Báo cáo khoa học: Comparative importance in vivo of conserved glutamate residues in the EX7E motif retaining glycosyltransferase Gpi3p, the UDP-GlcNAc-binding subunit of the first enzyme in glycosylphosphatidylinositol assembly doc

Comparative importance in vivo of conserved glutamate residues in the EX 7 E motif retaining glycosyltransferase Gpi3p, the UDP-GlcNAc-binding subunit of the first enzyme in glycosylphosphatidylinositol assembly Zlatka Kostova 1 , Benjamin C. Yan 1 , Saulius Vainauskas 2 , Roberta Schwartz 2 , Anant K. Menon 2 and Peter Orlean 1 1 Department of Biochemistry, University of Illinois at Urbana-Champaign, Urbana, IL, USA; 2 Department of Biochemistry, University of Wisconsin-Madison, Madison, WI, USA Saccharomyces cerevisiae Gpi3p is the UDP-GlcNAc-bind- ing and presumed catalytic subunit of the enzyme that forms GlcNAc-phosphatidylinositol in glycosylphosphatidylinosi- tol biosynthesis. It is an essential protein with an EX 7 Emotif that is conserved in four families of retaining glycosyl- transferases. All Gpi3ps contain a cysteine residue four residues C-terminal to EX 7 E. To test their importance for Gpi3p function in vivo, Glu289 and 297 in the EX 7 E motif of S. cerevisiae Gpi3p, as well as Cys301, were altered by site- specific mutagenesis, and the mutant proteins tested for their ability to complement nonviable GPI3-deleted haploids. Gpi3p-C301A supported growth but membranes from C301A-expressing cells had low in vitro N-acetylglucosami- nylphosphatidylinositol (GlcNAc-PI) synthetic activity. Haploids harboring Gpi3p-E289A proved viable, although slow growing but Gpi3-E297A did not support growth. The E289D and E297D mutants both supported growth at 25 °C, but, whereas the E289D strain grew at 37 °C, the E297D mutant did not. Membranes from E289D mutants had severely reduced in vitro GlcNAc-PI synthetic activity and E297D membranes had none. The mutation of the first GluintheEX 7 E motif of Schizosaccharomyces pombe Gpi3p (Glu277) to Asp complemented the lethal null mutation in gpi3 + and supported growth at 37 °C, but the E285D mutant was nonviable. Our results suggest that the second Glu residue of the EX 7 E motif in Gpi3p is of greater importance than the first for function in vivo. Further, our findings do not support previous suggestions that the first Glu of an EX 7 E protein is the nucleophile and that Cys301 has an important role in UDP-GlcNAc binding by Gpi3ps. Keywords: endoplasmic reticulum; glycosylphosphatidyl- inositol; glycosyltransferase; Saccharomyces cerevisiae; Schizosaccharomyces pombe. Glycosyltransferases can be classified into a range of families based on amino-acid sequence similarities, and these sequence alignments have led to the identification of signature motifs of amino acids [1–3]. Members of the large Pfam GT1F glycosyltransferase family, with representatives in the bacteria, archaea, and eukaryotes, have in turn been classified into subfamilies. Many Pfam GT1F glycosyl- transferases fall into family 4 of the classification proposed by Campbell and coworkers (CaZY) [2], and some into families 3 and 5. Most of these retaining glycosyltransferases have the signature motif EX 7 E. The conservation of these two acidic residues strongly suggests that they have key roles in glycosyltransferase activity, and this has been demon- strated in site-specific mutagenesis and in vitro activity assay studies of CaZY family 3 human muscle glycogen synthase and the family 4 a-mannosyltransferase from Acetobacter xylinum (AceA) [4,5]. The two glutamate residues have been proposed to be involved in catalysis, but their contributions have yet to be evaluated by 3D structural analysis or identification of enzyme–substrate complexes or reaction intermediates. Whereas Kaptinov & Yu [3] suggested that the second of the two Glu residues in the EX 7 Emotifmayserveasa nucleophile, and the first as an acid base catalyst, the results of site-directed mutagenesis studies of human muscle glycogen synthase and AceA indicated that, in both cases, mutations of the first glutamate had more severe effects on in vitro enzyme activity and on the ability of the mutant enzyme to catalyze its glycosyltransfer reaction when expressed in a heterologous system. These findings indicated that the first of the Glu residues is critical for enzyme activity, possibly as the nucleophile [4,5]. We showed recently that a yeast EX 7 E motif protein, Gpi3p, binds a photoactivatable sugar nucleotide analogue, consistent with its function as the substrate-binding and catalytic subunit of the enzyme complex that forms N-acetylglucosaminylphosphatidylinositol (GlcNAc-PI) in the first reaction of the pathway for glycosylphosphatidyl- Correspondence to P.Orlean, Department of Microbiology, University of Illinois at Urbana-Champaign, 309 Roger Adams Laboratory, 600 South Mathews Avenue, Urbana, IL 61801, USA. Fax: + 1 217 244 5858, Tel.: + 1 217 333 4139, E-mail: p-orlean@uiuc.edu Abbreviations: AceA, a-mannosyltransferase from Acetobacter xylinum; GPI, glycosylphosphatidylinositol; GlcNAc-PI, N-acetylglucosaminylphosphatidylinositol. (Received 26 July 2003, revised 11 September 2003, accepted 19 September 2003) Eur. J. Biochem. 