Expression Cloning of a Novel Suppressor of the Lec15 and Lec35 Glycosylation Mutations of Chinese Hamster Ovary Cells*

Lec15 and Lec35 are recessive Chinese hamster ovary (CHO) cell glycosylation mutations characterized by in-efficient synthesis and utilization, respectively, of man- nose-P-dolichol (MPD). Consequently, Lec15 and Lec35 cells accumulate Man 5 GlcNAc 2 -P-P-dolichol and glu- cosaminyl-acylphosphatidylinositol. This report describes the cloning of a suppressor (termed SL15) of the Lec15 and Lec35 mutations from a CHO cDNA library by functional expression in Lec15 cells, employing phyto- hemagglutinin/swainsonine selection. The SL15 protein has a predicted molecular weight of 26,693 with two potential membrane spanning regions and a likely C-terminal endoplasmic reticulum reten- tion signal (Lys-Lys-Glu-Gln). Lec15 cells transfected with SL15 have normal levels of MPD synthase activity in vitro and convert Man 5 GlcNAc 2 -P-P-dolichol to Glc 0–3 Man 9 GlcNAc 2 -P-P-dolichol in vivo . Surprisingly, SL15 also corrects the defective mannosylation in Lec35 cells. The SL15 protein bears no apparent similarity to Saccharomyces cerevisiae MPD synthase (the DPM1 pro-tein), but is highly similar to the hypothetical F38E1.9 protein encoded on Caenorhabditis elegans chromosome 5. These results indicate a novel function for the SL15 protein and suggest that MPD synthesis is more complex than previously suspected. This paper reported the cloning of a cDNA for a new gene, which we termed SL15. We reported that SL15 cDNA corrected two separate glycosylation-defective cell lines known as Lec15 and Lec35. In experiments to be reported in the future we have found that the Lec35 cells used in that study have a defect in the SL15 gene. This provides a clear explanation for correction of Lec35 cells by SL15 cDNA, a result that has now been obtained by three separate individuals in this laboratory as well as by another laboratory (T. Kinoshita, personal commu-nication). However, similar attempts to repeat the correction of Lec15 cells by SL15 cDNA have consistently failed. Based upon further analysis of cell samples from that study, we now con-clude that the Lec15 population was contaminated with Lec35 cells. Lec15 cells described in 2, and 4 as being transfected with SL15 cDNA and having normal glycosylation phenotypes have now been shown to be Lec15-Lec35 hybrids, most likely formed during electroporation. It is likely that SL15 cDNA was carried by contaminating Lec35 cells during expression cloning and that the contaminating Lec35 cells accounted for the greater number of colonies that survived lectin selection after transfection with SL15 cDNA as compared with vector controls.

Monoglycosylated polyisoprenol phosphates are essential sugar donors for prokaryotic and eukaryotic glycoconjugates (1). Eukaryotes use three such molecules, mannose-P-dolichol (MPD 1 ), glucose-P-dolichol (GPD), and GlcNAc-P-P-dolichol (GnPPD). All three are required for the synthesis of 14-sugar dolichol-P-P-linked oligosaccharides, the precursors of asparagine-linked oligosaccharides, contributing 4, 3, and 1 residues, respectively (2). MPD is also required for synthesis of glycosylphosphatidylinositols (GPIs), the precursors of GPI anchors, and donates all 3 mannose residues in GPIs (3). An essential issue is the coordination and regulation of the enzymes that synthesize these three sugar donors, all of which compete for the same limited pool of dolichol-P (4). Isolation and characterization of recombinant DNAs encoding these enzymes represent important steps toward resolving this issue. cDNAs or genes encoding GlcNAc-1-P transferase, the enzyme responsible for GnPPD synthesis, have been cloned from Saccharomyces cerevisiae (ALG7) (5,6), S. pombe (7), Leishmania (8), hamster (9,10), and mouse (11). In contrast, genes encoding the MPD (DPM1) (12) and GPD (ALG5) (13) synthases have been isolated only from S. cerevisiae. Thus, studies of the regulation of these enzymes in animal cells have been impeded.
To advance our understanding of MPD synthase (MPDS) regulation in animals, we designed an expression cloning strategy to isolate a hamster cDNA that could correct the defect in Lec15 Chinese hamster ovary (CHO) cell mutants. Cells of the Lec15 genotype lack MPDS activity in vitro (14) and, accordingly, accumulate Man 5 GlcNAc 2 -P-P-dolichol (15) and GlcN-(acyl)PI (16) in vivo. Thus, there is a good possibility that Lec15 cells lack MPDS or perhaps an accessory protein needed for MPDS activity. This report describes studies of a novel hamster suppressor of Lec15 cells, termed SL15 for suppressor of Lec15.
