JBC

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Melchers, K.
Right arrow Articles by Schäfer, K. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Melchers, K.
Right arrow Articles by Schäfer, K. P.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Volume 271, Number 1, Issue of January 5, 1996 pp. 446-457
©1996 by The American Society for Biochemistry and Molecular Biology, Inc.
Cloning and Membrane Topology of a P type ATPase from Helicobacter pylori(*)

(Received for publication, August 15, 1995; and in revised form, September 12, 1995)

Klaus Melchers (§),  ,  Thomas Weitzenegger ,  Anita Buhmann ,  Wolfram Steinhilber ,  George Sachs(¶), (**),  ,  Klaus P. Schäfer

From the From Byk Gulden Pharmaceuticals, Department of Molecular Biology, D-78462 Konstanz, Germany andUCLA/Wadsworth VA Hospital, Los Angeles, California 90073

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES

ABSTRACT

Southern blot screening of a genomic Helicobacter pylori library was employed to find a P type ATPase using a mixture of 16 DNA oligonucleotides coding for the DKTGT(I/L)T consensus sequence specific for the phosphorylation site of this family of ATPases. A positive clone, pRH439, was isolated and sequenced. The inserted 3.4-kb H. pylori DNA contained an intact open reading frame encoding a protein of 686 amino acids carrying the consensus sites for phosphorylation and ATP binding. The amino acid sequence exhibits a 25-30% identity with bacterial Cd and Cu ATPases. Genomic Southern blot analysis showed that this ATPase was present in all H. pylori strains examined, whereas it was not detectable in Campylobacter jejuni and other bacteria. The membrane topology of this ATPase was investigated using in vitro transcription/translation of fusion vectors to find signal anchor and/or stop transfer sequences. Eight regions of the H. pylori ATPase acted as signal anchor and/or stop transfer sequences and were ordered pairwise along the polypeptide chain placing the N and C-terminal amino acids in the cytoplasm. These transmembrane segments are contained between positions 73 and 92 (H1), 98 and 125 (H2), 128 and 148 (H3), 149 and 176 (H4), 309 and 327 (H5), 337 and 371 (H6), 637 and 658 (H7), and 659 and 685 (H8). The membrane domain of the ATPase, therefore, consists of at least four pairs of transmembrane segments with the phosphorylation site and ATP binding domain located in the large cytoplasmic loop between H6 and H7. The cytoplasmic domain contains several histidines and cysteines, perhaps indicative of divalent cation binding sites. There are several charged amino acids (3 Lys, 2 Glu, 2 Asp) predicted to be in the membrane domain mainly in H2, H3, and H4 and a Cys-Pro-Cys putative metal ion site in H6. The extracytoplasmic domain also has several charged amino acids (5 Glu, 1 Asp, 1 Lys, 1 Arg). It is likely that this novel protein is a heavy metal cation transporting ATPase and belongs to a family of P type ATPases containing eight transmembrane segments.

INTRODUCTION

P type ion motive ATPases are widely distributed in nature, occurring in bacteria, fungi, plants, and animals(1, 2, 3, 4) . They are polytopic integral membrane proteins, with usually one but occasionally more subunits(2) . An aspartate residue, which forms the beta aspartyl phosphate intermediate during the catalytic cycle, resides in a conserved DKTGT(I/T)T consensus sequence identified in all known P type sequences(1, 2, 3) . In contrast to this phosphorylation signature sequence and another sequence within the ATP binding domain, P type ATPases diverge widely in their amino acid sequences, in part reflecting their functional diversity in ion transport. Interpretation of hydropathy suggests that perhaps they can be divided into two groups, those containing eight and those containing 10 transmembrane helices.

Helicobacter pylori is a human gastric pathogen associated with peptic ulcer disease as well as chronic gastritis, which may predispose to gastric cancer(5, 6, 7, 8, 9, 10) . Several virulence factors are known to play a role in pathogenicity of H. pylori infections(6, 8, 9, 11) , but the means whereby H. pylori colonizes and survives in the human stomach is poorly understood. One possibility is the expression of selected P type ATPases. P type ATPases are involved in many bacterial functions such as maintenance of pH, turgor pressure, and intracellular ion composition(4, 12, 13) . The ions transported by P type ATPases of bacteria vary due to need for selective adaptation to a varying environment.