270, 4507–4514 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03844.x inositol (GPI) biosynthesis [6]. As Gpi3p is probably a glycosyltransferase, and because it is encoded by an essential gene in both Saccharomyces cerevisiae and, as we report here, Schizosaccharomyces pombe, it presents an excellent model with which to assess the effects of amino-acid replacements in the EX 7 E motif by testing mutated forms of Gpi3p for their ability to complement lethal null mutations in GPI3. The results of our studies indicate that both Glu residues are important for function, but that the second Glu in the EX 7 E motif is less tolerant of changes to other amino acids, and therefore comparatively more important for enzyme function. Gpi3 proteins from various organisms also have a conserved cysteine (Cys301 in S. cerevisiae Gpi3p). On account of its proximity to the proposed catalytic EX 7 E motif, and because GlcNAc-PI synthetic activity can be inhibited irreversibly by agents that alkylate thiol groups but protected from inhibition by uridine nucleotide compounds [7], it has been speculated that this cysteine is important for function, perhaps for UDP-GlcNAc binding [7,8]. Alter- ation of Cys301 to Ala has no obvious effect on the mutant protein’s ability to support cell growth, but membranes harboring the mutant protein had significantly lower GPI GlcNAc transferase activity than wild-type membranes. The results of our in vivo tests for Gpi3p function do not support previous suggestions that the first Glu of an EX 7 E glycosyltransferase is the nucleophile in the reactions cata- lyzed by members of this protein family or that Cys301 is involved in UDP-GlcNAc binding by Gpi3 proteins. Materials and methods Materials UDP-[U- 14 C]GlcNAc (specific radioactivity, 283 mCiÆ mmol )1 ) was purchased from NEN Life Science Products (Boston, MA, USA). Palmitoyl-CoA and tunicamycin were obtained from Sigma, and Nikkomycin-Z from Calbio- chem. Silica gel 60 TLC plates were supplied by Altech (Deerfield, IL, USA). X-OMAT X-ray film and Transcreen- LE intensifying screens were from Eastman-Kodak Com- pany (Rochester, NY, USA). Expand High Fidelity PCR mix, Pwo polymerase and EDTA-free protease inhibitor tablets were purchased from Roche Diagnostics (Indiana- polis, IN, USA). Geneticin (G418), Taq polymerase, and the restriction endonucleases were obtained from Gibco-BRL (now Invitrogen, Carlsbad, CA, USA), and DpnI from Stratagene (La Jolla, CA, USA). Oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA, USA), and DNA sequencing was performed at the University of Illinois Genetic Engineering and Sequencing Facility or at the University of Wisconsin-Madison Bio- technology Center. Yeast strains and culture media The temperature-sensitive S. cerevisiae gpi3-15C strain had the genotype MAT a, ade2, leu2-3,112, trp1-1, his3-11,15 [9]. Diploid strain YMW3 (MATa/a, ade2-1/ade2-1, ade3D22/ ade3D22, his3-11,15/his3-11,15, leu2-3,112/leu2-3,112, trp1- 1/trp1-1, ura3-1/ura3-1, can1-100/can1-100) is described in reference [10], and construction of the heterozygous GPI3/gpi3::kanMX4 diploid derived from YMW3 is detailed in reference [6]. Schizosaccharomyces pombe strains were derived from the wild-type heterothallic strains KGY246 (h – , ade6-M210, ura4-d18, leu1-32) and KGY249 (h + , ade6-M216, ura4-d18, leu1-32). YPD and SD media were prepared as described in reference [11], and EMM2 medium is described in reference [12]. The presence of the kanMX4 marker was verified by scoring for resistance to 200 lgG418ÆmL )1 on solid YPD medium. Expression and mutagenesis of S. cerevisiae GPI3 A 2624-bp XhoI–SacI fragment of S. cerevisiae genomic DNA, which contained the GPI3 gene, an additional 674 bp DNA containing the native GPI3 promoter at the gene’s 5¢ end, and 494 bp 3¢ flanking DNA, was cloned into the centromeric and 2l plasmids pRS415 and pRS425 [13]. The resulting plasmids, pRS415-GPI3 and pRS425-GPI3,were used as templates for mutagenesis and for expression of the GPI3 mutants in S. cerevisiae. The following mutations were made: E289A, E297A, E289D, E297D, E289G, E297G, E289D/E297D and C301A. Mutagenesis of GPI3 was performed using the Stratagene QuikChange site-directed mutagenesis strategy. For each mutation to be introduced, a mutagenic oligo- nucleotide and its inverse complement were designed that introduced the appropriate nucleotide changes and a diagnostic restriction site in the middle of the oligonucleo- tides. DNA amplification by PCR was carried out using Pwo polymerase. Potential mutagenized plasmids were identified by digestion with a restriction enzyme specific for the introduced site, and the GPI3 region on selected plasmids was sequenced to verify the presence of the desired mutation, and the absence of mutations introducing any further amino-acid changes. To