Cloning of a Cholesterol-α-glucosyltransferase from Helicobacter pylori*

O-Glycans of the human gastric mucosa show antimicrobial activity against the pathogenic bacterium Helicobacter pylori by inhibiting the bacterial cholesterol-α-glucosyltransferase (Kawakubo, M., Ito, Y., Okimura, Y., Kobayashi, M., Sakura, K., Kasama, S., Fukuda, M. N., Fukuda, M., Katsuyama, T., and Nakayama, J. (2004) Science 305, 1003–1006). This enzyme catalyzes the first step in the biosynthesis of four unusual glycolipids: cholesteryl-α-glucoside, cholesteryl-6′-O-acyl-α-glucoside, cholesteryl-6′-O-phosphatidyl-α-glucoside, and cholesteryl-6′-O-lysophosphatidyl-α-glucoside. Here we report the identification, cloning, and functional characterization of the cholesterol-α-glucosyltransferase from H. pylori. The hypothetical protein HP0421 from H. pylori belongs to the glycosyltransferase family 4 and shows similarities to some bacterial diacylglycerol-α-glucosyltransferases. Deletion of the HP0421 gene in H. pylori resulted in the loss of cholesteryl-α-glucoside and all of its three derivatives. Heterologous expression of HP0421 in the yeast Pichia pastoris led to the biosynthesis of ergosteryl-α-glucoside as demonstrated by purification of the lipid and subsequent structural analysis by nuclear magnetic resonance spectroscopy and mass spectrometry. In vitro enzyme assays were performed with cell-free homogenates obtained from cells of H. pylori or from transgenic Escherichia coli, which express HP0421. These assays revealed that the enzyme represents a membrane-bound, UDP-glucose-dependent cholesterol-α-glucosyltransferase.

So far only a few functions have been ascribed to steryl glycosides. The yeast Pichia pastoris requires a functional sterol glucosyltransferase for proper degradation of peroxisomes (15,16). A mutant of the plant pathogenic fungus Colletotrichum gloeosporioides defective in the sterol glucosyltransferase shows reduced pathogenicity on avocado plants (17). Sitosteryl glucoside in cotton was suggested to serve as a precursor for the biosynthesis of sitosteryl cellodextrins to be used for the initial step of cellulose biosynthesis (18).
The cholesteryl-␣-glucosides of H. pylori support the pathogenicity of this organism, because inhibition of the cholesterol glucosyltransferase by O-glycans of the human gastric mucosa suppresses growth of the bacterium (19). Interestingly, morphological changes of the bacterium or changes in colony variants are accompanied by alterations in the total amount of steryl glucosides and the relative proportions of ␣CG, ␣CAG, and ␣CPG (20,21).
These data suggest that a better understanding of cholesteryl glucoside biosynthesis in H. pylori would be useful for further studies on host-pathogen interactions. Therefore, the aim of our study was to identify the gene encoding the cholesterol-␣glucosyltransferase in the genome of the bacterium. A single candidate gene was selected and its function in lipid biosynthesis determined by the generation of a knock-out mutant of H. pylori. The activity of the corresponding enzyme was characterized by its heterologous expression in Escherichia coli and the yeast P. pastoris with subsequent lipid analyses and in vitro enzyme assays.

MATERIALS AND METHODS
Bacterial and Yeast Strains, Growth, and Recombinant DNA Techniques-E. coli strains XL1-Blue (MRFЈ) (Stratagene) and C41(DE3) (22) were routinely grown aerobically at 37°C in Luria-Bertani medium (23). Ampicillin (100 mg⅐liter Ϫ1 ) or kanamycin (45 mg⅐liter Ϫ1 ) were included for growth of plasmid-bearing cells. The yeast strain used in this study was P. pastoris JC 308 ⌬gcs/ ⌬ugt51B1 (24), grown at 30°C in YPD medium (10 g⅐liter Ϫ1 yeast extract, 20 g⅐liter Ϫ1 peptone, 20 g⅐liter Ϫ1 glucose). For gene expression driven by the AOX1 promoter, 0.5% methanol was added to minimal medium (13.4 g⅐liter Ϫ1 Yeast Nitrogen Base). H. pylori strain P12 was grown on agar plates containing 10% horse serum in a microaerophilic atmosphere (generated by Campy-Gen, Oxoid, Basingstoke, UK) at 37°C for 48 h. Bacteria were harvested, suspended to an optical density at 550 nm (A 550 ) of 0.1 in brain heart infusion broth containing 10% heat inactivated fetal calf serum, and grown for 18 h under microairophilic conditions. H. pylori was harvested, washed twice in ice-cold phosphate-buffered saline, and the bacterial pellet was used for further lipid extraction or in vitro assays. The vectors pBluescript (Stratagene), pET24d(ϩ) (Novagen), and pPIC3.5 (Invitrogen) were used for cloning. Standard methods were followed for DNA isolation, restriction endonuclease analysis, and ligation (23).