The first bacterial P type ATPase described was the three-subunit Kdp ATPase, a high affinity uptake transporter for K in Escherichia coli(14) . The ion-translocating activity of another putative K ATPase (15, 16) found in Enterococcus hirae (formerly Streptococcus faecalis) is now thought to have a role in H export(17) . This enzyme is composed of a single subunit of 78-kDa based on purification and reconstitution. Another E. hirae P type ATPase, CopB, was one of the earliest single subunit P type ATPases cloned(18) . Additional bacterial P type ATPases are now being cloned and expressed. The copper pumps, CopA and CopB(18, 19, 20) , and the Cd-ATPase (21) have significant homology to Menkes gene and other putative copper-transporting ATPases in mammalian cells(22, 23, 24, 25, 26, 27, 28) . One or more of these ATPases might be essential for the survival of H. pylori in the human stomach.

We used the homology in the phosphorylation consensus sequence for isolation of an H. pylori gene encoding a P type ATPase. The DNA predicts a 75-kDA protein including the conserved sites of phosphorylation and ATP binding. It has significant homology to the bacterial Cu and Cd transporting ion pumps (18, 19, 20, 21) as well as to the eukaryotic Cu pumps(26, 27, 28) . The membrane domain of the protein contains the ion transport pathway. The mammalian pumps, such as the SERCA family of Ca-ATPases(29) , the Na,K-ATPases(30) , and the gastric H,K-ATPase (31) are often expressed at sufficient levels to allow biochemical methods to be used for finding transmembrane segments. Bacterial P type ATPases are not expressed in large quantity, and a fusion protein approach was taken to define the 10 membrane segments of the Mg-ATPase of Salmonella typhimurium(32) . A fusion protein vector system in pGEM7sz+ has been developed that allows scanning of the membrane insertion properties of putative transmembrane segments of any protein by in vitro translation(33) .

In this work, we have cloned an H. pylori P type ATPase using a DNA oligonucleotide probe targeted to the phosphorylation consensus sequence of these pumps. Southern blot analysis indicated that the encoded ion pump may be unique to H. pylori. A variety of hydropathy algorithms predict six to eight transmembrane helices (H1 to H6 or H8) (^1)for this enzyme. In vitro transcription/translation, using a fusion vector containing putative transmembrane segments, showed the presence of at least eight transmembrane segments in this protein. Protein sequence suggests that this ATPase contains divalent cation binding sites and is a cation-transporting ATPase.

EXPERIMENTAL PROCEDURES

Gene Isolation and Cloning

A DNA oligonucleotide I-282, representing 16 possible sequences encoding the conserved phosphorylation signature box of P type ATPases, DKTGT(I/L)T, was used for screening of an H. pylori gene library. The nucleotide sequence was GAT AAA AC(A/C) GGC AC(C/G) (A/T)T(C/G) AC. Variable nucleotide positions are in parentheses. The DNA primer was labeled with a digoxigenin 3`-end-labeling kit (Boehringer Mannheim).

H. pylori Library

The library was constructed by Dr. R. Haas (Tübingen). It contains DNA fragments of H. pylori 69A from a partial Sau3A digest cloned into the vector pRH160 in E. coli HB101. The vector contains a tetracycline resistance gene. DNA inserts are on an average 3-4 times 10^3 base pairs in length and can be isolated by digestion with EcoRI and XhoI.

Preparation of Plasmid DNA Mixtures

An aliquot of the library, 1.4 times 10^3 colony forming units, representing about 2 genomic equivalents of H. pylori, was diluted in LB medium supplemented with 50 mg/ml tetracycline for inoculation of 20 cultures for subsequent mixed plasmid preparations. Inoculated vials, each containing approximately 70 different plasmids of the library, were incubated in a shaker incubator at 37 °C overnight for bacterial growth. Plasmid DNA was prepared (Qiagen) and digested by EcoRI and XhoI. DNA fragments were separated on agarose gels and blotted onto nylon membranes. In parallel, glycerol stocks were prepared from each mixed bacterial suspension.

Hybridization with Plasmid DNA and Purification of the P Type ATPase Clone

Southern blots containing restriction-digested DNA of mixed plasmid preparations were hybridized with digoxigenin-labeled DNA oligonucleotide I-282 according to a Boehringer Mannheim protocol for >6 h in 5 times SCC, 1 times blocking agent, 0.02% (w/v) SDS, 0.1% (w/v) N-lauroylsarcosine at 50 °C. Blots were washed in 5 times SSC, 0.1% (w/v) SDS at 53 °C. Positive plasmid mixtures were detected by chemiluminescence (Boehringer Mannheim). Aliquots taken from corresponding glycerol stocks were plated out on agar plates containing 50 µg/ml tetracycline, and 60 colonies were selected for preparation of plasmid DNA using anion exchange columns (Qiagen). Plasmid DNA was analyzed for sequences homologous to the primer I-282 by Southern blot analysis as described above.

