A Non-Golgi (cid:1) 1,2-Fucosyltransferase That Modifies Skp1 in the Cytoplasm of Dictyostelium *

Skp1 is a subunit of the SCF-E3 ubiquitin ligase that targets cell cycle and other regulatory factors for deg-radation. In Dictyostelium , Skp1 is modified by a pentasaccharide containing the type 1 blood group H trisac-charide at its core. To address how the third sugar, fucose (cid:1) 1,2-linked to galactose, is attached, a proteomics strategy was applied to determine the primary structure of FT85, previously shown to copurify with the GDP-Fuc:Skp1 (cid:1) 1,2-fucosyltransferase. Tryptic-generated peptides of FT85 were sequenced de novo using Q-TOF tandem mass spectrometry. Degenerate primers were used to amplify FT85 genomic DNA, which was further extended by a novel linker polymerase chain reaction method to yield an intronless open reading frame of 768 amino acids. Disruption of the FT85 gene by homologous recombination resulted in viable cells, which had al-tered light scattering properties as revealed by flow cytometry. FT85 was necessary and sufficient for Skp1 fucosylation, based on biochemical analysis of FT85 mutant cells and Escherichia coli that express FT85 recom-binantly. FT85 lacks sequence

Skp1 is an M r 21,000 protein found in the cytoplasm and nucleus, where it is a subunit of the SCF-E3 ubiquitin ligase family (1) and possibly other molecular complexes (2). In Dictyostelium, nearly all Skp1 is modified by a pentasaccharide, Gal␣1,6Gal␣1(Fuc␣1,2Gal␤1,3GlcNAc), 1 which is attached in O-linkage to HyPro143 (3,4). The HyPro attachment site is predicted to be located between a loop and an ␣-helix near to but projecting away from the interface with the F-box protein of the SCF-complex (5). Immunofluorescence studies localize Skp1 to the nucleus and regions of the cytoplasm (6,7,4), and mutant analysis suggests that at least the core disaccharide is required for concentration of Skp1 in the Dictyostelium nucleus (4).
Skp1 is predicted to be glycosylated by a novel 6-enzyme pathway, including a prolyl hydroxylase and five glycosyltransferases (GTases). 2 Two of the GTases, the GlcNAcTase that adds the first sugar and the ␣1,2-fucosyltransferase (FTase) that adds the third sugar, have been purified to near homogeneity from the cytosolic fraction of Dictyostelium cells and characterized. The GlcNAcTase transfers [ 3 H]GlcNAc from UDP-[ 3 H]GlcNAc to HyPro143 of a mutant Skp1 that is mostly hydroxylated but not glycosylated in vivo (8). This activity exhibits submicromolar K m values for both its donor and acceptor substrates, cannot be detected in the vesicular fraction of the cell, and requires the presence of a reducing agent in vitro. These properties suggest that this enzyme normally functions in the reducing environment of the cytoplasm, where substrates are expected to be less concentrated than in the secretory pathway.
A Skp1 ␣1,2-FTase activity (EC 2.4.1.69) has been assayed based on the transfer of [ 3 H]Fuc from GDP-[ 3 H]Fuc to pNO 2phenyl-Gal␤1,3GlcNAc (pNP-LNB) or to Gal␤1,3GlcNAc-Skp1 isolated from a GDP-Fuc synthesis mutant (9,10). Highly purified preparations of the enzyme are specific for the O-2 position of Gal that is in ␤1,3 linkage to an underlying HexNAc, which may be either GlcNAc or GalNAc, ␣or ␤-linked to an aglycone. Activity varies by an order of magnitude depending on the nature of the aglycone and displays a K m for the disaccharide on Skp1 that is three orders of magnitude lower than that of the best disaccharide aglycone. The acceptor carbohydrate substrate specificity profile of the Skp1 FTase suggests a closer relationship to the vertebrate Se-type compared with the H-type ␣1,2-FTases found in the Golgi apparatus. However, the Skp1 FTase is present in the cytosolic fraction with little detected in the vesicular fraction of the cell and, as for the Skp1 GlcNAcTase, has biochemical characteristics of a cytoplasmic rather than a Golgi protein. These properties suggest that the Skp1 FTase is also located in the cytoplasm of the cell with its acceptor substrate Skp1, not in the Golgi like the Se-FTase. The similar findings for both the GlcNAcTase and FTase suggest that the entire 6-enzyme pathway acts directly and sequentially on Skp1 in the cytoplasmic compartment.
Although eukaryotic cytoplasmic glycosylation is not as extensive and varied as glycosylation in the secretory pathway, a number of important examples are known. Dol-P-Man, Dol-P-Glc, and the Man 5 GlcNAc 2 -P-P-Dol precursor of N-glycsosylation are synthesized by cytoplasmic enzymes prior to flipping into the interior of the rER (11). The sugars of phosphatidylinositol GlcNH 2 and glucosylceramide each appear to be attached by cytoplasmically oriented enzymes (12,13). Glycogen is the product of two glucosyltransferases (14), and the extracellular polysaccharides cellulose, chitin, and hyaluronate are synthesized by enzymes whose active sites are cytoplasmically oriented, though their polysaccharide products are co-synthetically translocated across the plasma membrane (15)(16)(17). Analysis of the enzyme genes suggests that most of these pathways are topologically conserved from prokaryotes. Many proteins are modified at specific Ser or Thr residues with O-␤-GlcNAc by a Ser/Thr GlcNAcTase present in the cytoplasm and nucleus (18). There is also evidence for complex glycosylation of cytoplasmic proteins in addition to Skp1, such as parafusin (19) and Chlorella virus capsid proteins (20), but the enzyme genes have not been identified.
To confirm the cytoplasmic compartmentalization model for Skp1 glycosylation, we have turned to the GTases that carry out this process. In this study, we have cloned the gene for FT85, previously postulated to be the Skp1 FTase protein following million-fold purification and photoaffinity labeling (9), using a proteomics strategy based on the Q-TOF mass spectrometer that allowed femtomolar level sequencing of tryptic peptides from FT85 (21,22). The inferred sequence of FT85 confirms the cytosolic localization model. The Skp1-FTase is related to a large family of GTases (23)(24)(25) that contributes to the synthesis of lipid-linked glycans and polysaccharides, including some of those described above, in the cytoplasm of prokaryotes and eukaryotes prior to their translocation to the cell exterior.