make the E289D/E297D double mutant, 2l plasmids with each single mutation were mutagenized a second time using the oligonucleotide pairs designed to introduce the additional mutation. Double mutants were obtained with each starting mutant plasmid, and the correctness of the mutations and GPI3 sequence was confirmed by DNA sequencing. Cloning, disruption and site-directed mutagenesis of Sz. pombe gpi3 + A BLAST search [14] using the amino-acid sequence of S. cerevisiae Gpi3p as query identified Sz. pombe ORF SPBC3D5. This putative Sz. pombe gpi3 + gene contains four introns, and sequencing of a gpi3 + cDNA amplified from an Sz. pombe cDNA library [15] confirmed that the four introns are spliced as predicted. To disrupt the gpi3 + gene, DNA fragments of  1kb each of chromosomal DNA that flanks the 5¢ and 3¢ ends of the gpi3 + locus were amplified by PCR, and the PCR- amplified Sz. pombe ura4 + gene was cloned between the two gpi3 + -flanking fragments. The resulting 3.5-kb frag- ment, in which 88% of the gpi3 + sequence was replaced by ura4 + DNA, was used to transform an adenine-proto- trophic diploid created by mating haploid strains KGY246 and KGY249 to uracil prototrophy. Stable diploids were selected, and the presence of the disrupting fragment at the 4508 Z. Kostova et al.(Eur. J. Biochem. 270) Ó FEBS 2003 chromosomal gpi3 + locus verified by whole-cell PCR. Diploids were allowed to sporulate, and random spore analysis was carried out on EMM medium supplemented with limiting adenine but selective for uracil prototrophy to identify potential gpi3 + ::ura4 + haploids. Tetrad analysis was also carried out on asci derived from two independent gpi3 + /gpi3 + ::ura4 + diploids, and viable ade – haploid segregants were scored for uracil prototrophy. Genomic Sz. pombe DNA consisting of the gpi3 + locus and about 700 bp 5¢ flanking DNA and 1000 bp 3¢ flanking DNA was cloned into the LEU2-marked Sz. pombe expression vector pSP1 [16]. Diploids transformed with this plasmid, pSP1-gpi3 + , yielded ade – , uracil and leucine prototrophic haploids upon sporulation, indicating that the cloned gpi3 + gene complemented the gpi3 + ::ura4 + disruption. Plasmid pSP1-gpi3 + was used as template for site-directed mutagenesis of Glu277 and Glu285 to Asp as detailed for S. cerevisiae Gpi3p above, and the presence of the desired mutation, and the absence of mutations introducing any further amino-acid changes, were verified by DNA sequencing. Assay of GlcNAc-PI synthesis Washed mixed membranes were prepared and assayed for in vitro GlcNAc-PI synthetic activity as described previously [6, 17]. In assays to estimate the formation of [ 14 C]GlcNAc- PI with time, palmitoyl-CoA was omitted from the incuba- tion mixtures. Radiolabeled lipids were extracted, separated by TLC, and detected by fluorography. The chromato- grams were scanned by Phosphorimager to determine the relative amounts of 14 C in the GlcNAc-PI in each sample. Imaging and microscopy Images of yeast growth on solid YPD medium were obtained using a Bio-Rad Gel-Doc2000. Growth of individual colonies arising from spores that had germinated on solid YPD medium was monitored using a Nikon TE300 inverted microscope with a 40 · bright field objective. Results The importance of the conserved Glu residues in the EX 7 E motif of S. cerevisiae Gpi3p for in vivo function was tested by introducing mutations into the GPI3 gene that altered these residues, Glu289 and Glu297, to aspartates, glycines or alanines. The mutated genes were in turn introduced into a heterozygous GPI3/gpi3::kanMX4 diploid on low or high copy plasmids, the diploids induced to undergo meiosis and sporulation, and the resulting asci dissected to assess whether the mutated GPI3 gene permitted growth of otherwise nonviable haploid gpi3::kanMX4 segregants. The consequences of changing the Glu residues in the EX 7 E motif of Sz. pombe Gpi3p to Asp residues were exam- ined analogously. The C301A mutation in S. cerevisiae Gpi3p was also tested. E289A and E297A mutants in S. cerevisiae Gpi3p Tetrads arising from GPI3/gpi3::kanMX4 diploids transformed with centromeric or 2l plasmids expressing Gpi3p-E289A gave rise to two fast-growing segregants. However, after 5–6 days of incubation on YPD medium at 25 °C, many of the dissected tetrads yielded additional microcolonies, and a number of complete tetrads with two large colonies and two microcolonies were observed (Fig. 1A). Representative segregants that subsequently formed normal sized or microcolonies were examined by microscopy at intervals over several days, and these inspections confirmed that the segregants yielding micro- colonies were slow growing (Fig. 1B). Changing the first Glu of the EX 7 E motif of Gpi3p to Ala is therefore not lethal, although the mutation affects in vivo function, leading to a severe growth defect. The E297A mutation, however, abolishes in vivo function. Tetrads from GPI3/gpi3::kanMX4 diploids transformed with low or high copy plasmids expressing Gpi3p-E297A contained