DNA Sequencing

The DNA sequence of the entire insert of pRH439 was determined using the cycle sequencing procedure performed with the digoxigenin Taq polymerase DNA sequencing kit (Boehringer Mannheim). Digoxigenin-labeled sequencing primers, corresponding to selected regions along the nucleic acid sequence of the EcoRI-XhoI DNA insert of pRH439, were used. 25 cycles at 94 °C for 1 min, 40 °C for 1 min, and 72 °C for 1.5 min were performed in a thermal cycler, the temperature being maintained at 72 °C for 10 min at the end of the last cycle. Cycle sequencing reactions were subjected to direct blotting gel electrophoresis (GATC Konstanz) for DNA sequence determination.

Extraction of Chromosomal DNA from Bacterial Strains

Genomic DNA from several H. pylori strains and from H. felis, C. jejuni, Proteus vulgaris, and E. coli MM294 was prepared as described previously(34) . The DNA of a clinical isolate, H. pylori 69A was a gift of R. Haas (Tübingen).

Bacterial Strains

All bacterial strains used in this study were supplied by the American Type Culture Collection (ATCC). H. pylori strains used were ATCC 49503, ATCC 43526, ATCC 51110, ATCC 43629, ATCC 51111, ATCC 43504, and ATCC 43629. Additional strains used for comparative studies were H. felis (ATCC 49179), P. vulgaris (ATCC 13315), E. coli MM294 (ATCC 33625), and C. jejuni (ATCC 33560).

Culture Conditions

H. pylori was grown in brain heart infusion medium (Difco) supplemented with 6% horse serum and 0.25% yeast extract (Difco) in a CO(2) incubator (10% CO(2)) at 37 °C in 10-ml cell culture flasks. H. felis was grown in Columbia EH broth (Difco) supplemented with 6% horse serum in gas pak jars under microaerophilic conditions (Anaerocult C, Merck) at 37 °C in a shaker incubator. C. jejuni and P. vulgaris were grown as described for H. pylori. E. coli was cultured in LB broth for isolation of chromosomal DNA or plasmid DNA.

M0/M1 Vector Construction

The M0 and M1 cloning vectors are derivatives of the plasmid pGEM7zf+(DeltaHindIII) where the HindIII site has been removed (Promega, Madison, WI). These were originally used for detection of membrane-spanning segments of the gastric H,K-ATPase using in vitro transcription and translation and have been described in detail(33) . The N-terminal part of fusion proteins encoded by the vectors consists either of the first 101 (M0) or the first 139 amino acids (M1) of the alpha subunit of the gastric H,K-ATPase. The C-terminal region of both the M0 and M1 fusion protein contains the 177 terminal amino acids of the beta subunit of the gastric H,K-ATPase, which has five potential glycosylation sites. The M0 vector is used for detection of signal anchor sequences; the M1 vector, containing a single transmembrane segment, is used for detection of stop transfer sequences. The alpha and beta subunit regions of the fusion proteins are fused in frame by a short linker segment of DNA encoding four amino acids, which can be exchanged for any sequence of interest.

To insert amplified DNA fragments of the H. pylori P type ATPase, the DNA linker fragment of the M0 or M1 vector was first removed by BglII/HindIII restriction digestion. This was followed by in-frame insertion of selected DNA fragments carrying BglII/HindIII recognition sites at their 5`- and 3`-ends, obtained by PCR amplification of the H. pylori DNA with appropriate primers. These vectors are suitable for transcription-coupled in vitro translation of fusion proteins under control of the pGEM7zf+ T7 promoter.

DNA Oligonucleotides Used for PCR Amplification of Selected Sequences of the H. pylori P Type ATPase

The selection of regions in the H. pylori pump representing putative transmembrane-spanning sequences was based on hydropathy analysis using various algorithms(35, 36, 37, 38) . Selected putative transmembrane regions were amplified by PCR using appropriate DNA primers. Sense primers were extended by a preceding DNA (5`-GTCTACCTAGATCTC-3`) sequence containing a BglII restriction site for cloning, while antisense primers contained an extension (GCTCTACAAGCTT) including a HindIII restriction site for cloning into the M0 and M1 vectors. DNA oligonucleotides were targeted to DNA sequences encoding the first six (sense) or last six (antisense) amino acids of a segment of the H. pylori enzyme to be amplified. The corresponding amino acid sequences selected as target sites along the P type ATPase gene are listed in Table 1. DNA oligonucleotides were supplied by MWG-Biotec (Ebersberg, Germany).