EXPERIMENTAL PROCEDURES
Cells-The normal strain Ax3 and strain HL250, a mutant that cannot synthesize GDP-Fuc (26), were grown in HL-5 axenic growth medium until they achieved stationary phase (maximum cell density) for up to 1 day. To induce aggregation, exponentially growing cells were washed in KP (10 mM potassium phosphate, pH 6.5) and incubated in this buffer at 2 ϫ 10 7 cells/ml for up to 15 h on a gyratory shaker.
Purification of FT85-The FTase was purified from 300 liters of stationary phase cells (10 7 cells/ml). Cells were grown in batches of 30 liters, and S100 fractions from each were purified through the first DEAE anion exchange column chromatography step (9) and frozen at Ϫ80°C. These preparations were pooled for further purification as described previously (9). Aliquots of fractions from the final Superdex-200 gel filtration column were assayed for FTase activity using the pNP-LNB assay (see below), and electrophoresed in an SDS-PAGE gel and stained with silver to identify the elution position of FT85.
MS Sequencing of FT85-Fractions corresponding to the peak of FTase activity were subjected to an in-gel digest proteomics Q-TOF strategy (22) initially involving purification by 10% SDS-polyacrylamide gel electrophoresis, with Coomassie staining. A relatively lightly staining band with approximately M r 85,000 mobility (other bands were studied and identified as impurity proteins) was exised from the gel and destained in 0.1 M ammonium bicarbonate in acetonitrile for 10 min, followed by incubation with 1 g of porcine trypsin in 50 mM ammonium bicarbonate, pH 8.5, overnight at 37°C. Following extraction of the tryptic peptides from the gel in acetonitrile/0.1% trifluoroacetic acid (6:4 v/v) and reducing to 10% volume under vacuum, samples were loaded onto a 1 ϫ 10-mm C 18 reverse phase cartridge (Jones Chromatography) and eluted in a stepwise manner with 0.1% trifluoroacetic acid in H 2 O, 0.1% trifluroacetic acid in 30% acetonitrile, and finally 0.1% trifluroacetic acid in 60% acetonitrile to give 10-l samples.
Samples thus purified were studied initially in the MS mode (normal full spectrum from m/z 300 -2000) using a nanospray ion source on a Q-TOF mass spectrometer (Micromass UK) (27). Although signal strengths on the M r 85,000 band were weak, a number of doubly and triply charged signals were identified from isotopic spacings in the MS spectra. These signals were passed separately for collisional-activated decomposition tandem MS experiments (CAD MS/MS) into a collision cell filled with argon gas, using collision energies from 10 -40 eV as appropriate (27). Peptide sequences were interpreted manually from a knowledge of the low energy fragmentation pathways expected (22,27), and where protein data base library matches (OWL/SWISSPROT) failed to identify a protein precursor, these de novo sequences were assumed to be likely candidates for FTase primers. BLAST searches for homology matches gave moderate homology matches in some cases with GTases from other species.
Sequencing the FT85 Gene-Degenerate oligonucleotides were based on the peptide sequences derived from mass spectrometry and Dictyostelium codon usage frequencies. Multiple subsets of degenerate primers were synthesized as described in "Results," and primer lengths were 25-30 nucleotides to achieve T m values of 54 -64°C (GC ϩ AT method) given the high A/T content of Dictyostelium DNA. Primers were used in the touchdown (td) variation of PCR (28), in which the annealing temperature is reduced by 1°C/cycle during each of the first 15 cycles, followed by a return to 3°C below the initial annealing temperature for an additional 15 cycles. For amplification of td-pcr-1 (Fig. 4A), denaturation was at 94°C for 45 s (3 min for first cycle), the initial annealing temperature was 50°C, and extension was at 68°C for 2.5 min (10 min last cycle) using a Stratagene Robocycler. Degenerate primer pools were used at 100 pmol each in a 50-l reaction volume. CsCl-purified genomic DNA was isolated from nuclei of D. discoideum strain Ax3 (29), being careful to avoid the lower, intensely pigmented, mitochondrial region of the pellet when resuspending the nuclei.
To amplify unknown neighboring DNA, a linker-mediated PCR method was modified from a strategy originally employed by the Marathon and Gene-Walker methods by CLONTECH (30). 2.5 g of CsClpurified DNA was digested with either BamHI, BclI, BglII, or BstYI, phenol/chloroform extracted, ethanol-precipitated after the addition of 2.5 M ammonium acetate, washed in 70% ethanol and dried.
The following synthetic oligonucleotides (Integrated DNA Technologies, Coralville, IA) were used to form a double-stranded linker: 5Ј-GTCATCCAGTAAGCGTAGCCAGAGCGAAGGTCCCGTCCA-3Ј and 3Ј-H 2 N-GGGCAGGTCTAG-PO 4 -5Ј. Equal molar amounts dissolved in water were distributed into 100-l aliquots at 1 M, heated to 95°C, allowed to cool slowly to 37°C, and flash-frozen in liquid N 2 . These were expected to anneal via the underlined regions.
The digested genomic DNA was diluted to 5 ng/l in ligation buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 10 mM dithiothreitol, 1 mM ATP, 25 g/ml bovine serum albumin) with 4 units/l of T4 DNA ligase (New England Biolabs; Beverly, MA), and 4ϫ the calculated free molecular end concentration of annealed linker and ligated at 16°C for 16 h. 50 ng of library was used in each PCR reaction. The linkermodified libraries were typically amplified with a gene-specific primer (Fig. 4A) and LP-1, 5Ј-GTCATCCAGTAAGCGTAGC-3Ј using the touchdown protocol described above for 30 cycles (94°C, 45 s; 55°C, 45 s; 68 -72°C, 2.5-4 min) and Taq DNA polymerase. In principle, LP-1 was able to hybridize with only those DNA species that had been extended by reactions primed by the gene-specific primer. In some cases, the initial reaction was diluted 50-fold, and PCR was repeated using a nested gene-specific primer coupled with LP-1, LP-2, 5Ј-AGCGTAGC-CAGAGCGAAGGT-3Ј or LP-3, 5Ј-GTCATCCAGTAAGCGTAGCCA-GAGCGAA-3Ј based on which had the best T m match. PCR products were cloned into pCR4-TOPO (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions and sequenced in both directions.
Nucleotide sequences were also extended by searching for matches in genomic fragments sequenced by the International Dictyostelium Genome Sequencing Consortium, and in cDNA/ESTs sequenced by the Japanese cDNA project at Tsukuba University, using BLAST servers at genome.imb-jena.de/dictyostelium/, dicty.sdsc.edu/, or www.csm.biol. tsukuba.ac.jp/cDNAproject.html. Sequences were then confirmed on PCR-amplified full-length and domain-coding DNA, using primers P5, P4, P11, P10, P12, and P6 as indicated in Figs. 4 and 5.