only two normally growing segregants, and two that germinated and accomplished two or three cell divisions, but which did not continue to grow (Fig. 1A). All segregants giving rise to normal sized colonies were G418-sensitive, indicating that they contained the chromo- somal wild-type GPI3 gene. The slow growing or nonviable segregants in each tetrad were inferred to contain the gpi3::kanMX4 allele. In the case of slow growing segregants from the diploid transformed with plasmids expressing Gpi3p-E289A, this could be confirmed: cells from the microcolonies grew when restreaked on to G418-containing medium. The gpi3::kanMX4-Gpi3p-E289A segregants retained their slow growth phenotype when restreaked on to fresh YPD medium (Fig. 1C), but grew slightly better on YPD medium containing 0.6 M KCl, indicating partial relief of a cell wall defect. Neither the E289G nor E297G mutation supported growth of gpi3::kanMX4. E289D and E297D mutants in S. cerevisiae Gpi3p The EX 7 E Glu residues were changed to Asp and tested individually and in combination. Tetrads from GPI3/ gpi3::kanMX4 diploids transformed with centromeric or 2l plasmids expressing Gpi3p-E289D or Gpi3p-E297D gave rise to segregants that all grew at approximately compar- able rates at 25 °C (Fig. 2A). However, whereas the gpi3 ::kanMX4-Gpi3p-E289D segregants grew at 37 °Cwhen expressed from high or low copy plasmids, the gpi3::kan- MX4-Gpi3p-E297D segregants failed to grow at 37 °C, even when expressed on a 2l plasmid (Fig. 2A). Consistent with these results, Gpi3p-E289D restored ability of the tempera- ture-sensitive gpi3-15C strain [9] to grow at 37 °C, whereas E297D did not (not shown). The Gpi3p-E289D/E297D double mutant did not support growth of gpi3::kanMX4. The E289D and E297D mutations affected the in vitro transfer of [ 14 C]GlcNAc from UDP-[ 14 C]GlcNAc to endo- genous PI, but in different ways, with E289D having the less severe effect. Membranes from the two gpi3::kanMX4 segregants from a tetrad arising from a GPI3/gpi3::kanMX4 diploid transformed with low or high copy plasmids expres- sing Gpi3p-E289D retained the ability to synthesize Glc- NAc-PI, although at much lower levels than membranes from the wild-type siblings (Fig. 2B). The gpi3::kanMX4 segregants harboring Gpi3p-E297D had no detectable in vitro GlcNAc-PI synthetic activity. The copy number of the expression plasmid did not influence the in vitro GlcNAc-PI Ó FEBS 2003 EX 7 E motif in Gpi3p GlcNAc transferase subunit (Eur. J. Biochem. 270) 4509 synthetic activities. When plasmids expressing Gpi3-E297D were introduced into the temperature-sensitive gpi3-15C strain, weak restoration of in vitro GlcNAc-PI synthetic activity wasobtained (not shown), suggesting that the E297D mutation does not act as a dominant negative mutant. Thedifferenceinthein vitro GlcNAc-PI synthetic activities of wild-type and gpi3::kanMX4-Gpi3p-E289D membranes was quantified by incubating assays for shorter times, separating the [ 14 C]GlcNAc-PI formed by TLC, and estimating the amount of 14 C in the reaction product using a Phosphorimager. The initial rate of [ 14 C]GlcNAc-PI for- mation by wild-type membranes, estimated from the progress curves in Fig. 2C, is some 12-fold higher than the rate at which [ 14 C]GlcNAc-PI is formed by gpi3::kanMX4- Gpi3p-E289D membranes. E277D and E285D mutants in Sz. pombe Gpi3p The findings that the E289A and E289D mutations have a less severe effect on Gpi3p function than the E297A and E297D mutations prompted us to test whether the same trend holds for the corresponding Glu residues in another Gpi3 protein, the Gpi3p homologue from fission yeast. We cloned an Sz. pombe ORF encoding a protein of 456 amino acids with 52% identity with and 73% similarity to S. cerevisiae Gpi3p. This gene, which we designate gpi3 + , was disrupted by replacing 88% of the coding region of one gpi3 + allele in a wild-type diploid strain with the ura4 + gene. The resulting heterozygous diploid was induced to sporulate, and the sporulating diploid submitted to both random spore and tetrad analysis. No viable, uracil prototrophic haploids were recovered, indicating that disruption of gpi3 + is lethal. This lethality was due to disruption of the gpi3 + gene because viable gpi3 + ::ura4 + haploids were recovered from sporulated gpi3 + /gpi3 + ::ura4 + diploid harboring the gpi3 + gene on plasmid pSP1. Plasmids encoding the Gpi3p-E277D and Gpi3p-E285D mutations were introduced into heterozygous gpi3 + / gpi3 + ::ura4 + diploids, which were sporulated, and the meiotic segregants then submitted to random spore ana- lysis. Haploid uracil prototrophs were recovered from sporulated gpi3 + /gpi3 + ::ura4 + diploids that harbored the Fig. 1. Growth of meiotic segregants from GPI3/gpi3::kanMX4 diploids transformed with plasmids expressing Gpi3p-E289A and Gpi3p- E297A. (A) Six-day growth at 25 °Cof segregants dissected on to YPD agar. The Gpi3p-E289A protein was expressed on the centromeric plasmid pRS415 and the Gpi3p- E297A protein on the 2l plasmid