane fraction revealed a 8-12-fold overexpression of 4-methylumbelliferyl a-mannoside activity in COS cells transfected with MII-pXM when compared cells transfected with the vector alone or vector with an antisense insert (Table I) . This degree of overexpression is similar to the activity levels seen for the 01,4-galactosyltransferase expressed in COS cells (24) . The overexpressed Man II activity has a pH optimum of 5.5, identical to the endogenous enzyme in 3T3 cells (27), it is sensitive to inhibition by swainsonine, and is immunoprecipitable with antibody to the purified rat enzyme (Fig. 6 ) . Biosynthetic labeling of transfected COS cells resulted in a >10-fold overexpression ofimmunoprecipitable polypeptide (data not shown) reflecting the high rate of biosynthesis of the vector encoded polypeptide 48 h after transfection .

Immunocytochernistry at the EM level using antibodies to purified rat Man II has demonstrated the Golgi localization of the enzyme in rat tissues (44) . The antibodies have also been used recently as a marker for Golgi membrane components in cell trafficking studies (8, 22; and C. Zuber, A. Nakano, K. W. Moremen and J. Roth, manuscript submitted for publication) . Preliminary studies have shown that the antibody raised to the rat enzyme does not cross react with the COS cell enzyme (data not shown) allowing a minimal immunofluorescence background in the host cells. Transfection of COS cells with the pXM construct containing the Man II insert in the sense orientation (MII-pXM) resulted in an anti-Man II immunofluorescence pattern within 24 h ofthe initiation oftransfection. The immunofluorescence was largely restricted to a crescent shaped reticular pattern adjacent to the nucleus, but occasionally this reticular structure would extend well into the cytoplasm (Fig. 7, A and B ). This pattern of immunofluorescence has been previously shown to be characteristic ofthe Golgi complex (22, 31, 44) . Mock transfections, transfections with vector alone, or with the Man II insert in the antisense orientation resulted in no detectable immunofluorescence signal (data not shown) . Similarly, immunofluorescence offixed but not permeabilized cells transfected with MII-pXM resulted in no detectable fluorescence signal . The transfection efficiency ofthe MII-pXM construct, as measured by thepercentage of fluorescence-positive cells, was variable, but ranged from 5-25 R . By 36 h posttransfection the intensity of the peri-nuclear immunofluorescence increased substantially with many cells generating an additional diffuse particulate pattern of fluorescence in the cytoplasm consistent with ER staining (8, 22, 31) (Fig. 7 , C-E) . By 48 h postuansfection immunofluorescence staining adjacent to the nucleus remained intense but the particulate cyto- The human chromosome compliment of each hybrid was compared with the ability to generate the 178-bp human Man II amplification fragment following PCR. Concordant segregationof the amplification fragment was observed only for chromosome 5. All other chromosomes were excluded as the site for the Man II gene by discordancies in at least 15 hybrids . Hybrid clones containing the indicated chromosome in at least 5% of the cells were considered positive for the chromosome in the discordancy analysis . plasmic staining greatly increased . With the high level of Man II synthesis late in the transient expression, detection of early biosynthetic intermediates in the ER or an accumulation of aggregated enzyme in the ER might be expected.

Chromosome Mapping of the Human Man II Gene Comparison of the murine Man II cDNA with the sequence of the partial human Man II clone revealed that the DNA sequences are 84 % identical at the DNA level and 85 % identical at the amino acid level (data not shown) . Primers were designed to a region ofthe human Man II cDNA which was distinct from the murine cDNA sequence (3/17 and 7/19 mismatches, respectively) and produced a clear 178-bp PCR product with the human Man II cDNA sequence upon amplification with human genomic DNA. Amplification with hamster genomic DNA resulted in a faint band at 178 pb that was distinguishable from the human amplification product both in quantity and by the absence ofthe HindIII and EcoRI sites characteristic of the human Man II genomic sequence. DNA from a human/hamster somatic cell hybrid panel (4) was tested for the amplification of the 178-bp human Man II fragment. Cell lines which scored positive for the human Man II PCR fragment were also all found to contain human chromosome 5 (Table I) . One of the lines (hybrid line 212) contained multiple deletions in 5q and several lines contained deletions in the 5p15.1-5pl5.2 region . All of these Moremen and Robbins cDNA Cloning and Expression of Golgi a-Mannosidase 11 1531 lines scored positive for amplification of the human Man II fragment. All other chromosomes were excluded as the site for the Man II gene by discordancies in at least 15 of the hybrid lines.