Polymerase Chain Reaction

Amplification of the DNA was carried on pRH439 plasmid DNA by performing PCR in 60 mM Tris/HCl, pH 9.0, 15 mM (NH(4))(2)SO(4), 2.5 mM MgCl(2), 0.25 µM each dNTP, 100 pg of plasmid DNA in the presence of 0.07 units/ml Taq polymerase reaction volume in a Perkin-Elmer thermal cycler (Cetus model). Cycling was carried out at 94 °C for 1 min, 37 °C for 1 min, and 72 °C for 3 min including a 10-min extension at 72 °C during the last cycle. PCR products were purified using PCR purification kits (Qiagen). Purified PCR products representing selected regions of the H. pylori pump were ligated into the M0 and M1 vectors as detailed below.

DNA Cloning and Construction of PY Vectors

PCR products were digested with BglII and HindIII restriction enzymes. DNA fragments were separated on agarose gels for purification using a PCR purification kit (Qiagen). For construction of the PY vectors, the purified PCR fragments were ligated into BglII/HindIII-digested M0 and M1 vectors using T4 DNA ligase (Boehringer Mannheim). Ligated plasmid DNA was transformed and replicated in E. coli HB101. Cloned vectors were isolated using anion exchange chromatography (Qiagen). The DNA sequence inserted into M0 and M1 vector was verified by DNA dideoxy sequencing. The PY vectors containing different putative transmembrane segments of the H. pylori ATPase are listed in Table 1.

In Vitro Transcription/Translation

The M0 or M1 plasmids containing the DNA insertions encoding the putative transmembrane segments of the H. pylori P type ATPase (PY vectors) were used for synthesis of the corresponding S-labeled fusion protein products. In vitro protein synthesis was carried out using a troponin T-coupled reticulocyte lysate system in the absence or presence of canine pancreatic microsomes (Promega) according to the manufacturer's protocol. In each transcription/translation experiment both the M0 and M1 vectors without any insertions were used to control the efficiency of the reaction and to verify the quality of membranes used. After translation, products synthesized in the presence of membranes were purified by centrifugation through 250 mM sucrose, 50 mM Tris/HCl, pH 7.5, and the pellets were resuspended in 0.5% SDS, 10 mM Tris/HCl, pH 7.5.

SDS-PAGE and Autoradiography

The S-labeled protein synthesized in vitro was separated electrophoretically on 10% SDS-polyacrylamide gels(39) . Products synthesized in the absence and presence of membranes using the same vector were run on lanes next to each other. Because the translation efficiency was decreased when membranes were present, those experiments were analyzed on gels using 2 to 10 times more material than in the absence of membranes. Bio-Rad M(r) standards (6,500-200,000) were used to calibrate the M(r) of the translation products. After electrophoresis, gels were equilibrated in 50% methanol, 10% acetic acid and then stained in the same buffer supplemented with 0.1% Coomassie Brilliant Blue R-250. Subsequently gels were incubated in 20% methanol, 7.5% acetic acid before soaking for 30 min in Amplify solution (Amersham Corp.) to prepare the gels for autoradiography.

Gels were dried and placed into cassettes. X-ray films (Hyperfilm MP, Amersham) were exposed for 6-24 h at -80 °C. The presence of glycosylation due to a signal anchor sequence was seen by the increase in M(r) of the product when the M0 vector with insert was translated in the presence of membranes. The presence of a stop transfer sequence was shown by the inhibition of normal glycosylation of the M1 vector in the presence of membranes.

Materials

The chemicals used were all analytical grade or higher. Molecular biological reagents were obtained from Amersham, Qiagen, Promega, and Boehringer Mannheim as specified.