pRS425. (B) Microscopic examination of GPI3 and gpi3::kanMX4-pRS425-GPI3-E289A segre- gants at intervals after micromanipulation of ascospores on to YPD agar. (C) Growth of restreaked segregants from a representative tetrad from a GPI3/gpi3::kanMX4 diploid transformed with pRS425-GPI3-E289A at 25 °C on YPD medium or YPD medium supplemented with 0.6 M KCl. 4510 Z. Kostova et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Gpi3p-E277D-expressing plasmid, and these complemented disruptants grew as well as gpi3 + haploids at 37 °C. In contrast, no viable gpi3 + ::ura4 + haploids were recovered from sporulated gpi3 + /gpi3 + ::ura4 + diploids expressing Gpi3p-E285D, even when the sporulated diploids were plated on selective medium supplemented with high con- centrations of salt, glucose, glycerol, or sorbitol, and incubated at lower temperatures. These results, which indicate that Sz. pombe Gpi3p cannot tolerate the conservative Glu to Asp substitution in the second of the two Glu residues of its EX 7 Emotif,are consistent with those obtained with S. cerevisiae Gpi3p, although the effect on Sz. pombe Gpi3p is more severe. C301A mutant in S. cerevisiae Gpi3p Tetrads from GPI3/gpi3::kanMX4 diploids transformed with centromeric or 2l plasmids expressing Gpi3p-C301A gave rise, in most cases, to three or four viable segregants that all grew at approximately comparable rates at 25 °C and 37 °C, indicating that Gpi3p-C301A can complement the lethal gpi3::kanMX4 mutation (Fig. 3A). Moreover, introduction of a plasmid-borne copy of Gpi3p-C301A restored the ability of a temperature-sensitive gpi3 strain to grow at 37 °C (Fig. 3B). Alteration of Cys301 to Ala therefore has no obvious effect on the protein’s ability to support growth. The C301A mutation did, however, lower in vitro GlcNAc-PI synthetic activity: mixed membranes from a gpi3::kanMX4 segregant from a tetrad arising from a GPI3/gpi3::kanMX4 diploid transformed with a 2l plasmid expressing Gpi3p-C301A had about 20% of the in vitro GlcNAc-PI synthetic activity of membranes from a wild- type sibling (Fig. 3C,D). Discussion The EX 7 E motif is conserved among the members of four families of retaining glycosyltransferases, suggesting key roles for the two Glu residues in enzyme function. We exploited the fact that S. cerevisiae GPI3 and Sz. pombe gpi3 + are essential genes to test the importance of the conserved Glu residues in the EX 7 E motif of these proteins by the stringent criterion of their ability to support cell growth. Our results indicate that both Glu residues are important for function, but that the second one in the EX 7 E motif is less tolerant of change to other amino acids, and therefore is most critical for enzyme function in vivo.Our Fig. 2. Growth and in vitro GlcNAc-PI synthetic activity of meiotic segregants from GPI3/gpi3::kanMX4 diploids transformed with plas- mids expressing Gpi3p-E289D and Gpi3p-E297D. (A) Drops of liquid containing dilute suspensions of cells of the four viable segregants from six asci were placed on YPD agar and plates were incubated at 25 °C or 37 °C. (B) In vitro synthesis of [ 14 C]GlcNAc-containing lipids by membranes from the haploid wild-type (WT) and gpi3::kanMX4 (Dgpi3) segregants from a diploid transformed with pRS425-Gpi3p- E289D or pRS425-Gpi3p-E297D. Mixed membranes were prepared and incubated with UDP-[ 14 C]GlcNAc, and 14 C-labeled GlcNAc-PI, GlcN-PI, and GlcN-(acyl-Ins)PI were extracted, separated by TLC, and detected by fluorography. Identities of the 14 C-labeled lipids were assigned as previously [23]. (C) Mixed membranes were prepared from a wild-type (WT) and two gpi3::kanMX4 segregants harboring pRS425-Gpi3p-E289D and incubated with UDP-[ 14 C]GlcNAc for different times, after which radiolabeled lipids were extracted, separ- ated by TLC, and detected by fluorography. Amounts of 14 Csignalin each sample migrating at the position of [ 14 C]GlcNAc-PI on the chromatogram were quantified using a Phosphorimager and plotted as relative intensities. (s) Mean of Phosphorimager signals from [ 14 C]GlcNAc-PI formed by membranes from two gpi3::kanMX4 seg- regants containing Gpi3p-E289D. Ó FEBS 2003 EX 7 E motif in Gpi3p GlcNAc transferase subunit (Eur. J. Biochem. 270) 4511 in vivo findings do not support previous suggestions that the first Glu of an EX 7 E motif protein is the nucleophile in the reactions catalyzed by members of this protein family and that Cys301 has an important role in UDP-GlcNAc binding by Gpi3ps. Changing the first Glu of S. cerevisiae Gpi3p to Ala was not lethal, whereas the E297A change was. The E289A mutant presumably retains a level of function in vivo that allows it to support growth, albeit weakly. Conservative changes of the EX 7 E Glu residues to aspartates were much less deleterious to S. cerevisiae Gpi3p: haploid segregants harboring the E289D and E297D mutations grew about as well as their wild-type siblings at 25 °C, consistent with the importance of an acidic side chain at both positions in the protein. However, by