FT85 Mutant Cells-The blasticidin-S resistance (bsr) cassette was excised from pBsR519 (31) with PstI and ligated into the compatible NsiI site of td-pcr-1 in pCR4-TOPO, as shown in Fig. 4B. To induce homologous recombination, the resulting DNA insert was excised with EcoRI and electroporated into Dictyostelium strain Ax3 cells (32).
Transformants were selected in 10 g/ml blasticidin-S (Life Technologies, Inc.) and cloned on SM agar plates in the presence of Aerobacter aerogenes. Transformants were screened for insertion of bsr into the FT85 locus by amplification of genomic DNA using primers P9 and P10 (Fig. 4A) as described (33).
Expression of FT85 in E. coli-The full-length coding region of FT85 predicted from the nucleotide sequence was amplified using primers 5 and 6 or 7 and 8 (Figs. 4C and 5), cloned into pCR4-TOPO, and amplified in E. coli strain TOP10 One Shot chemically competent cells (Invitrogen). Restriction sites designed into the 5Ј-ends of these primers were used to excise the PCR-amplified DNA inserts for ligation between the NdeI and SapI sites of pTYB1, or the SapI and PstI sites of pTYB11, creating an in-frame fusion at the N terminus or C terminus, respectively, of the IMPACT-CN chitin-binding domain (CBD) via an intein linker (Ref. 34; New England Biolabs). pMYB5, which contained the maltose-binding protein fused to the intein tag, was used as a control. The expression plasmids, pTYFT-CBD and pTYCBD-FT, were produced in TOP10 One Shot chemically competent E. coli cells and then transfected into E. coli strain ER2566 for expression, which was induced when cells achieved an A 600 of 0.5, by incubation in 0.5 mM IPTG for 20 h at 15°C. Soluble extracts were prepared by freeze-thawing cell pellets, probe sonication until the suspension became viscous, and centrifugation at 13,000 ϫ g for 15 min at 5°C.
The underlined regions correspond to FT85 DNA in Figs. 4C and 5. Primers 13 and 15 encode a 16-nucleotide sequence upstream of the start codon designed to support translation initiation, and P15 in addition encodes a start codon prior to codon 397 for the C-terminal domain. The PCR products were digested with KpnI and BglII (shown in bold), and cloned into the KpnI and BglII sites of the expression vector pVS4 (33), resulting in the replacement of its start codon and signal peptide with the new coding region fused to a C-terminal c-Myc epitope tag. This plasmid directs expression under control of the discoidin promoter, which is constitutive in our hands (4), and the actin 8 terminator. The new plasmids, pVFT85N and pVFT85C, were cloned in TOP10 cells, purified using a QIAprep Spin mini-prep kit (Qiagen, Valencia, CA), and electroporated into HW260 cells as described (4). Transformants, which typically integrate the plasmid chromosomally in multicopy tandem arrays, were selected in 5-7.5 g/ml G418 (Life Technologies, Inc.) in HL-5ϩ growth medium. DNA sequencing showed that pVFT85N was modified from pTYFT-CBD by the absence of codons for 22 of the 29 contiguous Asn residues encoded at the C terminus of the construct; pVFT85C was likewise deleted in 20 of these codons.

SDS-PAGE and Western
Blotting-For M r analysis of Skp1, stationary phase cells were lysed by sonic disruption with a probe sonifier in 50 mM Tris-HCl, pH 7.4, with protease inhibitors (9) and centrifuged at 13,000 ϫ g for 1 h. The supernatant from 1.5 ϫ 10 6 cells (100 g of protein) was loaded onto 13-cm long, 15-20% linear gradient polyacrylamide SDS-gels and electrophoresed as previously described (4). Gels were Western blotted and probed with mAb 3F9 to detect Skp1 (4).
Fucosylation Assays in Cell Extracts-Dictyostelium cells (stationary stage) were washed and suspended in 0.25 M sucrose, 50 mM Tris-HCl, pH 7.4, lysed by forced passage through a 5-or 3-m diameter pore size Nuclepore filter, centrifuged at 13,000 ϫ g for 1 h, and the supernatant (S13) was used for assays (9). To assay fucosylation of the acceptor pNP-LNB, the S13 fraction of Dictyostelium or E. coli (see above), whose protein content had been determined by comparison to bovine serum albumin using a Coomassie Blue dye binding assay (Pierce), were incubated at a final concentration of 1.0 -1.2 M GDP-[ 3 H]Fuc, prepared from a mixture of GDP-[1-3 H]Fuc (PerkinElmer Life Sciences 16.5 Ci/mmol) and unlabeled GDP-␤-L-Fuc (Sigma Chemical Co.), in 50 l at 30°C for 1 h. Incorporation was measured as dpm that were adsorbed to a C 18 -SepPak cartridge and released by the first MeOH wash, after subtraction of the dpm eluted by the final water wash prior to MeOH elution. Unless indicated, the result from a blank assay that contained no added pNP-LNB was subsequently subtracted, as described previously (9).
To assay fucosylation of Skp1, Skp1 purified from the GDP-Fuc Ϫ strain HL250 (9) was added at the concentration indicated to E. coli or Dictyostelium S13 fractions. Incorporation was measured by liquid scintillation counting of trichloroacetic acid precipitates or counting of the Skp1 band after SDS-PAGE, as described previously (8).
Metabolic Labeling with [ 3 H]Fuc-Cells were incubated either in HL-5 growth medium during exponential growth or in KP-buffer to induce aggregation for 16 h in 0.1-0.4 mCi/ml L-[6-3 H]Fuc (85.2 Ci/ mmol, PerkinElmer Life Sciences; or 60 Ci/mmol, American Radiolabeled Corp.) which had been previously dried down. Cells were washed in KP, lysed and centrifuged to produce S100 and P100 extracts as described above. 100 g of extract protein was subjected to SDS-PAGE and, after staining with Coomassie Blue and destaining, the gel was cut into 55-60 equal parts for scintillation counting as above.
Flow Cytometry-Cells growing at a density of 1-3 ϫ 10 6 cells/ml in HL-5 were introduced directly into a BD Biosciences FACScan flow cytometer (San Jose, CA). Forward (low angle) and side (right angle) light scatter values were determined using illumination from an Ar ion laser (488 nm) operating at 15 milliwatts. 60,000 cells were counted in each sample, and data were analyzed using CellQuest software from BD Biosciences.