RESULTS

ATPase Cloning

Detection and Isolation of pRH439

The DNA oligonucleotide I-282 was hybridized on Southern blots carrying restriction-digested DNA of H. pylori and E. coli. The probe gave a strong signal with H. pylori DNA, but blotting of E. coli DNA also resulted in a prominent hybridization signal (data not shown). Since E. coli was positive with I-282 we did not use the conventional colony filter hybridization assay for gene library screening, anticipating many false positives. Instead, the isolated plasmid DNA was used in order to prevent the background due to E. coli genes. The H. pylori genomic DNA library in pRH160 was isolated from E. coli. The inserted DNA was screened using the oligonucleotide I-282 under the same conditions used in the genomic Southern blot analysis. Screening resulted in the isolation of a positive clone, pRH439, containing an H. pylori DNA fragment 3.4 kilobase pairs in length.

DNA Sequence Analysis of pRH439

DNA sequencing of the entire H. pylori insert provided a complete open reading frame encoding a protein of 686 amino acids with a predicted M(r) of 74,937 kDa (Fig. 1). Along the DNA sequence, the 75-kDa ORF comprises the nucleotide sequence starting at position 1214 and ending at nucleotide 3271. The 5`-flanking DNA sequence of the 75-kDa ORF carries an additional ORF, disrupted by the BglII-Sau3A cloning sequence, which ends with the UAA stop codon at positions 1187-1189. Located between the truncated ORF and the 75-kDa ORF, starting 7 nucleotides upstream of the 75-kDa ORF ATG initiation codon, there is a AGGGA sequence resembling a Shine/Dalgarno sequence element as found in other isolated genes of H. pylori(40, 41, 42, 43) . In the 3`-region of the inserted DNA, near the EcoRI restriction site of the pRH160 cloning vector, there is a potential termination sequence.

Figure 1: DNA sequence and predicted amino acid sequence of pRH439 isolated from H. pylori DNA library. The figure shows the 3,399-bp H. pylori DNA sequence of the Sau3A insertion of pRH439 as determined by DNA sequencing. There is one complete open reading frame beginning with an ATG translational start codon at positions 1,214-1,216 predicting a putative P type ATPase 686 amino acids in length. Amino acids are shown by single letter code. The 5` adjacent sequence of the 74,937-kDa open reading frame contains an AGGGA ribosome binding site (underlined). Potential transcriptional termination sequences are contained in the 3`-flanking DNA sequence (underlined). The putative P type ATPase gene is preceded by an additional open reading frame, which is disrupted by the BglII-Sau3A restriction site (not translated).


Sequence Analysis of the 75-kDa ORF-encoded Protein

The predicted amino acid sequence of H. pylori P type ATPase exhibits two sequence motifs typical for this type of pump, namely the signature sequence of phosphorylation with the aspartate in amino acid 388 and the conserved GDGXNDXP region in the ATP binding domain (GDGINDAP, positions 582-589). A feature of the DNA-deduced amino acid sequence of the novel H. pylori protein is the presence of 11 cysteine residues (1.6%) and 14 histidines (2%) mainly clustered in the N-terminal region and the central region of the enzyme around the predicted sites of phosphorylation and ATP binding. In the middle part of the ATPase a CPC motif (Cys, Pro, Cys) is located as part of a hydrophobic amino acid cluster postulated to be part of the ion-translocating channel in related pumps(12) . In the N-terminal region there are two cysteine residues (Cys and Cys^14), as part of a CXXC motif, also present in other bacterial P type ATPases. In the cloned H. pylori pump the CXXC motif is preceded by two histidines (His^5 and His^7).

Homology with Other Bacterial P Type ATPases

The predicted protein sequence was shown to be most closely related to the Cd-exporting P type ATPase of Staphylococcus aureus(21) , CadA (31% overall sequence identity), and the Cu-transporting pumps CopA of Enterococcus hirae(19) (28%) and hpCopA, recently cloned from H. pylori(20) (27%). The homology was about 50% for all three pumps. In the case of enterococcal CopA (19) the homology was least in the N-terminal region containing a CXXC motif. The predicted H. pylori CopA ATPase lacks an N-terminal region with a CXXC motif, while a CSHC sequence is present in the operon-associated CopP gene product(20) , displaying some sequence identity when compared with the N-terminal region of the novel P type ATPase. When compared with the other copper-transporting pump of E. hirae, CopB(18, 19) , sequence similarity was even more reduced in the N-terminal 200 amino acid residues of this pump, while sequence identity with the C-terminal part was found to be 28% as observed for the enterococcal CopA protein. The sequence identity with the Rhizobium meliloti FixI P type ATPase, a pump predicted to be specifically required for symbiotic nitrogen fixation(44) , was significant (24%) but restricted to an internal stretch of 400 amino acids. In addition, there is significant sequence identity to two recently cloned histidine-rich P type ATPases, HRA-1 and HRA-2, supposedly derived from E. coli(45) . These two P type pumps, which are similar to bacterial copper-transporting ATPases do not contain a CXXC motif in their N-terminal region, as observed in CopA, hpCopP, CadA, and the H. pylori pump analyzed here, but contain N-terminal domains rich in histidine, methionine, glutamate, and aspartate residues. Significant sequence identity with the N-terminal region of HRA-1 and HRA-2 was not found. These homology data are summarized in Fig. 2.