Preparation and Characterization of Polyoma Large T-antigen Transfected Lec15 Mutants-A MPDS synthase-deficient mutant line of the Lec15 genotype (termed Lec15.2) was isolated by selection of ethylmethanesulfonate-treated CHO-K1 cells with conA/swainsonine (16). The MPDS deficiency was verified biochemically (16), and the complementation group was confirmed as Lec15 (20). CaPO 4 (21) was used to co-transfect pSVE1-B1a encoding the polyoma virus large T-antigen (22) with pB3 encoding a hygromycin resistance gene (20). Hygromycinresistant colonies were picked (20) and screened for large T-antigen function by expression of G418 resistance after electroporation of pcDNAI/neo (Invitrogen Corp.), which contains the polyoma virus origin of replication. The colony with the highest transfection efficiency (termed ptLec15.2D6 and referred to as ptLec15 throughout this paper) was used for expression cloning. Lec35 cells (16) were transfected with polyoma large T-antigen and prepared as described for ptLec15 cells, yielding ptLec35 cells.
Expression Cloning of pLW1 in ptLec15 cells, Sequencing, and Biochemical Analysis-A unidirectional CHO-K1 cell cDNA library of 1.8 ϫ 10 7 primary recombinants in the pcDNAI vector was purchased from Invitrogen Corp (catalogue number A950-01, lot number 100930). The library (40 g of DNA) was electroporated into 5 ϫ 10 5 ptLec15 cells (600 V, 71 microfarads, 50 mA, 50 watts, infinite resistance) with an Invitrogen electroporator. The cells were selected in medium (see Fig. 1 legend) with 20 g/ml phytohemagglutinin E (PHA) plus 1 g/ml swainsonine (Sw) to obtain cells with normal phenotypes as well as Lec15/ Lec1 double mutants. This took 2 weeks. Lec15/Lec1 double mutants were expected to arise spontaneously in Lec15 cells by loss of Golgi GlcNAc transferase I (19). Therefore, cells surviving PHA/Sw were counterselected with conA (no swainsonine) to remove the Lec15/Lec1 double mutants (19), requiring 1 week. DNA was recovered from surviving cells by the Hirt procedure (23). The DNA preparation was enriched for a single plasmid at this point, and this was introduced into Escherichia coli strain MC1061/P3 by electroporation. 10 transformed * This work was generously supported by National Institutes of Health Grant GM38545 and Robert Welch Foundation Grant I-1168. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U55387.
E. coli colonies picked at random all had the same plasmid (however, see "Results and Discussion") which was termed pLW1.
Restriction fragments of the cloned insert of pLW1 were ligated into pUC18 and pUC19 and sequenced on both strands using specific primers with fluorescent terminators on an Applied Biosystems Model 373A Sequencer.
For biochemical experiments, pLW1 was transfected into ptLec15 or ptLec35 cells. After selection with PHA/Sw, there were many resistant colonies which were pooled for further use. No colonies were typically obtained with PHA/Sw selection if vector alone was transfected (data not shown), although in rare cases a presumed Lec15/Lec1 or Lec35/ Lec1 double mutant was detected.

RESULTS AND DISCUSSION
Expression Cloning-The method employed was first suggested by Heffernan and Dennis (22) and subsequently used by others to clone cDNAs encoding several proteins involved in glycoconjugate synthesis (24 -28). In this method, cells with recessive mutations are engineered to express polyoma virus large T-antigen. This allows plasmids containing the polyoma virus origin of DNA replication to replicate freely without incorporation into the genome. Upon transfection of a library of such plasmids and selection for correction of phenotype, freely replicating plasmids are easily recovered and enriched by retransfection if necessary.