two criteria, the change of Glu297 to Asp had a more severe effect on Gpi3p function. First, haploids complemented by Gpi3p-E289D grew at 37 °C, but the E297D-expressing strains were temperature-sensi- tive. Secondly, membranes containing Gpi3p-E297D had no detectable GlcNAc-PI synthetic activity, whereas those containing Gpi3p-E289D retained in vitro activity at about one twelfth the level seen with wild-type membranes. The differential effects of the glutamate to aspartate mutations in Sz. pombe Gpi3p highlighted the greater relative import- ance of the second glutamate: the E285D mutation was lethal, whereas the E277D mutation had no discernible effect in vivo. A potential concern with site-directed mutagenesis approaches is that the mutations introduced in the test protein may affect the protein’s structure, localization, or its ability to participate in a complex, and so may only indirectly affect enzyme function. However, our genetic data showing retention of function of various key mutants in vivo render a demonstration of misfolding, instability, or mislo- calization of the protein, or of impaired complex formation by the protein, redundant, for such additional findings could not alter – and would have little bearing – on our conclusions. Thus, for example, mutation of E289 in Gpi3p would be expected to yield a nonfunctional protein accord- ing to current models [4,5]. In this event, to make sure that the point mutation was the sole cause of nonfunctionality, we would be obliged to investigate expression level of the protein, its ability to form a complex with other GlcNAc-PI synthase subunits, and its subcellular localization. None of thesetestsarenecessarybecauseweshowthattheE289A and E289D mutants function in vivo and are able to sustain cell growth, albeit weakly in the case of E289A. Indeed, were the effects of the E289A mutation to be indirect ones on protein folding, stability, or localization, or on the ability of Fig. 3. Growth and in vitro GlcNAc-PI synthetic activity of meiotic segregants from GPI3/gpi3::kanMX4 diploids transformed with plas- mids expressing Gpi3p-C301A. (A) Colonies arising upon germination of spores from asci formed from GPI3/gpi3::kanMX4 diploids trans- formed with pRS415-Gpi3p-C301A or pRS425-Gpi3p-C301A. (B) Ability of pRS415-Gpi3p-C301A and pRS425-Gpi3p-C301A to restore the ability of the temperature-sensitive gpi3-15C strain to grow at 37 °C. WT, wild-type strain; gpi3 t-s,mutant.(C)[ 14 C]GlcNAc-PI synthesis. Mixed membranes were prepared from a wild-type (WT) and a gpi3::kanMX4 segregant (Dgpi3) harboring pRS425-Gpi3p- C301A and incubated with UDP-[ 14 C]GlcNAc for the times indicated, after which radiolabeled lipids were extracted, separated by TLC, and detected by fluorography. (D) Quantification of [ 14 C]GlcNAc-PI. Amounts of 14 C signal in each sample migrating at the position of [ 14 C]GlcNAc-PI on the chromatogram in (C) were quantified using a Phosphorimager and plotted as relative intensities. 4512 Z. Kostova et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Gpi3p to be incorporated into and function in a complex, then the mutant protein’s actual catalytic activity would, if anything, be higher in vivo. The only mutant where such additional tests may be required is the nonfunctional E297A protein. However, the ability of the related mutant E297D to function at 25 °C despite its inability to function at 37 °C suggests that E297D at least is properly folded and localized in the cell and that the E297A point mutant is likely to be similar. Although gene dosage effects might have been expected, expression of the mutant proteins from high copy plasmids did not result in elevated in vitro GlcNAc-PI synthetic activity or improved cell growth compared with strains expressing the same mutant proteins on low copy plasmids. However, because Gpi3p functions in a protein complex [6,9,17,18], the availability of the other subunits may be limiting, such that the number of functional complexes is not significantly increased when one subunit is overexpressed. Consistent with this, even very high level expression of wild- type Gpi3p in a gpi3 deletion background using a galactose- inducible promoter resulted in only a slight elevation of in vitro GlcNAc-PI synthesis [6]. Likewise, overexpression of Gpi1p or Gpi2 does not significantly increase GlcNAc-PI synthetic activity [6,19]. Failure of excess Gpi3p to be incorporated into a GlcNAc-PI synthetic complex may also lead to its degradation. Such is the case with the catalytic subunit of the GPI transamidase complex, Gpi8p: mono- meric Gpi8p subunits that are excluded from complete complexes are turned over rapidly [20]. The high degree of conservation of the EX 7 EGlu residues, and their potential to function as nucleophiles or general acid/base catalysts or to participate in sugar nucleotide binding suggest that these two Glu residues are active-site residues and participate in catalysis of glycosyl transfer from a sugar nucleotide donor to