Deletion of HP0421 in H. pylori-For the construction of the P12⌬0421 strain two 0.5-kb DNA fragments upstream and downstream of the HP0421 open reading frame (ORF) (GeneID: 900074) were amplified by polymerase chain reaction and cloned into pBluescript, separated by a chloramphenicol resistance cassette. The plasmid was introduced into H. pylori strain P12 by natural transformation. Transformation and homologous recombination was performed by harvesting bacteria from serum plates and suspending to an A 550 of 0.1 in brain heart infusion broth containing 10% fetal calf serum. DNA was added (1 g), and incubation was extended for 5 h under microaerophilic conditions before the suspension was plated on selective serum plates. Correct allelic exchange of the HP0421 gene with the resistance gene was verified by polymerase chain reaction. Fractionation of H. pylori Homogenate-To isolate cytosolic and membrane fractions of H. pylori, bacteria were resuspended in 50 mM Tris-HCl, pH 7.4, 1 mM MgCl 2 including a mixture of protease inhibitors (Complete TM , Roche) and disrupted by 4 passages through a French press at 15,000 lb/in 2 . Removal of intact bacteria by two centrifugations at 4,000 ϫ g and 4°C for 10 min resulted in a cell-free homogenate. This homogenate was subjected to centrifugation at 164,000 ϫ g at 4°C for 1 h. The supernatant represented the cytosolic fraction, whereas the pellet representing the membrane fraction was washed and resuspended in buffer (see above).
Cloning of the Cholesterol-␣-glucosyltransferase from H. pylori-The ORF sequence of the cholesterol-␣-glucosyltransferase gene HP0421 from H. pylori 26695 was amplified from genomic DNA by PCR using the specific oligonucleotide primer pairs HP0421_F (5Ј-GTACGGA-TCCACCATGGTTATTGTTTTAGTCGTGGATAG-3Ј) and HP0421_R (5Ј-ACGTGGATCCCGTGCGGCCGCTGATAAG-GTTTTAAAGAGATGGGGG-3Ј). Herculase-enhanced DNA polymerase (Stratagene) was used for the amplification of the 1169-bp product containing the entire HP0421 ORF sequence. This amplicon was digested with NcoI/BamHI and inserted into pET24d(NcoI/BamHI) leading to pET24d-HP0421, which was used for transformation of E. coli C41(DE3) cells. The amplicon was also digested with BamHI/NotI to be inserted into pPIC3.5(BamHI/NotI) resulting in pPIC3.5-HP0421 which was used to transform P. pastoris. Exact in-frame cloning and identity of the PCR-cloned fragments were confirmed by sequencing.
Heterologous Expression of the ORF HP0421 in E. coli and P. pastoris-E. coli C41(DE3) (22) was used as expression host for the plasmid pET24d-HP0421. After transformation, 50-ml cultures of E. coli were grown at 37°C to an A 600 of 0.8 -1.2. Induction was performed by adding 0.4 mM isopropyl ␤-D-thiogalactoside and further incubation for 3 h at 30°C. A P. pastoris strain deficient in glycolipid biosynthesis was used to express HP0421: JC 308 ⌬gcs::Sh ble, ⌬ugt51B1::URA3, ade1, arg4, his4 (24). In a first step, this strain was transformed with the empty vectors pBLADE and pBLARG (25) to increase its rate of growth. Subsequently, the resulting strain, DKO (JC 308 ⌬gcs::Sh ble, ⌬ugt51B1::URA3, his4) was used for transformation with pPIC3.5-HP0421 leading to DKO-Hp. Transformed cells were grown at 30°C in 50 ml of YPD medium to an A 600 between 1 and 2. Expression was driven by the strong AOX1 promoter and induced by transferring the cells to minimal medium with 0.5% methanol as sole carbon source followed by additional incubation for 20 h.