RESULTS
Sequencing the FT85 Polypeptide-FT85 was purified nearly 1 million-fold as described previously (9). Aliquots from the final gel filtration step were assayed for FTase activity using pNP-LNB and analyzed by SDS-PAGE and silver staining. Although the present purification yielded higher levels of lower M r contaminating bands, the staining intensity of an M r 85,000 band correlated with the level of FTase activity (Fig. 1). The peak fractions 27-29 were pooled and purified by SDS-PAGE followed by staining with Coomassie Blue. The lightly staining 85-kDa band (together with other bands, which were later found to be impurities) was subjected to in-gel tryptic digestion, and the purified peptide mixture was studied by high sensitivity Q-TOF tandem MS procedures (21,22,27). Fig. 2A shows a typical nanoelectrospray partial mass spectrum (approximately m/z 450 -750) using 1 l of the 30% acetonitrile eluate from the cartridge elution of the gel peptide extract (see "Experimental Procedures"). Tryptic peptides, conventionally carrying an N-terminal amino group together with a C-terminal side chain amino on Lys or Arg will typically be doubly charged, showing half-mass unit spacings in their natural 13 C isotope satellites (27). The low level of protein in the M r 85,000 band is illustrated in this spectrum where most of  200) were assayed for FTase activity using pNP-LNB as an acceptor (A), and for protein by SDS-PAGE followed by silver staining (B). The band marked FT85 co-elutes with activity and corresponds to a band that was previously photoaffinity-labeled with GDP-hexanolamine-azido-125 I-salicylate (9). the signal intensity is singly charged background. However, some weak doubly charged signals are evident, as shown for m/z 461, 645, and 727, with m/z 645 expanded in the inset to illustrate the half-mass unit spacing visible within the background.
Selecting m/z 645 for MS/MS analysis produced the spectrum shown in Fig. 2B, where an excellent series of N-terminal and C-terminal (b and yЉ) fragment ions (22)  MS/MS analysis of the doubly charged m/z 727 ion in Fig. 2A produced a fragment ion spectrum, which was interpreted to give a sequence of YYFTL(I)L(I)DA. A search of the protein data bases showed this to be derived from Elongation Factor 1␣. The sequence study of numerous peptides from two independent purifications showed many to be derived from this impurity protein or from keratin background (common when working at this level), but some 15 peptide sequences were not identified in data base searches and were therefore presumed to derive from FT85. Of these, which included a mix of full and partial predicted sequences, the confidence limit ranged from good to very high, and from the latter, three were chosen for primer candidate sequences. Of these, two were thought suitable for design of degenerate oligonucleotides for PCR: MDSD-DL(I)SHPTR and NHPNR (peptides 1 and 2). The latter high confidence sequence was interpreted (Fig. 3, A and B) as being preceded most probably by DGL(I) or NGL(I) in MS/MS spectra of m/z 461 obtained from different fractions of the cartridge purification.
Sequencing the FT85 Gene-Oligonucleotide pools that encode candidate sequences from peptides 1 and 2 in either the forward or reverse directions were synthesized to represent all codons used in Dictyostelium at a frequency of Ն25%. Three pools of oligonucleotides corresponding to the forward and reverse reads of peptide 1 were synthesized so that the number of sequences present in each pool was 8. Similarly, 3 pools were synthesized corresponding to each direction for each of the candidate peptide 2 sequences, and the degeneracy ranged from 16-to 128-fold. For PCR, each forward direction oligonucleotide pool derived from one peptide was mixed with each reverse direction oligonucleotide pool from the other peptide and vice versa. These primer pool pairs were used to amplify DNA from genomic DNA, using the touchdown protocol described in the "Experimental Procedures" section. One combination of a forward primer pool from peptide 1 (primer 1, 5Ј-ATGGATTCWGATGATATYTCW) and a reverse primer pool from peptide 2 (primer 2, based on the peptide sequence DG-L(I)NHPNR, 5Ј-RTTTGGRTGRTTNARACCATC) uniquely amplified a 1.7-kilobase DNA, td-pcr-1 (Fig. 4A) that was not seen with either of the primer pools alone. Its nucleotide sequence belonged to a single ORF in the forward direction. Adjacent DNA was amplified in both directions from linkermodified sublibraries of BstYI-digested genomic DNA using primers P3 and P4, and linker-specific primers, as shown in Fig. 4A.
A total of 3584 nucleotides were sequenced by PCR (Fig. 4A).
The termini of this sequence overlapped with cDNA and unconfirmed genomic sequences obtained from the Dictyostelium cDNA and genome sequencing projects, which permitted tentative extension of the overall sequence to 4701 nucleotides. Conceptual translation yielded a single, long, forward-directed ORF of 2304 nucleotides in the PCR-derived sequence, with a G/C-content of 21% and codon usage typical for Dictyostelium coding regions. This main ORF was predicted to encode a 768-residue protein with a calculated M r of 89,735 (Fig. 5), which compared favorably with the predicted M r of FT85, 85,000 (Fig. 1B). The first Met codon of this ORF was the only ATG in the vicinity, occurred at a boundary between G/C-poor and G/C-richer sequence, and had a sequence context similar to that of other Dictyostelium genes (35). Sequences that potentially correspond to a TATA-box and an adjacent oligo(dT) box are found ϳ130 nucleotides upstream of the assigned start codon (Fig. 5, bold). A second forward-directed ORF was found at the 5Ј-end of the PCR-derived sequence, separated from the main ORF by 262 nucleotides that have a G/C-content of 17% (Figs. 4A and 5). This ORF was extended in the upstream direction by the genome data base sequences and had a length of at least 1377 nucleotides (Ͼ459 amino acids) and G/C-content of 27%. It is unlikely that the upstream ORF is a separate exon of the main ORF, as: 1) there are no consensus donor and acceptor splice sequences that can be used to create a conventional intron near the ORF boundaries, 2) its 3Ј-end is found in a cDNA/EST clone (see Fig. 4A) suggesting that it is a separate transcription product, and 3) a consensus polyadenylation signal (AATAAA) for the upstream ORF lies 88 nucleotides upstream of the start of the main ORF (Fig. 4A, bold). Interestingly, this suggests that the 3Ј-untranslated sequence of the upstream gene overlaps with the promoter and 5Ј-untranslated sequence of the FT85 gene.