an acceptor glycan [1,3–5]. Studies of two other EX 7 E motif proteins led to the conclusion that the first of the two Glu residues is more important for in vitro activity of both AceA and human muscle glycogen synthase, and, in the case of AceA, for the ability to transfer mannose to an endogenous acceptor glycan in vivo when expressed in a heterologous system. These studies led to the proposal that the first Glu functions as a nucleophile [4,5]. Our finding that the E289A mutation does not abolish Gpi3p function in vivo calls into question the possible role of Glu289 as the nucleophile in the Gpi3p-catalyzed reaction. Although there are differences between the three enzymes in their acceptor glycans and sugar nucleotide donors, with Gpi3p using UDP-GlcNAc, glycogen synthase using UDP- Glc, and AceA using GDP-Man, it is unlikely that the proteins would have entirely different catalytic sites. Like- wise, although the yeast Gpi3 protein functions in the context of a complex that contains at least three other proteins [6,9,17,18], it seems unlikely that the participation of Gpi3p in an enzyme complex would reverse the roles of the EX 7 E Glu residues. The present in vivo assessment of protein function differs from the analyses used in the mutagenesis studies of glycogen synthase and AceA in that the importance of the EX 7 E Glu residues of human muscle glycogen synthase and AceA could not be tested in a context in which these proteins were essential for cell growth. Were such tests possible, conclusions about the relative importance of the EX 7 E Glu residues of glycogen synthase and AceA might be reversed. We note that it is unlikely that the ability of yeast Gpi3p mutants such as E289A to grow is due to a bypass of a requirement for GPI synthesis, because gpi3::kanMX4 haploids – alone, or harboring Gpi3p-E297A – are nonviable. Other amino-acid residues appear to be important for Gpi3p function. Mutagenesis studies of the human sequence homologue of Gpi3p, Pig-A, have established that Gly48, His128, Ser129, and Ser155 are important for function [8,21], but mutations in the EX 7 E Glu residues were not examined in these studies. Gpi3 proteins all contain a cysteine four residues C-terminal to the second EX 7 E glutamate. Because of its location near residues proposed to be involved in catalysis, and because in vitro GlcNAc-PI synthetic activity is inhibited by alkylating agents but protected from inhibition by uridine nucleotides, it has been suggested that this cysteine may have a role in UDP-GlcNAc binding [7,8]. Mutation of Cys301 in S. cerevisiae Gpi3p to Ala would therefore be expected to have a severe, if not lethal, effect. Our finding that this mutation has no apparent effect on the protein’s ability to support cell growth indicates that this Cys, which is conserved in Gpi3/Pig-A proteins although not in other EX 7 E-containing proteins, is not important for enzyme activity in vivo. As was the case with E289A, our results with the C301A mutant indicate that the expressed protein is properly folded and localized in the cell because the mutant is functional in vivo. The lowering of in vitro GlcNAc-PI synthetic activity upon introduction of the C301A mutation into Gpi3p may mimic the observed in vitro inactivation of GlcNAc-PI synthesis by alkylating agents [7]. However, it is also possible that the sensitivity of in vitro activity to alkylating agents is due instead to modification of other Cys residues in the GlcNAc-PI synthetic complex. In summary, the results of our mutagenesis studies with Gpi3p suggest that roles for the conserved EX 7 EGlu residues, as well as the nearby Cys301, cannot yet be assigned with confidence for this protein. Therefore, assignments of exact functions for the Glu residues of other EX 7 E-containing proteins may be open to question. Indeed, questions about the identity of catalytically important residues have recently also been raised for the NRD1b glycosyltransferase family: here too, a proposed catalytic glutamate could be changed to Ala without abolition of the catalytic activity of the representative enzyme studied [22]. Definition of the roles of the EX 7 E Glu residues, and of signature amino acids in other glycosyltransferases, will require 3D structures of members of this family of glycosyltransferases. Acknowledgements This work was supported by National Institutes of Health Grant GM46220 to P.O., by National Institutes of Health Grant GM55427 and grant 020026 from the Mizutani Foundation for Glycoscience to A.K.M., and by American Heart Association postdoctoral fellowship 0120565Z to S.V. We thank M. Glaser, B. Ng, and P. Rodriguez- Waitkus for assistance with microscopy, and B. Dylan and Axel Heyst for stimulation. Ó FEBS 2003 EX 7 E motif in Gpi3p GlcNAc transferase subunit (Eur. J. Biochem. 270) 4513 References 1. Geremia, R.A., Petroni, E.A., Ielpi, L. & Henrissat, B. (1996) Towards a classification of glycosyltransferases based on amino acid sequence similarities: prokaryotic a-mannosyltransferases. Biochem. J. 318, 133–138. 