In Vitro Sterol Glucosyltransferase Assay-E. coli cells were harvested by centrifugation (4°C, 10 min, 3200 ϫ g), resuspended in 2-3 ml buffer (50 mM Tris-HCl, pH 7.5, 5 mM 1,4-dithiothreitol), and cooled in an ice bath. Disruption of E. coli cells was performed by ultrasonication (probe tip, 10 times for 10 s). Cell debris were removed by centrifugation (4°C, 10 min, 3200 ϫ g), and the supernatant fractions representing the cellfree homogenates were used for in vitro assays. The assay for the cholesterol-␣-glucosyltransferase activity was performed similar to the assay used for the eukaryotic sterol-␤-glucosyltransferases (3,26). Shortly, the assay mixture contained in a total volume of 100 l: . After incubating for 30 min at 30°C the reaction was terminated by the addition of 0.7 ml of 0.45% NaCl solution and either 2 ml of ethyl acetate or 3 ml of chloroform/methanol 2:1. After vortexing and phase separation by centrifugation, the radioactivity in the organic phase was determined by scintillation counting or the extracted lipids were separated by thin-layer chromatography (TLC) or high performance TLC (HPTLC). The radioactivity on the silica gel plate was detected by radioscanning with a BAS-1000 Bio Imaging Analyser (Raytest, Straubenhardt, Germany).
Lipid Extraction and Analysis-P. pastoris or H. pylori cells were harvested by centrifugation (4°C, 10 min, 3200 ϫ g) and the sedimented cells were boiled for 10 min in a water bath. Lipid extraction was performed with chloroform/methanol 1:2 (v/v) and chloroform/methanol 2:1 (v/v). The lipid extract was washed by Folch partitioning (27) (chloroform/methanol/ 0.45% NaCl solution, 2:1:0.75) and the organic phase was evaporated. The residue was redissolved in pyridine (because of the low solubility of steryl glucosides in chloroform/methanol mixtures) and subjected to analytical and preparative TLC. For NMR spectroscopy and mass spectrometry (MS), the purified glycolipids were acetylated (with acetic anhydride in pyridine, 1:1) overnight at room temperature and subjected to repurification by preparative TLC in diethyl ether.
Electrospray Ionization Fourier Transform Mass Spectrometry-High resolution electrospray ionization Fourier-transform mass spectrometric (ESI FT-MS) analyses of acetylated glycolipids were performed in the positive and/or negative ion mode using an APEX II-Instrument (Bruker Daltonics, Billerica, MA) equipped with an actively shielded 7 Tesla magnet and an Apollo II ESI source. Mass spectra were acquired using standard experimental sequences as provided by the manufacturer. For the positive ion mode samples (ϳ10 ng⅐l Ϫ1 ) were dissolved in a 50:50:0.03 (v/v/v) mixture of 2-propyl alcohol, water, 30 mM ammonium acetate adjusted with acetic acid to pH 4.5. For the negative ion mode samples (ϳ10 ng⅐l Ϫ1 ) were dissolved in a 50:50:0.001 (v/v/v) mixture of 2-propyl alcohol, water, and triethylamine (pH 8.5). The samples were sprayed at a flow rate of 2 l⅐min Ϫ1 . Capillary entrance voltage was set to 3.8 kV, and drying gas temperature to 150°C. Because the spectra comprise the molecular species in different charge states, the spectra were charge-deconvoluted, using the XMASS-6.1 software, and mass numbers given refer to monoisotopic masses of the neutral molecules.