Figure 2: Comparison of P type ATPases: regions of significant sequence identity of prominent members of the bacterial P type ATPase family. P type ATPases analyzed were the novel H. pylori P type ATPase (hpATPase), S. aureus CadA ATPase, the hpCopA/P ATPase of H. pylori (hpCopA, hpCopP), CopA and CopB of E. hirae (ehCopA, ehCopB), FixI ATPase of R. meliloti (rmFixI), and the histidine-rich P type ATPase Hra1 of E. coli (ecHra1 ATPase). All amino acid sequences were originally compared for homologies using the Genetics Computer Group programs PILEUP and PRETTY. P Type ATPases were aligned manually to show similar domains along the proteins. This revealed a block of very strong homology around the phosphorylation site (P, dark vertical bar) and the ATP binding region (ATP, dark vertical bar). This region was used as an anchor region for the manual alignment of all sequences. Indicated in the sequences are positions of histidine residues (light vertical bars) and cysteine residues (circles). The eight putative transmembrane helical regions of the novel H. pylori ATPase are shown as open boxes as determined using the in vitro translation mapping technique as described in this paper.


Significant homology was seen to portions of other P type ATPases, such as the E. coli KdpB gene product (46) and in particular to the human copper-transporting pumps associated with Menkes disease (26, 27, 28) , and the gastric H,K-ATPase (47) (data not shown).

Genomic Southern Blot Analysis

Southern blot analysis using the 2.1-kilobase pair EcoRI-HindIII DNA restriction fragment as a probe revealed that the corresponding DNA fragments are present in all H. pylori strains examined. Chromosomal DNA fractions prepared from clinical H. pylori isolates were also shown to be positive using the ATPase probe (data not shown). The gene seems to be widely distributed among H. pylori strains. Using H. felis chromosomal DNA a weak hybridization signal was obtained using less stringent conditions of hybridization (data not shown). The blot was negative with the DNA of C. jejuni, P. vulgaris, and E. coli (Fig. 3). As described for other genes of H. pylori, the hybridization pattern was variable with respect to the length of DNA fragments reacting, indicating restriction length polymorphism of the P type ATPase gene. These data suggest that the cloned P type ATPase gene is relatively selective for Helicobacter and is well expressed in H. pylori as compared with H. felis and not in the other bacteria examined.

Figure 3: Fluorogram of Southern blot analysis using DNA isolated from various H. pylori strains and related and unrelated bacteria. Genomic DNAs were prepared from H. pylori ATCC 49503 (lane 5), ATCC 43526 (lane 6), ATCC 43629 (lane 7) ATCC 51111 (lane 8), ATCC 51110 (lane 9), ATCC 43504 (lane 10) and clinical isolate H. pylori 69A (lane 11). Other DNAs were isolated from H. felis ATCC 49179 (lane 1), P. vulgaris ATCC 13315 (lane 2), E. coli MM294 ATCC 33625 (lane 3), and C. jejuni ATCC 33560 (lane 4). DNAs were digested with HindIII and separated by agarose gel electrophoresis. Lanes containing the different DNAs are given in parentheses. For detection of the cloned P type ATPase gene, a digoxigenin-labeled 2.1-kilobase pair EcoRI-HindIII restriction fragment of pRH439 was used for hybridization. Under stringent hybridization conditions used here, pRH439-derived P type ATPase sequences were detectable in all H. pylori strains analyzed but not in genomic DNA of H. felis, P. vulgaris, C. jejuni, and E. coli. Using less stringent hybridization conditions, a DNA fragment was detected in H. felis DNA (data not shown).


Analysis of Membrane Topology

Hydropathy

Various hydropathy-based algorithms predict several transmembrane segments in the H. pylori P type ATPase but are not in agreement for all such segments. These sequence predictions, as shown in Table 2, provided templates for the PY fusion vectors used for analysis of putative transmembrane segments. In Fig. 4the selected segments are shown along the amino acid sequence in combination with the Kyte/Doolittle hydropathy plot. These segments were analyzed for their signal anchor and stop transfer properties to enable experimental definition of the membrane domain of this particular P type ATPase.