2. Campbell, J.A., Davies, G.J., Bulone, V. & Henrissat, B. (1997) A classification of nucleotide-diphospho-sugar glycosyltransferases based on amino acid sequence similarities. Biochem. J. 326, 929–939. 3. Kaptinov, D. & Yu, R.K. (1999) Conserved domains of glyco- syltransferases. Glycobiology 9, 961–978. 4. Cid, E., Gomis, R.R., Geremia, R.A., Guinovart, J.J. & Ferrer, J.C. (2000) Identification of two essential glutamic acid residues in glycogen synthase. J. Biol. Chem. 275, 33614–33621. 5. Abdian, P.L., Lellouch, A.C., Gautier, C., Ielpi, L. & Geremia, R.A. (2000) Identification of essential amino acids in the bacterial a-mannosyltransferase aceA. J. Biol. Chem. 275, 40568–40575. 6. Kostova, Z., Rancour, D., Menon, A.K. & Orlean, P. (2000) Photoaffinity labeling with P3-(4-azidoanilido) uridine 5¢-triphos- phate identifies Gpi3p as the UDPGlcNAc-binding subunit of the enzyme that catalyzes formation of N-acetylglucosaminyl phos- phatidylinositol, the first glycolipid intermediate in glycosyl phosphatidylinositol synthesis. Biochem. J. 350, 815–822. 7. Milne, K.G., Ferguson, M.A. & Masterson, W.J. (1992) Inhibi- tion of the GlcNAc transferase of the glycosylphosphatidylinisitol anchor biosynthesis in African trypanosomes. Eur. J. Biochem. 208, 309–314. 8. Tiede, A., Nischan, C., Schubert, J. & Schmidt, R.E. (2000) Characterization of the enzymatic complex for the first step in glycosylphosphatidylinositol biosynthesis. Int. J. Biochem. Cell. Biol. 32, 339–350. 9. Leidich, S.D., Kostova, Z., Latek, R.R., Costello, L.C., Drapp, D.A., Gray, W., Fassler, J.S. & Orlean, P. (1995) Temperature- sensitive yeast GPI anchoring mutants gpi2 and gpi3 are defective in the synthesis of N-acetylglucosaminyl phosphatidylinositol: cloning of the GPI2 gene. J. Biol. Chem. 270, 13029–13035. 10. Zieler, H.A., Walberg, M. & Berg, P. (1995) Suppression of mutations in two Saccharomyces cerevisiae genes by the adeno- virus E1A protein. Mol. Cell. Biol. 15, 3227–3237. 11. Sherman, F. (1991) Getting started with yeast. Methods Enzymol. 194, 3–21. 12. Alfa, C., Fantes, P., Hyams, J., McLeod, M. & Warbrick, E. (1993) Experiments with Fission Yeast. A Laboratory Course Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 13. Christianson, T.W., Sikorski, R.S., Dante, M., Shero, J.H. & Hieter, P. (1992) Multifunctional yeast high-copy-number shuttle vectors. Gene 110, 119–122. 14. Altschul, S.F., Gish, W., Miller, W., Myers, E.W. & Lipman, D.J. (1990) Basic local alignment search tool. J. Mol. Biol. 215, 403–410. 15. Minet, M., Dufour, M E. & Lacroute, F. (1992) Complementa- tion of Saccharomyces cerevisiae auxotrophic mutants by Arabidopisis thaliana cDNAs. Plant J. 2, 417–422. 16. Cottarel, G., Beach, D. & Deuschle, U. (1993) Two new multi- purpose multicopy Schizosaccharomyces pombe shuttle vectors, pSP1 and pSP2. Curr. Genet. 23, 547–548. 17. Yan, B.C., Westfall, B.A. & Orlean, P. (2001) YNL038wp (Gpi15p) is the Saccharomyces cerevisiae homologue of human Pig-Hp and participates in the first step in glycosylphos- phatidylinositol assembly. Yeast 18, 1383–1389. 18. Watanabe, R., Inoue, N., Westfall, B., Taron, C.H., Orlean, P., Takeda, J. & Kinoshita, T. (1998) The first step of glycosylpho- sphatidylinositol biosynthesis is mediated by a complex of PIG-A, PIG-H, PIG-C, and GPI1. EMBO J. 17, 877–885. 19. Leidich, S.D. & Orlean, P. (1996) Gpi1, a Saccharomyces cerevi- siae protein that participates in the first step in glycosylphos- phatidylinositol anchor synthesis. J. Biol. Chem. 271, 27829– 27837. 20. Fraering, P., Imhof, I., Meyer, U., Strub, J.M., van Dorsselaer, A., Vionnet, C. & Conzelmann, A. (2001) The GPI transamidase complex of Saccharomyces cerevisiae contains Gaa1p, Gpi8p, and Gpi16p. Mol. Biol. Cell 12, 3295–3306. 21. Norris, E.R., Howard, T.A., Marcus, S.J. & Ware, R.E. (1997) Structural and functional analysis of the Pig-A protein that is mutated in paroxysmal nocturnal hemoglobinuria. Blood Cells Mol. Dis. 23, 350–360. 22.Hefner,T.&Sto ¨ kigt, J. (2003) Probing suggested catalytic domains of glycosyltransferases by site-directed mutagenesis. Eur. J. Biochem. 270, 533–538. 23. Costello, L.C. & Orlean, P. (1992) Inositol acylation of a potential glycosyl phosphoinositol anchor precursor from yeast requires acyl coenzyme A. J. Biol. Chem. 267, 8599–8603. 4514 Z. Kostova et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . Comparative importance in vivo of conserved glutamate residues in the EX 7 E motif retaining glycosyltransferase Gpi3p, the UDP-GlcNAc-binding subunit of the first enzyme in glycosylphosphatidylinositol. test the importance of the conserved Glu residues in the EX 7 E motif of these proteins by the stringent criterion of their ability to support cell growth. Our results indicate that both Glu residues. that the second Glu residue of the EX 7 E motif in Gpi3p is of greater importance than the first for function in vivo. Further, our findings do not support previous suggestions that the first Glu of