A Single Candidate Gene in the Genome of H. pylori Was Identified as the Putative Cholesterol-␣-glucosyltransferase Gene-
To discover putative cholesterol-␣-glucosyltransferase genes in H. pylori, a BLAST data base search (29) with the amino acid sequence of the sterol ␤-glucosyltransferase from Arabidopsisthaliana (1) was performed. No sequence of significant similarity was found in the genome of H. pylori. This result was not surprising considering that glycosyltransferases operate with two different catalytic mechanisms and are classified according to the stereochemistries of the reaction substrates and products as either retaining or inverting enzymes (30). While the plant ␤-glucosyltransferase belongs to the inverting enzymes, the ␣-glucosyltransferase from H. pylori probably works by the retaining mechanism.
As a result, candidate cholesterol-␣-glucosyltransferase genes from H. pylori had to be identified by a different approach. Cholesterol biosynthesis is a typical eukaryotic fea-ture and this lipid is available for bacteria only from eukaryotic sources. Therefore, we speculated that the cholesterol-␣-glucosyltransferase from H. pylori may have developed from an ancient bacterial ␣-glycosyltransferase of originally different acceptor specificity. This hypothesis is supported by a similar situation in another group of enzymes, the acyl-CoA-dependent acyltransferases. Whereas an acyl-CoA-dependent cholesterol acyltransferase had been cloned and characterized in 1993 (31), all efforts to clone a diacylglycerol acyltransferase failed for many years. This important enzyme was finally identified, since a protein with amino acid sequence similarities to the cholesterol acyltransferase turned out to transfer the acyl group not to cholesterol, but to diacylglycerol (32). Thus, cholesterol and diacylglycerol acyltransferases have a common ancestor.
By extrapolating this cholesterol/diacylglycerol correlation to glycosyltransferases, we concluded that the H. pylori cholesterol-␣-glucosyltransferase might have developed from a bacterial diacylglycerol-␣-glycosyltransferase. Indeed, such diacylglycerol ␣-glucosyl and galactosyltransferases have been identified in Acholeplasma laidlawii, Streptococcus pneumoniae, Deinococcus radiodurans, and Thermotoga maritima (33,34). A BLAST search with the sequence of the diacylglyc- For the alignment with the glycosyltransferase from H. pylori studied in the present work, eight diacylglycerol glycosyltransferases, and monoglycosyldiacylglycerol glycosyltransferases, respectively, from the GT4 were selected (boxed names) (33,34,(41)(42)(43). Three additional sequences represent hypothetical polypeptides, which show the highest similarities to the glycosyltransferase from H. pylori. The dendrogram has been constructed from pairwise similarities of amino acid sequences using ClustalX (44). The sequences used for the alignment are present at the Carbohydrate-Active enZYmes server and are from: A. laidlawii erol ␣-glucosyltransferase from A. laidlawii (33) revealed a single sequence in H. pylori with low but significant similarity (42% similarity, 19% identity). This hypothetical protein, HP0421, was the most promising candidate to be tested for cholesterol-␣-glycosyltransferase activity. The selection of this candidate was checked by a survey of hypothetical glycosyltransferases at the Carbohydrate Active enZYme server (35). This data base identified 21 hypothetical glycosyltransferase genes in the genome of H. pylori 26695 (23 in H. pylori J99), from which 5 (7 in H. pylori J99) represent retaining glycosyltransferases. The hypothetical protein HP0421 consists of 389 amino acids, is a member of the glycosyltransferase family 4 (GT4), and shows sequence similarities to several members of the family with diacylglycerol glycosyltransferase activity (Fig.  2). To elucidate the function of HP0421, we generated a knockout mutant of H. pylori and expressed HP0421 in E. coli and P. pastoris.