Man II occupies a central position in the Asn-linked oligosaccharide processing pathway acting as the committed step in the synthesis ofcomplex type structures (29) . The enzyme has long been considered a marker for the Golgi complex by cell fractionation (14) and immunocytochemistry (35) , as well as being the biochemical marker responsible for the conversion of N-linked glycans to structures which are resistant to cleavage by endoglycosidase H (50) . Intact Man II has been purified and characterized from rat liver (29) and shown to be a disulfide-linked homodimer (30) with a catalytic domain oriented toward the Golgi lumen. Protease protection experiments in intact Golgi membranes were unable to detect any cytoplasmically oriented polypeptide (28) . The release of a soluble form of the enzyme following chymotrypsin digestion ofsolubilized Golgi membranes allowed us to purify and compare the intact and soluble forms of the enzyme demonstrating that the soluble catalytic domain retains all of the catalytic characteristics of the intact enzyme. The cloning of the full-length murine Man II open reading frame has allowed us to demonstrate that Man II shares several of the biochemical features of the Golgi glycosyltransferases (37) , namely a type II transmembrane orientation with an NH2-terminal membrane anchor domain and susceptibility to proteolytic release of a soluble, catalytically active form of the enzyme.

Several lines of evidence indicate that the isolated cDNA clone MII-8 encodes the authentic Man II polypeptide. The NH2-terminal sequence of the intact purified enzyme aligns to amino acid position 6 of the translated cDNA at the junction of the cytoplasmic tail and the transmembrane domain. Whether the cleavage of the five amino acids segment represents an in vivo or in vitro event is difficult to determine since it is not possible to distinguish the cleaved from uncleaved forms based on size. The remaining six rat polypeptide sequences can be identified in the translation of the murine cDNA with several conservative amino acid substitutions between the two species . The Man II cDNA clone transcribed and translated in vitro was found to yield a polypeptide identical in size, glycosylation pattern, and topology in microsomal membranes to the rat liver (28) or biosynthetically labeled 3T3 enzyme (27) . Expression of the Man II cDNA in COS cells resulted in a 8-12-fold overexpression of enzyme activity with the same pH optimum, sensitivity to inhibition by swainsonine, and crossreactivity with the anti-rat Man II antibody as the endogenous 3T3 enzyme (27). Finally, the enzyme expressed in COS cells was localized by immunofluorescence to a perinuclear membrane array characteristic of the Golgi complex suggesting an appropriate localization of the transfected cDNA translation product . The open reading frame predicts a type II transmembrane topology with a 5 amino acid cytoplasmic tail, a single transmembrane domain, a "stem" region of at least 80 amino acids based on proteolysis studies, and a 1,044 amino acid catalytic domain. Although the combined "stem" and catalytic domains are two to three times larger than most Chromosome  Concordant  Discordant  Percent discordant   1   6  19  76  2  4  21  84  3  5  20  80  4  5  20  80  5  25  0  0  6  7  18  72  7  5  20  80  8  6  19  76  9  6  19  76  10  6  19  76  11  6  19  76  12  7  18  72  13  9  16  64  14  10  15  60  15  7  18  72  16  6  19  76  17  3  22  88  18  3  22  88  19  10  15  60  20  6  19  76  21  10  15  60  22  7  18  72  X  6  19  76  Y  7  18  72 of the glycosyltransferases, the general features of membrane topology are identical between the two classes of enzymes (37) . Transmembrane topologies of plasma membrane and ER proteins are quite varied (55) suggesting thatthere is no obvious selective advantage of a single topology for membrane protein function. Many of the Golgi glycosyltransferases (37) , as well as Man II (28) , retain full catalytic activity when released from their membrane anchoring domains by selective proteolysis either in vivo or in vitro suggesting that the cleaved regions exert little, if any, influence on the catalytic activities of these enzymes . The conservation of topological features between the collection of Golgi glycosyltransferases and a processing hydrolase, Man II, might therefore suggest a functional role for the common domain structure beyond a simple attachment ofthe catalytic domain to the membrane surface. One function that has been proposed (37) for the "stem" region is to confer flexibility to the catalytic domains ofthe Golgi enzymes in order to allow accessibility to lumenal and membrane associated substrates. Although this hypothesis provides a logical resolution to the problem of substrate accessibility, evidence for this "hinge" or "stem" has yet to be demonstrated either in vitro or in vivo. A Golgi GDPase from S. cereWsiae (57) which cleaves the soluble substrate GDP in the yeast Golgi lumen has recently been cloned and sequenced (C. Abeijon, K. Yanagisawa, K. W. Moremen, C. B. Hirschberg, and P W Robbins, unpublished results) and was found to contain a similar type II transmembrane structure . Since this enzyme presumably cleaves only soluble substrates it would suggest that the conserved topological features of Golgi enzymes might have a function distinct from providing flexibility to the catalytic domain.