A third, reverse-oriented ORF was found at the 3Ј-end of the PCR-derived sequence, which was extended an additional 541 nucleotides for an overall length of at least 652 nucleotides, using overlapping cDNA/EST sequence reads from the Japa-  Fig. 2, showing the fragment ions used to deduce the peptide sequence. B, the interpretation of how the yЉ ion masses were used to derive the sequence NHPNR with high confidence, which was most probably preceded by NGL(I) or DGL(I) (possibly resulting from deamidation of N), or possibly RE (not shown). Each of these sequences was used to design pairs of degenerate primer pools for the PCR studies. nese cDNA sequencing project (Fig. 4A). This ORF had a G/Ccontent of 28%, consistent with the G/C-contents of the other ORFs in this overall sequence. The intergenic region between the main ORF and this 3Ј-ORF was 103 nucleotides with a G/C-content of 3%, and contained consensus polyadenylation signals in both directions.
The main ORF was concluded to correspond to the entire coding region of FT85, with no introns, based on the DNA sequence evidence that the flanking regions were composed of intergenic DNA. The DNA sequence of the ORF was confirmed from a full-length PCR product produced for expression studies (see below). Twenty-five percent of this predicted protein sequence was then confirmed using the remainder of the original mass spectrometric data. This included matches of exact mass values, and the full or partial sequence interpretations made, together with residual (tag) masses in the case of partial interpretations. Importantly, the predicted C-terminal sequence SVHL(I)GEL(I)FL(I)S of FT85 was found in this data set. The matches are shown in bold in Fig. 5.
Properties of the Predicted FT85 Sequence-A search for sequences similar to FT85 was carried out by BLAST analysis of SwissProt and conceptually translated GenBank TM sequences, yielding a large number of weak matches between its N-terminal 270 amino acids and Family 2 (23) or Family E (24) GTases. Family 2 represents a wide variety of mostly prokaryotic GTases whose catalytic domains are about 260 amino acids in length, and whose catalytic mechanisms involve inversion of the donor sugar linkage as for the Skp1 FTase (36). Similarity was highest within four amino acid clusters (Fig. 6). The first three are associated with the previously described NRD2 domain (nucleotide recognition domain-2; Ref. 25), and includes the DXD motif or one of its variations in motif 3 (asterisks). The fourth sequence cluster contains the catalytic Asp residue (see "Discussion"). The C terminus of FT85 is separated from the rest of the protein by a string of 29 consecutive Asn residues starting at codon 409, a motif found in other Dictyostelium proteins (37) that is not expected to encode structural information. Although no significant matches were detected by BLAST for the C-terminal region of FT85, sequences related to the Family 2 motifs found in the N-terminal domain were recognized (Fig. 6). Thus both the N-and C-terminal parts of FT85 have the potential for GTase activity, as depicted in Fig. 4C. Previously described sequence motifs that are found in other ␣1,2-FTases (38) were not recognized in FT85.
No hydrophobic sequences were found near the N terminus that might target FT85 into the rER as for other known eukaryotic ␣1,2FTases. In addition, no other stretches of amino acids satisfied criteria for predicted membrane association, consistent with the soluble character of the protein when isolated from cell extracts (9).
FT85 Is Required for Skp1 Fucosylation-To determine whether FT85 was required for Skp1 fucosylation, the FT85 locus was disrupted by homologous recombination as outlined in Fig. 4B and described in "Experimental Procedures." Insertion of the bsr locus was targeted to codon 276 of the gene, which was expected to destabilize FT85 mRNA resulting in a null phenotype. Transformants were selected in the presence of blasticidin-S, and clones were initially screened for the presence of a lower M r form of Skp1 based on SDS-PAGE/Western blot analysis (see below). Clonal strain HW260 was analyzed by PCR using FT85-specific primers P9 and P10 that flanked the bsr insertion site and was found to contain an insertion in the FT85 gene whose length corresponded to that of the bsr locus (Fig. 7A).
Skp1 produced by mutant strain HW260 migrated slightly ahead of normal Skp1 on a SDS-PAGE gel (Fig. 7B), close to the position of Skp1 produced by a strain (HL250) unable to fucosylate Skp1 (3) because of a mutation in GDP-Fuc synthesis (26). To verify that Skp1(HW260) was not fucosylated, extracts of HW260 and normal strain Ax3 cells were compared after metabolic-labeling with [ 3 H]Fuc during growth in HL-5 or differentiation in KP buffer. Total radioactivity incorporated after FIG. 4. Strategies for sequencing, disrupting, and expressing the FT85 gene. A, degenerate oligonucleotides corresponding to peptides 1 and 2 were used to prime amplification of td-pcr-1 (see text). Specific primers derived from the td-pcr-1 sequence were used to extend the sequence in either direction by linker-mediated PCR. Overlapping sequence reads from the International Dictyostelium Genome Project (1-11) and from the Japanese cDNA/EST project at Tsukuba University (12-15) are shown below. Arrows denote the directions of transcription of the predicted ORFs. B, to inactivate the FT85 gene, the bsr-resistance casette from pBsr519 was ligated into the NsiI site of td-pcr-1 (A) in pCR4-TOPO. The disruption DNA fragment was excised with EcoRI and electroporated into strain Ax3. C, predicted domain structure of FT85 based on sequence analysis (Fig. 6). The full-length coding region was amplified by PCR and cloned into pTYB1 and pTYB11 for expression in E. coli, and the N-and C-terminal domains were amplified by PCR and cloned into pVS4 for expression in Dictyostelium strain HW260. precipitation with trichloroacetic acid was highly enriched in the P100 fraction, as previously observed (36) and was similar between the two strains (data not shown). The apparent M r profiles were compared after separation on an SDS-PAGE gel and quantitation of radioactivity by scintillation counting of individual gel slices. Similar profiles of incorporated [ 3 H]Fuc were observed for the P100 fractions of each strain grown on HL-5 (Fig. 8A), and similar profiles, with slightly greater mobility, were obtained for cells starved in KP (data not shown). Thus the majority of fucosylation was not affected in the mutant. In the S100 fractions from KP-starved cells, incorporation was detected at the Skp1 position of Ax3 but not FT85 mutant HW260 cells (Fig. 8B), confirming that Skp1, which accumulated at a normal level, was not fucosylated in the FT85mutant strain.