Figure 4: Kyte/Doolittle hydropathy profile of the H. pylori P type ATPase and the primary amino acid sequence. The predicted Kyte/Doolittle hydropathy plot of the P type ATPase using a moving average of 11 amino acids is shown. The numbers at the top indicate the eight segments of the pump that were shown to act as transmembrane segments by using in vitro transcription and translation of insertion detection vectors (this study). The corresponding peaks of the plot are shaded. Also marked is the site of phosphorylation (P) at position Asp. The primary amino acid sequence consists of 686 residues, as predicted from the isolated DNA clone pRH439. The putative membrane-spanning segments are shaded differentially for sequences that behaved as both signal anchor and stop transfer sequences and for those that behaved only as stop transfer sequences.


In Fig. 5A, the M0 and M1 fusion protein products are shown as they were obtained by in vitro translation of the M0 and M1 vectors without inserts from the H. pylori P type ATPase. The M1 fusion protein was glycosylated in the presence of membranes, while the M0 protein was not, as shown previously(33) .

Figure 5: In vitro transcription/translation of M0 and M1 vectors without and with insertions corresponding to H1 (amino acids 73-92), H2(98-125), and H1+H2(73-125). S-Methionine-labeled protein products were visualized by autoradiography. Protein was obtained by in vitro transcription/translation of DNA vectors in the presence of S-methionine in the absence(-) or presence of microsomal membranes. The protein products were separated by SDS-PAGE followed by autoradiography. Panel A shows the result when the basic M0 vector and M1 vector, without carrying any DNA insertions, were subjected to in vitro translation as control. These M0/M1 control reactions were carried out in every experiment. Whereas the M(r) of the M0 alpha-beta fusion protein product is unaffected by the presence of membranes, it is shown that the presence of the first signal anchor transmembrane segment of the gastric H,K-ATPase resulted in a significant shift in molecular weight. Panel B shows the autoradiogram obtained by translation of the putative TM1 region (amino acids Leu-Leu) of the H. pylori P type ATPase inserted into M0 (PY-36) and M1 vector (PY-89). C, the result obtained by translation of H2 (amino acids Pro-Arg) in M0 (PY-37) and M1 vector (PY-51) is shown. D, translation of H1+H2 (amino acids Leu-Arg) in M0 (PY-63) and M1 vector (PY-102). Panel E shows the result obtained when the modified H1+H2 sequence element, Leu-Val, was used in M0 (PY-40).


In the following we describe a series of experiments using various arrangements of putative transmembrane sequences as defined by hydrophobicity, singly, in pairs, or in larger combinations, to determine their ability to membrane-insert during translation.

H1 (Amino Acids 73-92)

This sequence, the first putative transmembrane segment of the cloned H. pylori P type ATPase, was cloned into the M0 (PY-36) and M1 vectors (PY-89). Translation of PY-36 resulted in a fusion protein product of about 35 kDa as detected by SDS-PAGE and autoradiography. When PY-36 was translated in the presence of microsomes, a significant increase in the molecular weight was observed, indicating that the beta subunit part of the fusion protein was translocated across the microsomal membrane and glycosylated due to the signal anchor properties of H1 (Fig. 5B). Translation of the corresponding M1 vector with the Leu-Leu insert, PY-89, led to nonglycosylated protein products in the presence of microsomal membranes (Fig. 5B). The Leu-Leu insert can therefore behave as a stop transfer sequence with the first membrane spanning sequence of the gastric H,K-ATPase.

H2 (Amino Acids 98-125)

The next sequence analyzed was the second possible transmembrane segment (Pro-Arg) of the H. pylori protein inserted into the two vectors, M0 (PY-37) and M1 (PY-51). In the presence of membranes, the translation product of PY-37 was glycosylated. The product of PY-51, the sequence inserted into the M1 vector, was not glycosylated. Hence the region corresponding to the second predicted transmembrane segment of this H. pylori ATPase has both signal anchor and stop transfer properties as found for H1 (Fig. 5C).