Another potential role for the tail/transmembrane domain/stem regions of these Golgi enzymes would be in the recognition and retention of the polypeptides in the Golgi stacks. Recognition signals for soluble and membrane bound polypeptides in the ER have been described . The tetrapeptide recognition signal, KDEL, on the COOH termini of soluble, lumenal ER proteins results in their recognition and retention in the ER (31) . Recently, a recognition signal for ER type I transmembrane proteins has also been described (16, 33, 36, 47) with the recognition sequence KKXX at the COOH termini of the polypeptide sequences being recognized on the cytoplasmic face of the ER (47) . Deletions in the first of three transmembrane domains of the avian coronavirus El glycoprotein disrupted the cis-Golgi localization of this polypeptide suggesting that a membraneassociated region contains at least a portion of the information of Golgi targeting of this viral glycoprotein (23) . The Golgi cisternae have an additional level of complexity, however, since many of the endogenous Golgi enzymes exhibit distinctive sub-Golgi compartmentation . The mechanistic requirements necessary to yield these subtle differences in Golgi localization may therefore be more complex than the simple linear sequences involved in ER protein retention.

The similarity in sequence between murine Man II, the rat ER a-mannosidase, and the yeast vacuolar a-mannosidase was surprising considering the distinctions between the enzymes on substrate specificity and sensitivity to alkaloid inhibitors. All three enzymes recognize the small synthetic The Journal of Cell Biology, Volume 115, 1991 substrates, p-nitrophenyl a-D-mannoside and 4-methylumbelliferyl a-D-mannoside, although with different Ks (29) . These results suggest that the region of similarity might reflect a portion of the catalytic domain involved in a-mannoside recognition . The mammalian lysosomal a-mannosidase (5) also cleaves the small synthetic a-mannoside substrates and, like Man II, is inhibited by swainsonine. When the lysosomal enzyme is cloned and the sequence is determined a comparison with the consensus sequence of the other a-mannosidases would be a critical test of the function of the consensus sequence. A match in the proposed substrate recognition region between the three disparate mammalian a-mannosidases would provide strong evidence that this region is involved in the active site of the enzymes .

The 3' end of Man II was found to be heterogeneous in length with the isolation ofthree distinct size classes of polyadenylated clones . The longest class of clones extends 2,543 by in the 3' direction from the end of the open reading frame before terminating in a poly(A) tract. Comparison between the longest aggregate clone sequence and the message size on Northern blots suggests that the predominant transcript likely reads through the region containing the three polyadenylation signals and uses a termination and polyadenylation signal an additional -1.4 kb further downstream. Repetitive element sequences were found in the long 3' untranslated region of both murine Man II and murine 51,4-galactosyltransferase (46) . Although this type of extended 3' untranslated region is common among the glycosyltransferases the functional significance of this extended transcript is not clear.

The identification of a deficiency in Man II expression as the causative agent in one form of HEMPAS disease (12), a heterogeneous autosomal disease characterized by a defect in the synthesis of cellular and secreted glycoproteins (13) , has focused our interest on the genomic structure of Man II and the regulation ofgene expression . As a first step we have presented the cloning and expression of murine Man II cDNA. In addition we have localized the human Man II gene to chromosome 5. Further characterization of full length human Man II cDNA and genomic clones should allow us to determine the molecular basis of the Man II deficiency in HEMPAS disease and the regulatory features of the processing enzymes responsible for these maturation of cellular and secreted N-glycans.

Isolatio n of a cDNA coding for human galactosyltransferase

Isolation, characterization, and expression of cDNA encoding a rat liver endoplasmic reticulum a-mannosidase

Carbohydrate ligands of the LEC cell adhesion molecules

Genetic caunterselective procedure to isolate interspecific cell hybrids containing single human chromosomes : construction of cell hybrids and recombinant DNA libraries specific for human chromosomes 3 and 4

Purification and comparison of the structures of human liver acid a-n-mannosidases A and B

Genomic sequencing

Cloning of cDNA encoding the membrane-bound form of bovine ,81,4-galactosyltramferase

Dissociatio n of a 110-kD peripheral membrane protein from the Golgi apparatus is an early event in Brefeldin A action

Promoters for housekeeping genes

Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are onco-developmental antigens

Carbohydrate differentiation antigens : probable ligands for cell adhesion molecules

Incomplete synthesis of N-glycans in congenital dyserythropoetic anemia type II caused by a defect in the gene encoding a-mannosidase II

HEMPAS disease : genetic defect of glycosylation

Evidence for extensive subcellular organization of asparagine-linked oligosaccharide processing and lysosomal enzyme phosphorylation