To test for Skp1 FTase activity directly, soluble cell extracts were initially assayed using a simple disaccharide conjugate, pNP-LNB, shown previously to be an acceptor substrate (9). In the presence of GDP-[ 3 H]Fuc, a high level of activity was observed in normal strain Ax3 extracts (Fig. 9A). A diminished level of incorporation was seen for strain HL250, the GDP-Fuc synthesis mutant, probably because non-fucosylated Skp1 present in this strain is competitive with pNP-LNB and is not detected by the assay (9). In contrast, no incorporation was detected in the FT85 mutant strain HW260. This was not due to expression of the bsr marker, because another strain (HW264) that had incorporated this marker elsewhere in the genome expressed a normal level of activity. Negligible activity in HW260 was confirmed using Skp1(HL250), which contains only the disaccharide core, as the acceptor substrate (Fig. 9A). Thus no Skp1 FTase activity could be detected in the S13 extracts, indicating that FT85 is necessary for FTase activity.
FT85 Exhibits Skp1 FTase Activity in E. coli-To determine whether FT85 alone was able to fucosylate Skp1, Skp1 FTase activity was assayed in soluble extracts of E. coli expressing full-length Dictyostelium FT85 recombinantly. As described in "Experimental Procedures," FT85 was fused at either its N terminus or C terminus to a chitin-binding domain via a cleavable intein linker and placed under the control of the T7 promoter (34). Induction of expression of the CBD-intein-FT85 and the FT85-intein-CBD fusions resulted in substantial levels of FTase activity using either pNP-LNB or Skp1(HL250) as acceptors (Fig. 9B). Activity was greater than that detected in wild-type extracts of Dictyostelium on a per cell protein basis.  Fig. 4, is shown. The predicted FT85 ORF consists of a single exon. Amino acid sequences in bold were confirmed by de novo sequencing (for the primer work) or from the remaining MS/MS data on the tryptic peptides generated by in-gel digestion of FT85. The positions of PCR primers shown in Fig. 4 are underlined (Note: the degeneracy of P1 and P2 and the 5Јextensions of P5-P8 and P13-P16 are not shown). A predicted TATA-box and oligo(dT) region for the FT85 ORF, and a putative polyadenylation signal for the upstream gene, are in bold upstream of the FT85 ORF; a putative polyadenylation signal is in bold downstream of the FT85 ORF. The entire PCR-confirmed sequence corresponds to GenBank TM accession AF279134.
Negligible activity was detected in extracts in the absence of pNP-LNB or Skp1(HL250), and in extracts from an induced MBP-intein-CBD expressing control strain assayed in the presence of the substrates. Detailed kinetic analyses of FT85 purified from these extracts and cleaved from its intein-CBD tag will be required to understand the basis for quantitative differences between substrates, which may result from interference by the attached intein-CBD domains. These results strongly suggest that FT85 alone, in the absence of other interacting proteins, is sufficient for fucosylation of Skp1.
The FTase Activity Maps to the C-terminal Domain of FT85-Because FT85 appears to be a multidomain protein, its N-and C-terminal regions were separately expressed in the FT85 mutant strain HW260 to determine whether either was able to reconstitute FTase activity. Coding DNA for these regions was amplified using PCR as described in Fig. 4C and cloned into a modified version of a previously described constitutive expression vector (see "Experimental Procedures"). Plasmids were electroporated into HW260 cells and transfectants were selected in 5 g/ml G418. After 2 weeks of growth, S13 extracts were prepared and assayed for FTase activity using the pNP-LNB assay described above. Extracts from cells transfected with pVFT85C encoding the C-terminal domain exhibited a 7-fold higher level of activity than wild-type Ax3 cells (Fig. 9C). The higher activity level seen in the pVFT85C-transfected extracts was probably because of multicopy integration of pVFT85C in some of the cells and the higher activity of the plasmid's discoidin promoter compared with the endogenous FT85 promoter. In contrast, extracts from parental HW260 cells and cells transfected with pVFT85N exhibited negligible activity. Similar results were seen in independently transfected cultures. In contrast, extracts expressing the C-terminal domain exhibited only very low activity with respect to Skp1(HL250). Thus the C-terminal domain of FT85 appears to contain the catalytic domain of the FTase, but another part of the protein is required to efficiently fucosylate Skp1.
Light Scattering Properties of FT85 Mutant Cells-FT85 mutant HW260 and normal Ax3 cells proliferated at similar rates in HL-5 growth medium and developed similar fruiting bodies (data not shown). However, HW260 cells appeared larger under FIG. 5-continued phase contrast microscopy. To quantify this apparent difference, the light scattering of growing cells at 488 nm was compared by flow cytometry. Forward light scatter depends on cell size and refractive index (39). HW260 exhibited a forward light scatter profile that was shifted to higher values than that of parental Ax3 cells (Fig. 10A). Similar results were seen for two other FT85 mutant clones, suggesting that the difference was specific to the loss of FT85. The GDP-Fuc Ϫ strain HL250 showed an even greater shift, possibly caused by a failure to fucosylate other acceptors in addition to FT85.
In addition, all 3 FT85 mutant strains showed an increased number of cells with larger side light scatter values compared with Ax3 (Fig. 10B). Strain HL250 was similar to the FT85 mutant strains though there were minor differences in distribution, suggesting that the change in FT85 mutant cells was because of the loss of FTase activity. Side light scattering depends in part on texture or granularity (39), which might be influenced by cell shape or organelle density.