Deletion of the HP0421 Gene in H. pylori Led to the Loss of Cholesteryl-␣-glucoside and Its Three Derivatives-A HP0421 knock-out strain of H. pylori P12 was generated by insertion of a chloramphenicol resistance cassette. The deletion of HP0421 in H. pylori resulted in the loss of cholesterol-␣-glucosyltransferase activity determined by in vitro enzyme assays with homogenates of H. pylori cells and radiolabeled substrates (Fig. 3A). In addition, TLC of lipid extracts of the cells revealed that 4 glycolipids present in wild-type extracts were absent in the extracts of the knock-out mutants (Fig. 3B). Three of them, ␣CG, ␣CAG, and ␣CPG have been identified before (9). Analysis of the purified and peracetylated lipids by MS and NMRspectroscopy confirmed their structures (data not shown). On the other hand, the compound with the lowest R f value had not been described previously. Therefore, it was isolated, and its structure elucidated by MS. The mass spectrum (supplemental Fig. S1) comprised two abundant and one minor molecular species the masses of which were in agreement with the mass of peracetylated cholesteryl-lysophosphatidyl-hexoside which carried either a C 18:1, a cyclopropane C 19:0 or a C 14:0 fatty acid (supplemental Fig. S1). From these data we conclude that the glycolipid of H. pylori with the smallest R f value represents cholesteryl-6Ј-O-lysophosphatidyl-␣-glucopyranoside derived from ␣-CPG by the loss of one fatty acid residue. The deletion of HP0421 in H. pylori and lipid analysis of this mutant demonstrate that this gene is essential for the biosynthesis of ␣CG and its three derivatives in H. pylori.
Heterologous Expression of HP0421 in P. pastoris Resulted in the Biosynthesis of Ergosteryl-␣-glucoside-P. pastoris constitutively synthesizes ceramide, diacylglycerol, and sterols either as intermediates or end products of lipid metabolism. Therefore, P. pastoris represents an appropriate host for the expression of glycosyltransferases, which require one of these lipids as glycosyl acceptors (28, 34, 36 -38). We used a double null mutant of P. pastoris (⌬gcs/⌬ugt51B1; (24), which is devoid of both steryl glucosides and glucosyl ceramides to express HP0421 from H. pylori. Lipid analysis of the transformed cells showed the formation of a new glycolipid  in the strain expressing HP0421 (Fig. 4). This glycolipid was isolated, peracetylated and appeared as two partially separated compounds after purification by TLC, which were subjected to MS and NMR spectroscopy.
A parallel expression of a bacterial GT4 member from T. maritima in P. pastoris led to the biosynthesis of diacylglycerol-␣glucoside as described before (34). These data demonstrate that HP0421 from H. pylori is a sterol-␣-glucosyltransferase.
HP0421 Is a Membrane-bound UDP-glucose-dependent Cholesterol-␣-glucosyltransferase-Further functional characterization of the cholesterol-␣-glucosyltransferase from H. pylori was performed by in vitro determination of enzyme activity. E. coli cells expressing HP0421 were disrupted by ultrasonication and the sterol glycosyltransferase activity in cell-free homogenates was measured using radiolabeled substrates. After HPTLC of the lipophilic reaction products, radioscanning revealed that cholesteryl-␣-glucosides were synthesized, which differ from the well known steryl-␤-glucosides by a smaller R f value (Fig. 5). The clear resolution of the two anomeric steryl glucosides was achieved by HPTLC. Equivalent separations were observed before with diacylglycerol glycosides (34).
In the sterol glucosyltransferase assay, the recombinant HP0421 exhibited a pH optimum of 8.5, and the apparent K m for UDP-Glc was 25 M. Table 2 shows the substrate specificities of the recombinant enzyme HP0421 for glucosyl acceptor substrates. Despite the fact that the assays were performed under linearized conditions, these data should not be considered as a quantitative determination of substrate affinities. Because the lipophilic substrates in the assay were not uniformly dissolved, but were present as both monomers and components of lipid vesicles, the exact determination of their concentrations in the assay was not possible. Therefore, these results indicate which substrates are accepted or discriminated by the cholesterol-␣-glu-  cosyltransferase. With UDP-[ 14 C]glucose as the donor, the glycosyltransferase used various sterols as sugar acceptors, but neither ceramide nor diacylglycerol were glucosylated. Under linearized assay conditions the enzyme showed a distinct preference for cholesterol. Whereas all other sterols led to at most 13% of the incorporation into steryl glucosides compared with cholesterol. Despite this preference for cholesterol, the enzyme is able to synthezise substantial amounts of ergosteryl-␣-glucoside in transgenic P. pastoris expressing HP0421 (Fig. 4).