To obtain a Lec15 suppressor cDNA we used a Lec15 CHO-K1 mutant line, isolated previously in this laboratory (16), which lacked MPDS activity in vitro and accumulated both Man 5 GlcNAc 2 -PP-dolichol and GlcN-(acyl)PI in vivo. This line was stably transfected with the polyoma large T-antigen and used for expression cloning with a CHO-K1 cDNA library as described under "Experimental Procedures." To isolate positive clones, the transfectants were selected with a mixture of PHA/Sw as described earlier (19). This selection relies upon the observation that loss of MPD results in truncated Man 5 GlcNAc 2 dolichol-linked oligosaccharides. After transfer to protein, such oligosaccharides do not require the action of Golgi mannosidase II for processing from high mannose (conA binding) into complex (PHA-binding) oligosaccharides. Thus, cells lacking MPD are resistant to the effects of Sw, a specific inhibitor of Golgi mannosidase II. Cells can be selected for or against the ability to synthesize MPD depending on whether Sw, PHA, and/or conA are added (29). To aid the reader, these selections are demonstrated with normal CHO cells and Lec15 mutants in Fig. 1. These methods are also applicable to other mutations resulting in Man 5 GlcNAc 2 -PP-dolichol accumulation, such as Lec9 and Lec35 (20). To simulate the anticipated properties of Lec15/Lec1 double mutants, which were expected to interfere with expression cloning, Fig. 1 also includes a Lec35/Lec1 double mutant (19).
Colonies resistant to PHA/Sw were readily isolated after transfection of the CHO-K1 library into ptLec15. These could have been (i) positive transfectants, i.e. Lec15 cells carrying MPDS cDNA or a suppressor cDNA; (ii) genotypic revertants of the Lec15 mutation; or (iii) Lec15/Lec1 double mutants. Indeed, the PHA/Sw-resistant population included a large fraction of presumed Lec15/Lec1 double mutants which were routinely detected and removed by counter-selection with 10 g/ml conA. Analysis of dolichol-linked oligosaccharides (data not shown) in surviving cells indicated a phenotypic correction, consistent with possibilities i or ii. Plasmids were recovered by the Hirt procedure (23) and analyzed after electroporation into Escherichia coli MC1061/P3 cells.
The Hirt DNA and plasmid preparations from individual E. coli colonies all contained the same cDNA species, eliminating the need for further rounds of enrichment or "sib" selection. The plasmid carrying the cDNA, termed pLW1, efficiently corrected the Lec15 phenotype (see below) whereas the vector did not, demonstrating that possibility i was correct. The functional insert cDNA of pLW1 was 1.3 kb and designated SL15 for suppressor of Lec15. It was noted, however, that pLW1 had a tendency to rearrange in E. coli (data not shown). The transfected E. coli MC1061/P3 colonies had plasmids with inserts of 1.3 kb and 1.6 kb in various ratios. By isolating and reintroducing these plasmids into E. coli MC1061/P3 and ptLec15, it was determined that the 1.3-kb insert, but not the 1.6-kb insert, corrected the Lec15 phenotype and that the 1.3-kb insert rearranged irreversibly into the 1.6-kb species. Use of E. coli DH10B/P3 (Life Technologies, Inc.), a strain similar to MC1061/P3 but lacking various genes associated with DNA restriction and rearrangement, gave the same results as E. coli MC1061/P3 (data not shown).
Biochemical Effects of SL15 Expression in Lec15 Cells-pLW1 corrected the accumulation of Man 5 GlcNAc 2 -PP-dolichol in ptLec15 cells (Fig. 2, A-C). The predominant oligosaccharides (C) were all of the Glc 0 -3 Man 9 GlcNAc 2 type. Expression of MPDS activity was examined in vitro with microsomal membranes as shown in Fig. 3. MPDS activities in pLW1-transfected ptLec15 cells were essentially normal and greatly in- FIG. 1. Selections used to study mannosylation mutants. 200 CHO-K1 cells of the indicated type (normal, ptLec15.2D6, Lec35/Lec1 (19)) were plated in wells of a 24-well plate with 1.5 ml of Ham's F-12 medium buffered at pH 7.2 with 20 mM Na-HEPES, serum (10% fetal bovine serum for conA and controls, 3% calf serum for PHA), 100 units/ml penicillin, and 100 g/ml streptomycin plus, where indicated, 1 g/ml swainsonine (Sw) and/or 10 g/ml concanavalin A (conA) or 20 g/ml phytohemagglutinin E (PHA). Colonies were stained after 7 days of growth at 37°C in a humidified 5% CO 2 atmosphere. Lec35/Lec1 cells were expected to behave the same as Lec15/Lec1 cells in these selections. Suppressor of Lec15 and Lec35 Cells 13936 creased over untransfected ptLec15 cells. In addition, the principal lipid product made by membranes from pLW1-transfected ptLec15 cells (Fig. 4, lane 3) comigrated with MPD from normal membranes (Fig. 4, lane 1). None of the missing MPDS activity of ptLec15 membranes was restored by mixing with T-ptLec15 membranes (Fig. 3), indicating that SL15 did not behave as a soluble trans-acting factor. When assayed under conditions similar to those in Fig. 3, the GPD and GnPPD synthases were not significantly altered in T-ptLec15 membranes compared with control or ptLec15 membranes (data not shown). This indicated that SL15 was not a general transferase activator or a promiscuous transferase.