DISCUSSION
FT85 Is the Skp1 FTase-FT85 was originally implicated as the Skp1 FTase based on its chromatographic copurification with the enzyme activity (Fig. 1) and comigration on SDS-PAGE gels with a protein that could be photoaffinity-labeled with the donor substrate analog GDP-hexanolamine-azido-125 Isalicylate (9). Tryptic peptides from gel purified FT85 were used to design degenerate oligonucleotides for use in PCR amplifications that ultimately led to the identification of FT85 genomic DNA. The in-gel digest Q-TOF-based proteomics strategy was shown to be particularly powerful here, where sample quantity was small and there is a law of diminishing returns to consider in deciding on further protein purification. Provided that sufficient doubly or triply charged peptides are studied in the tryptic mixture generated, there is a good probability, as illustrated here (Figs. 2, 3), of generating sufficient de novo sequence to initiate primer studies, and also of identifying the presence of contaminating proteins of known sequence. The unknown FT85 sequences determined were estimated to be at the few tens of femtomole level. In addition to generating de novo the peptide sequences for the FIG. 6. Family 2 GTase sequence motifs in FT85. Sequences from five prokaryotic Family 2 GTases (in blue) are aligned to show the regions of greatest similarity (motifs A-D). In red are compared sequences from the N-and C-terminal domains of FT85, and two predicted eukaryotic Family 2 GTases. An initial alignment generated by Clustal was manually refined to optimize regions of identity or similarity across 20 enzyme sequences in this group. To facilitate comparison, hydrophobic residues are green, positively charged residues are dark red, negatively charged residues are blue, and Pro and Gly are bright red; residues that are identical in the majority of sequences are in bold, and residues that are similar in the majority of sequences are highlighted. As in hydrophobic cluster analysis, A, V, L. I, M, F, Y and W are considered similar; the other similarity groups consist of structure-breaking residues (P, G), small residues (G, A, C, S, T), negatively charged residues or amides (D, E, N, Q), and positively charged residues (K, R). Asterisks refer to positions cited in the text that are occupied by highly conserved Asp residues. The prokaryotic sequences include a GTase involved in O-antigen synthesis (45) in Vibrio cholerae, a hypothetical GTase from Hemophilus influenzae, SpsA, involved in glycosylation of the spore coat of B. subtilis (40), ExoM, a ␤1,4-glucosyltransferase (41) from S. meliloti involved in succinoglycan synthesis, and a ␤1,3-GalTase (46) from Campylobacter jejuni. The eukaryotic sequences include, in addition to the two domains of Dictyostelium FT85, hypothetical GTases from C. elegans and A. thaliana. Numbers denote positions relative to the N terminus, or number of amino acids (in parentheses) between the motifs. FIG. 7. Molecular characterization of the FT85 disruption strain. A, DNA from strains Ax3 (normal) and HW260 (FT85 Ϫ ) was amplified by PCR using primers P9 and P10 and separated on a 1% agarose gel. The length of the PCR product from HW260 DNA was longer by 1.4 kilobases, corresponding to an insertion of the bsr locus. As the shorter product from Ax3 DNA amplified more efficiently, lane 3 contains 10% the amount of DNA amplified in lanes 2 and 4. B, as an indication of Skp1 FTase activity in vivo, the mobilities of Skp1s from HW260, Ax3, and HL250 (GDP-Fuc Ϫ ) were compared by SDS-PAGE of their S13 fractions followed by Western blotting using mAb 3F9 (Ref. 47; upper panel). HW260 and HL250 Skp1s migrated slightly more rapidly than Ax3 Skp1, consistent with the absence of three sugars from Skp1(HL250). The slight apparent difference between Skp1s from HW260 and HL250 can be explained by a variation in total protein migration between the 2 lanes, as determined by postblot staining of the same gel with Coomassie Blue (lower panel). initial primer studies, the MS-MS sequence analysis confirmed 25% of the predicted amino acid sequence of FT85.
To confirm that FT85 was required for Skp1 fucosylation, the FT85 locus was targeted by insertion of the blasticidin S resist-ance marker into codon 276 by homologous recombination (Figs. 4B and 7A). As a result, Skp1 was expressed at a slightly lower apparent M r (Fig. 7B), consistent with the absence of the terminal fucose-dependent trisaccharide. Metabolic labeling of Incorporation of radioactivity into pNP-LNB was determined using the C 18 -SepPak assay. Incorporation of radioactivity in the Skp1 reactions was assayed using the TCA assay. B, equal amounts of protein from S13 extracts of IPTG-induced E. coli expression strains were incubated for 2 h with substrates as in A (except Skp1(HL250) was at 0.3 M). Incorporation into pNP-LNB was assayed as in A, and incorporation into Skp1(HL250) was assayed using the SDS-PAGE method. Data values shown represent the difference between ϩ and Ϫ substrate and are representative of two independent trials. Note that the value for E. coli expressing recombinant CBD-FT85 is offscale. CBD, chitin-binding domain; MBP, maltose-binding protein. C, S13 extracts from Dictyostelium strain Ax3 (labeled A3), strain HW260 (labeled 0), or HW260 cells transfected with either pVFT85N (N-a, N-b; 2 transfections) or pVFT85C (C-a, C-b; 2 transfections), encoding the N-or C-terminal domain of FT85 as shown in Fig. 4 (Fig. 8B). Soluble extracts of HW260 had only negligible levels of FTase activity (Fig. 9A).
Expression of FT85 in E. coli as a gene fusion with an intein-CBD tag at either its N or C terminus rendered soluble extracts prepared from these cells highly active in the fucosylation of pNP-LNB and Skp1, compared with extracts of control E. coli or even Dictyostelium (Fig. 9B). Thus FT85 appeared to be both necessary and sufficient for Skp1 FTase activity, based on activity after purification (9), photoaffinity labeling (9), absence of activity in FT85 mutant cells (Fig. 9A), failure of Skp1 to be fucosylated in the FT85 mutant in vivo (Figs. 7B and 8B), ability to reconstitute activity when expressed in E. coli (Fig.  9B), and the presence of GTase-like motifs in its predicted sequence (see below).
Domain Organization of FT85-The PCR studies led to a 2304-base pair open reading frame that conceptually encodes a 768 amino acid protein with a predicted M r of 89,735 (Fig. 5), comparing favorably with FT85's apparent M r of 85,000, based on SDS-PAGE (Fig. 1). Near the middle of the protein, at positions 409 -441, lies a nearly homopolymeric stretch of Asn residues that might separate FT85 into two domains. BLAST analysis suggests that the N-terminal 260 amino acids is homologous to the full-length of numerous Family 2 GTases, and this is supported by the higher level of similarity at positions that are most conserved within the family. This includes a characteristic hydrophobic region at the very N terminus (motif A in Fig. 6), a pair of aspartates preceded by hydrophobic residues in motif B (see asterisks), a DXD-like motif preceded by hydrophobic residues in motif C, and a predicted catalytic aspartate in the C-terminal half of the domain (motif D). Sequences related to motifs A-D, with a similar spacing in the protein, can also be recognized in the C-terminal domain downstream of the Asn-rich stretch (Fig. 6). Motifs A-C are characteristic of the NRD2 domain (27) seen in GTase families 2, 23, and 27 according to the scheme of Henrissat and co-workers (23) (afmb.cnrs-mrs.fr/ϳpedro/CAZY/gtf_2.html). The NRD2 region comprises a half-domain that contains most of the residues that contact the sugar nucleotide donor, based on diffraction studies on SpsA from Bacillus subtilis (40), and site-specific mutagenesis of ExoM from Sinorhizobium meliloti (41). These similarities suggest that FT85 contains two domains that are each related to a large family of inverting GTases from eubacteria, archeabacteria, and eukaryotes, which can now be confirmed by site-specific mutagenesis of the asterisked residues in Fig. 6. Recent structural data suggest that these GTases belong to an ancient superfamily that includes GlcNAc-Tase-T1, ␤1,4GalTases, ␣1,3GalTases, and ␣GlcATases (42). Thus the sequence suggests that FT85 may be a bifunctional GTase, whose architecture resembles that of class 2 hyaluronate synthase (43) and other bifunctional prokaryotic GTases, as depicted in Fig. 4C.