To assess the intracellular localization of the cholesterol-␣glucosyltransferase, a cell-free homogenate from H. pylori cells was centrifuged for 1 h at 164,000 ϫ g, and the cholesterol glucosyltransferase activity was determined in the supernatant (cytosolic fraction) and in the resuspended pellet (membrane fraction). The activity of the cholesterol glucosyltransferase was present in the homogenate of H. pylori cells and in the membrane fraction, but was barely detectable in the cytosolic fraction (Fig. 5). In summary, these in vitro data demonstrate that HP0421 from H. pylori represents a membrane-bound UDPglucose-dependent cholesterol-␣-glucosyltransferase.

DISCUSSION
To identify the enzyme, which is responsible for the biosynthesis of cholesteryl-␣-glucoside in H. pylori, we have employed a strategy based upon both insertional inactivation in H. pylori and heterologous expression in E. coli and P. pastoris. This has led to the characterization of HP0421, a glycosyltransferase of the GT4 family (35), the members of which display a retaining chemistry. The complete loss of cholesteryl-␣-glucoside and its three derivatives in the HP0421-inactivated mutant showed that this enzyme is solely responsible for cholesterol ␣-glycosylation in H. pylori (Fig. 3). Heterologous expression of HP0421 in P. pastoris led to the biosynthesis of steryl-␣-glucoside, but not of diacylglycerol-␣-glucoside. In contrast, other bacterial enzymes of GT4 e.g. TM0744 from T. maritima synthesized diacylglycerol-␣-glucoside, but not steryl-␣-glucoside (Fig. 4) (33, 34). The unique activity of HP0421 from H. pylori was confirmed by in vitro assays, because the enzyme glucosylated different sterols with a preference for cholesterol. In contrast, diacylglycerol or ceramide were not accepted by HP0421. However, the sequence similarities of HP0421 to bacterial diacylglycerol-␣-glucosyltransferases support the assumption that HP0421 might have developed from such an enzyme after a new substrate, cholesterol, was available for the pathogenic bacterium. Like other bacteria, H. pylori is not able to synthesize cholesterol, but it is able to acquire the lipid from epithelial cells of its host organisms. 5 Because the bacterial cholesterol glucosyltransferase has been proposed to be accessible by extracellular ␣-1,4-GlcNAccapped O-glycans (19), we assessed the intracellular localization of the enzyme by differential centrifugation. The activity of the cholesterol glucosyltransferase was found in total H. pylori lysates and in the membrane fraction, but was barely detectable in the cytosolic fraction. These data are in agreement with the assumption that an enzyme using a membrane lipid, cholesterol, as substrate would be membrane-bound. On the other hand, it is doubtful whether HP0421 contains a transmembrane domain, because the results of various prediction methods at the ExPASy Molecular Biology server are contradictory (data not shown). Berg et al. (33) found similar inconsistencies with the DAG-␣-glucosyl-transferase from A. laidlawii. Fold predictions indicated several amphipathic ␣-helices, which may attach the enzyme to negatively charged domains of the bacterial membrane (33). Regardless of the mechanism of membrane attachment/integration, it is not yet clear whether the enzyme is present in the plasma membrane or the outer membrane or both. Because of its UDP-glucose-dependence it seems likely that the enzyme localizes to the cytoplasmic face of the plasma membrane. Regarding the inhibitory effect of ␣-1,4-GlcNAccapped O-glycans on the cholesterol-␣-glucosyltransferase, we suggest either a signaling effect by O-glycans, or the internalization of O-glycans by the bacterium to permit access to the enzyme.
The biosynthesis of cholesteryl-␣-glucoside is of particular importance for H. pylori, since it allows the pathogen to evade the immune response of the mammalian host, and H. pylori mutants lacking cholesteryl glucosides show impaired infection in a murine model. 5 Therefore, the identification of the cholesterol-␣-glucosyltransferase provides the basis for further studies on a new aspect of the interaction between the pathogenic bacterium and its hosts. The availability of the DNA sequence of HP0421 allows the manipulation of cholesteryl glucoside synthesis in H. pylori. On the other hand, we demonstrated the formation of recombinant HP0421 protein in two different expression hosts. These expression systems will be used for the purification of the recombinant protein, which could be subjected to a high throughput screen for the identification of specific inhibitors of cholesterol-␣-glucosyltransferase activity (40).