Suppression of the Lec35 Mutation-Lec35 CHO-K1 mutants synthesize normal levels of MPD in vivo and in vitro, but fail to utilize it efficiently in vivo (29,30). Thus, Lec35 cells accumulate both Man 5 GlcNAc 2 -PP-dolichol and GlcN-(acyl)PI (16). However, these molecules are readily mannosylated in vitro in Lec35 membranes. This has led to the idea that MPD may fail to be flipped across the ER membrane or be transported efficiently in Lec35 cells, such that the mechanical disruption associated with membrane isolation could weaken physical barriers and enhance mobility of MPD.

for examining the formation of [ 3 H]Man-GlcN-(acyl)PI in Lec35 membranes from endogenous GlcN-(acyl)PI, with [ 3 H]MPD generated with exogenous GDP-[ 3 H]mannose. Lec35 cells failed to mannosylate GlcN-(acyl)PI in vivo and yielded membranes which made abundant amounts of [ 3 H]Man-GlcN-(acyl)PI in vitro
Sequence Analysis-The effects of SL15 in Lec15 and Lec35 cells could be explained if MPD synthesis was increased in vivo, correcting the Lec15 defect directly and overriding the defect in Lec35 by mass action. It was therefore of interest to determine whether the SL15 protein resembled the MPDS from S. cerevisiae encoded by the DPM1 gene. As shown in Fig. 5, the SL15 cDNA had 15 nt of 5Ј-untranslated sequence, 526 nt of nonpoly(A) 3Ј-untranslated sequence, and 744 nt of coding region preceded by a pentanucleotide (CCAAC) in excellent agreement with the consensus for eukaryotic translation initiation sites (31). The SL15 protein is predicted to have 248 amino acid residues (43% hydrophobic) with a molecular weight of 26,693  (29), except that the volume was 0.2 ml, the incubation time was 10 min, and no greater than 10 g of membrane protein was used per assay. The 10-min time point was in the linear range (data not shown). CHO-K1, ptLec15, and T-ptLec15 microsomes were assayed alone or in combination in the quantities indicated. Each bar represents the mean of 6 independent determinations Ϯ S.E. Assays were performed with either endogenous dolichol-P alone (upper panel) or supplemented with 0.4 g/assay exogenous dolichol-P (lower panel), which was shown to be saturating (data not shown). A 0.7 mg/ml stock of dolichol-P (Sigma) was prepared in 0.35% (w/v) Nonidet P-40 and added to reaction mixtures containing microsomes. The resulting amount of Nonidet P-40 (0.001%) by itself did not affect MPDS activity with endogenous dolichol-P (data not shown).

FIG. 4. Thin-layer chromatography.
Samples were obtained in assays performed as described in Fig. 3 (endogenous dolichol-P), except that incubations were 30 min. Lanes are labeled as in Fig. 2. TLC was performed as described (16), and lipids were identified by autoradiography. The upper and lower arrows indicate mannose-P-dolichol and mannosyl-glucosaminyl-acylphosphatidylinositol (16), respectively.