Isolation of 13 and 15 kilobase human genomic DNA clones containing the gene for UDP-N-acetylglucos amine :a-3-D-mannoside 5-1,2-N-acetylglucosaminyltransferase I

Identification of a consensus motif for retention of transmembrane proteins in the endoplasmic reticulum

Bovine a1,3-galactosyltransferase : isolation and characterization of a cDNA clone

Assembly of asparagine-linked oligosaccharides

The scanning model for translation: an update

The nucleotide sequence of the ubiquitous repetitive DNA sequence BI complementary to the most abundant class of mouse fold-back RNA

Isolation of a cDNA encoding a murine UDPgalactose :O-D-galactosyl-1,4-N-acetyl-D-glucosaminide a-1,3 galactosyltransferase : expression cloning by gene transfer

Micrombule-dependent retrograde transport of proteins into the ER in the presence of Brefendin A suggests an ER recycling pathway

The El glycoprotein of an avian coronavirus is targeted to the cis Golgi complex

Expression of bovine 5-1,4-galactosyltransferase cDNA in COS-7 cells

Identification of the full-length coding sequence for human galactosyltransferase (8-N-acetylglucosaminide : 5-1,4-galactosyltransfemse)

Isolation of a rat liver Golgi mannosidase II clone by mixed oligonucleotide-primed amplification of cDNA

Moremen and Robbins cDNA Cloning and Expression of Golgi a-Mannosidase ii 153 3 27

Topology of mannosidase II in rat liver Golgi membrane;and release of the catalytic domain by selective proteolysis

Mannosidases in mammalian glycoprotein processing

Novel purification of the catalytic domain of Golgi a-mannosidase II

A C-terminal signal prevents secretion of lumenal ER proteins

Cloning and sequencing of cDNA of bovine N-acetylglucosamine (sl-4)galactosyltransferase

Shor t cytoplasmic sequences serve as retention signals for transmembrane proteins in the endoplasmic reticulum

Control of glycoprotein synthesis : purification and characterization of rabbit liver UDP-N-acetylglucosamine :ce-3-D-mannoside R-1,2-N-acetylglucosaminyltransferase I

Immunocytochemical localization of cY-D-mannosidase II in the Golgi apparatus of rat liver

A short sequence in the COOH-terminus makes an adenovirus membrane glycoproteins a resident of the endoplasmic reticulum

Glycosyltransferases: structure, localization, and control of cell type-specific glycosylation

Glycoproteins : what are the sugar chains for? 77BS (Trends

Improved tools for biological sequence comparison

Sequence and domain structure of talin

Transport of protein between cytoplasmic membranes of fused cells: correspondence to processes reconstituted in a cell-free system

DNA sequencing with chain-terminating inhibitors

Antibodie s to rat pancreas Golgi subfractions : identification of a 58-kD cis-Golgi protein

Molecular cloning and expression of cDNA encoding the enzyme that controls conversion of high mannose to hybrid and complex N-glycans : UDP-N-acetylglucosamine :a-3-D-mannoside 0-1,2-N-acetylglucosaminyltransferase 1

Characterization of the full length cDNA for the murine 0-1,4-galactosyltransferase

Signals for retention of transmembrane proteins in the endoplasmic reticulum studied with CD4 truncation mutants

Highly repeated sequences in mammalian genomes

Computer handling of DNA sequencing projects . in Nucleic Acid and protein sequence analysis

ci-D-mannosidases of the rat Golgi membranes

Characterization of a novel a-D-mannosidase from rat brain microsomes

Swainsonine inhibits the biosynthesis of complex glycoproteins by the inhibition of Golgi mannosidase II

Primar y structure of 5-galactoside a2,6-sialyltransferase

Role of the conserved AAUAAA sequence : Four AAUAAA point mutants prevent messenger RNA 3' end formation

Multiple mechanisms of protein insertion into and across membranes

Cloning and characterization of DNA complementary to human UDP-GaINAc: Fucal,2GalaI,3GalNAc Transferase (histo-blood group A transferase) mRNA

A guanosine diphosphatase enriched in Golgi vesicles of Saccharmnyces cerevisiae

Huma n IL-3 (Multi-CSF) : Identification by expression cloning of a novel hematopoietic growth factor related to murine IL-3

Nucleotide sequence of AMS1, the structural gene of vacuolar a-mannosidase of Saccharomyces cerevisiae

A novel pathway of import of a-mannosidase, a marker enzyme of vacuolar membrane, in Saccharomyces cerevisiae

We would like to express our thanks for patient advice from Dr. D. J. G. Rees on many aspects of the cloning and sequencing project and his kind giftofthe 3T3 cDNA library. We would also liketo acknowledge the help-