This two-domain model is supported experimentally by the FTase activity studies. Previous biochemical studies showed that during purification, FT85 was susceptible to conversion to a lower M r form of about 40,000 (FT40), based on gel filtration, that retained substantial activity (9). A time-dependent conversion of FT85 to species with apparent M r values of 40,000 and 28,000, based on SDS-PAGE, was also observed when the protein preparation was photoaffinity-labeled with GDP-hexanolamine-azido-125 I-salicylate. This suggested that the FTase activity of FT85 is mediated by a catalytic domain that may be as small as M r 28,000, a typical size for catalytic domains of Family 2 GTases. These results are consistent with the present finding that expression of the C-terminal domain, starting immediately prior to the homopolymeric region of 32 Asn residues, reconstitutes in vitro FTase activity with respect to pNP-LNB when overexpressed in FT85 mutant cells (which have no detectable FTase activity). However, this domain exhibits only weak activity with respect to Gal␤1,3GlcNAc-Skp1 (Skp1(HL250)), suggesting that another part of the protein is important for efficient recognition of a protein substrate. Recently, we have obtained evidence that purified FT85 and recombinant FT85 also encode a UDP-Gal:Skp1 ␤1,3-galactosyltransferase activity, and that this activity is absent from FT85 mutant cells. 3 This activity is reconstituted by expressing the N-terminal but not the C-terminal portion of FT85. 3 Thus the C-terminal region of FT85 downstream of the polyasparagine stretch contains the catalytic domain of the FTase, but this region of the protein may act in concert with the rest of the protein for efficient processing of Skp1.
Relationship to other FTases-The specificity of the FT85 FTase for the 2-O position of Gal in Gal␤1,3HexNAc disaccharides is remarkably similar to that of the human Secretor ␣1,2-FTase (9). The length of the proposed FTase catalytic domain of FT85, about 270 residues, is similar to that of the Se-FTase, about 330 residues. However, FT85 differs from the Se-and other mammalian ␣1,2-FTases in its 1) requirement for a divalent cation and a reducing agent for FTase activity in vitro, 2) greater affinities for its donor and acceptor substrates in vitro, and 3) compartmentalization in the cytoplasm rather than the Golgi lumen and absence of a membrane anchor 3 H. van der Wel and C. M. West, unpublished data. domain (see below). Furthermore, several sequence motifs that are highly conserved in prokaryotic, microbial, and mammalian ␣1,2-FTases (38) are not recognizable in FT85. However, further study is required to determine whether the Skp1 FTase is related to known ␣1,2-FTases at the structural level.
Relationship to Other Family 2 GTases-Family 2 GTases catalyze the transfer of many kinds of sugars, including Glc-NAc, GalNAc, Glc, Gal, Rha, GlcA, abequose, and altrose, from either purine and pyrimidine sugar nucleotides by a mechanism that involves inversion of the anomeric sugar linkage. Genetic analysis of prokaryotic Family 2 GTases suggests that they modify membrane-associated lipid-linked precursors that are oriented toward the cytoplasmic compartment (44). These GTases associate either directly or indirectly with the plasma membrane via their C termini. The product glycolipids are subsequently translocated across the plasma membrane to the extracellular space where they contribute to cell wall layers, capsules, and exopolysaccharides. The Family 2 GTase domain is also embedded in cytoplasmically exposed regions of polytopic transmembrane proteins that polymerize and translocate cellulose, callose, chitin, or hyaluronan across the plasma membrane in prokaryotes and eukaryotes (15)(16)(17). In eukaryotes, this domain is also found in Dol-P-Man synthase, Dol-P-Glc synthase, and ceramide glucosyltransferase; all enzymes that are oriented toward the cytoplasm and whose products are then translocated across the membrane of the rER or Golgi (11)(12)(13). The FT85-FTase has conserved the cytoplasmic compartmentalization associated with this family, but it is not apparently associated with a membrane and its product is not translocated across a membrane. This evolutionary advancement has exposed a large, new set of cytoplasmic and nuclear proteins to potential complex O-linked glycosylation.
BLAST and motif searches of public domain data bases suggest that genes encoding enzymes similar to the Skp1 FTase may exist in other eukaryotes. The proteins predicted to be encoded by these putative genes lack motifs for targeting to the rER or other membranes and are not homologous to Dol-P-Hex synthases, ceramide glucyosyltransferases, cellulose cynthases, chitin synthases, or hyaluronan synthases. Examples of two of these putative genes, from Caenorhabditis elegans and A. thaliana, are shown in the alignment in Fig. 6. These sequences are not similar enough to suggest that they are FT85 orthologs. Another enzyme in the Skp1 glycosylation pathway, the Skp1 GlcNAcTase (8), is also related to Family 2 GTases in its NRD2 domain. 3 Role in Cell Physiology-Glycosylation is important for nuclear accumulation of Skp1 based on immunofluorescence studies on mutant Skp1s, and either the native pentasaccharide or the disaccharide produced in GDP-Fuc Ϫ mutant cells is sufficient for localization (4). Nevertheless, FT85-mediated glycosylation is important physiologically, because FT85 mutant cells exhibit increased forward and side light scattering properties suggesting increased size and granularity (Fig. 10). These changes appear to result from loss of FT85's FTase activity, as qualitatively similar changes in light scattering were observed in GDP-Fuc Ϫ mutant cells. A previous study showed that Skp1 was the most abundant fucoprotein in the cytosolic cell fraction based on metabolic labeling (36), as confirmed here (Fig. 8B).
That study also showed that Skp1 was the most abundant protein acceptor for the Skp1 FTase that accumulated in the cytosolic fraction of GDP-Fuc Ϫ mutant cells. Thus FT85-dependent fucosylation of Skp1 may affect the physical state of the cell, consistent with the emerging broad roles of this protein in cellular regulation (1,2).