Suppressor of Lec15 and Lec35 Cells 13937
and a pI of 8.65. The amino-terminal sequence contains a segment of 8 apolar residues (residues 12-19) preceded by 2 positively charged residues (residues 10 and 11), a common feature of cleaved eukaryotic leader sequences (32). However, since the 8 residues following the apolar segment do not conform to the "-1,-3" rule (32), it is doubtful that this is a true cleaved leader sequence. There are two hydrophobic segments in the SL15 protein long enough to span the membrane, from residues 130 to 157 and residues 211 to 236. Applying the rules of Hartmann et al. (33), a speculative structure for the SL15 protein would have cytosolic amino and carboxyl termini with the two membrane spanning regions connected by a lumenal loop (Fig. 6). In accord with this model, there are lysines at positions 245-246, in excellent agreement with the consensus for a class of endoplasmic reticulum retention signals which are oriented toward the cytoplasm (34). An attractive feature of this model is that the bulk of the SL15 protein would face the cytosol, similar to the orientation proposed for S. cerevisiae MPDS (35) and other early reactions of oligosaccharide synthesis (36). A BLAST search of combined nucleic acid data bases revealed a number of human EST sequences with apparent homology to the SL15 cDNA (accession numbers H27246, H51950, H77811, R05987, R48750, T16638, T81897, T81898) in GenBank release 92. Furthermore, a BLAST search of combined peptide data bases revealed excellent similarity between the SL15 protein and the gene F38E1.9 protein from chromosome 5 of C. elegans (GenBank TM release 92), shown in Fig. 7. The function of F38E1.9 is currently unknown. There were no other reasonable similarities between the SL15 DNA or protein and other entries in current data bases (data not shown). Most notably, there was no significant similarity between the SL15 protein and S. cerevisiae MPDS by using Bestfit or Gap in the Wisconsin Genetics Computer Group package or by performing various homology and alignment searches with Microgenie Version 7 (Beckman Corp.). Only weak similarities were identified that appeared to be no better than similarities found by deliberately comparing the wrong reading frames of DPM1 and SL15 (data not shown).
These results raise several possibilities for SL15 protein function, including the functional hamster equivalent of yeast MPDS, a distinct MPDS, an activator of MPDS, or a protein that stimulates a distinct enzyme to mimic MPDS. It is not yet known whether SL15 is mutated in Lec15 cells or whether Lec15 cells lack the actual MPDS enzyme. In any case, these results suggest that MPD synthesis may be more complex than previously thought.

Felecia E. Ware and Mark A. Lehrman
This paper reported the cloning of a cDNA for a new gene, which we termed SL15. We reported that SL15 cDNA corrected two separate glycosylation-defective cell lines known as Lec15 and Lec35. In experiments to be reported in the future we have found that the Lec35 cells used in that study have a defect in the SL15 gene. This provides a clear explanation for correction of Lec35 cells by SL15 cDNA, a result that has now been obtained by three separate individuals in this laboratory as well as by another laboratory (T. Kinoshita, personal communication). However, similar attempts to repeat the correction of Lec15 cells by SL15 cDNA have consistently failed. Based upon further analysis of cell samples from that study, we now conclude that the Lec15 population was contaminated with Lec35 cells. Lec15 cells described in Figs. 2, 3, and 4 as being transfected with SL15 cDNA and having normal glycosylation phenotypes have now been shown to be Lec15-Lec35 hybrids, most likely formed during electroporation. It is likely that SL15 cDNA was carried by contaminating Lec35 cells during expression cloning and that the contaminating Lec35 cells accounted for the greater number of colonies that survived lectin selection after transfection with SL15 cDNA as compared with vector controls.
Since there is no evidence that SL15 cDNA can correct the Lec15 phenotype certain aspects of our paper must be rescinded, specifically Fig. 2, panel C; Fig. 3, all data with T-ptLec15 microsomes; Fig. 4, lane 3; and all sections of the text indicating that SL15 can correct Lec15 cells. However, we remain confident that the correction of Lec35 cells by SL15 cDNA is a valid result. We sincerely apologize to the readers of this journal for any confusion or inconvenience this may have caused.

Myounghee Yu, Lilin Zhong, Arun K. Rishi, Mohammed Khadeer, Giuseppe Inesi, and Arif Hussain
Dr. Zhong's name was misspelled. It is correct in the author line above.
Pages 7597 and 7600: Due to an error in assembling Figs. 1F and 7C, the two D28 blot lanes in Fig. 1F were mounted upside down and Fig. 7C was mistaken as LOX melanoma seprase but it was actually derived from placental seprase data. The former had been shown previously (Monsky, W. L., Lin, C.-Y., Aoyama, A., Kelly, T., Mueller, S. C., Akiyama, S. K., and Chen, W.-T. (1994) Cancer Res. 54, 5702-5710), and the latter should be referenced with similar figures derived from two LOX seprase experiments done on July 9, 1992 and January 15,1993. In all of these changes, there is no new scientific information added or changed, and there is no change in the conclusions of the paper.
We suggest that subscribers photocopy these corrections and insert the photocopies at the appropriate places where the article to be corrected originally appeared. Authors are urged to introduce these corrections into any reprints they distribute. Secondary (abstract) services are urged to carry notice of these corrections as prominently as they carried the original abstracts.