H1+H2 (Amino Acids 73-125)

The properties of Leu-Leu as a signal anchor sequence and Pro-Arg as a stop transfer sequence, as demonstrated above, suggest that these sequences represent the first membrane-spanning pair of alpha-helices of the H. pylori P type ATPase. Accordingly, when the M0 vector, Leu-Arg (PY-63), containing the sequence encoding the first pair of putative transmembrane-spanning segments, was translated in the presence of membranes, glycosylation was not observed (Fig. 5D). The putative H2 region (Pro-Arg) is therefore able to act as a stop transfer sequence not only for the TM1 membrane-spanning sequence of the gastric H,K-ATPase (PY-51) but also for the putative TM1 element (Leu-Leu) of the H. pylori ATPase. Removal of the C-terminal arginine in this vector, and substitution of valine for phenylalanine (to be compatible with HindIII site cloning) as was done in the M0 vector PY-40 (Leu-Val; Val is substituted for Phe) resulted in absence of glycosylation as well. Thus, the stop transfer property of the putative M2 membrane-spanning sequence does not depend on the positively charged arginine residue localized at position 125 (Fig. 5E). The first pair of hydrophobic sequences on the deduced sequence of this P type ATPase appears to be a pair of TM segments.

H3 (Amino Acids 128-148)

When the DNA encoding for amino acids Phe-Cys was inserted into the M0 vector (PY-94) a significant fraction of the fusion protein was glycosylated, showing that this sequence was able to act as a signal anchor sequence. The presence of this sequence in M1 vector (PY-95) prevented most of the fusion protein product from being glycosylated. These data, as shown in Fig. 6A, demonstrate that the sequence between Phe and Cys is able to have both signal anchor and stop transfer properties. The N-terminal end of the insertion sequence was extended back to Arg, and the C-terminal end was extended to Glu, as shown in Fig. 6B, its efficacy as a signal anchor sequence improved, as did its efficacy as a stop transfer sequence. This suggests that the signal for the membrane insertion of this region begins at or near Arg.

Figure 6: In vitro transcription/translation of the H3 (amino acids 128-148), H4(149-176), and H3+H4(128-177) segments of the H. pylori pump using the M0/M1 vector system. The autoradiograms shown were obtained by in vitro transcription/translation of vectors in the absence (-) and presence (+) of membranes followed by SDS-PAGE to separate the S-methionine-labeled translation products. Panel A shows the translation products when the H3 segment of the H. pylori P type ATPase, Phe-Cys, was present in the M0 (PY-94) and the M1 vector (PY-95), respectively. Panel B shows the autoradiogram obtained with the extended H3 segment consisting of amino acids from position Arg to Glu. Using PY-90, the sequence was expressed as insertion of the M0 protein, whereas in PY-91 it was expressed as internal part of the M1 protein. C, in these lanes the products from translation of the H4 sequence are shown, starting with a valine residue in position 149 and ending with the serine residue of position 176. Translation of PY-83 resulted in the expression of the Val-Ser containing M0 fusion protein. Using PY-84, the same insert was expressed adjacent to the first transmembrane segment of the gastric ATPase cloned in the M1 vector. D, translation products obtained using the C-terminal truncated Glu-Lys region when expressed in M0 (PY-92) and M1 vector (PY-93). Panel E shows the products that were expressed with the vector PY-100 containing the putative second pair of transmembrane sequences, Arg-Lys, cloned into M0. Products containing the same sequence inserted into the M1 protein are obtained by using vector PY-101.


H4 (Amino Acids 149-176)

When the M0 vector containing the putative fourth TM segment (PY-83) was translated in the presence of microsomes, no glycosylation was found, showing that this stretch of amino acids in the PY-83 vector did not act as a signal anchor sequence. However, when present in the M1 vector (PY-84) this sequence was able to partially inhibit glycosylation, as shown by the appearance of a nonglycosylated product. These data show that this segment is able to act as a stop transfer sequence (Fig. 6C). If this sequence was truncated at the N-terminal end, starting therefore at Glu, as in the vectors PY-92 and PY-93, neither signal anchor nor stop transfer properties were displayed. These data imply that the four amino acids prior to Glu are important for the stop transfer properties of the fourth transmembrane segment (Fig. 6D).

H3+H4 (Amino Acids 128-177)

When the sequence spanning from Arg to Lys was inserted into the M0 vector (PY-100), only slight glycosylation was observed, consistent with the presence of two membrane spanning segments in the sequence. When present in the M1 vector (PY-101), some glycosylation was still observed, perhaps due to the odd number of membrane spanning segments in this vector (Fig. 6E). The third and fourth hydrophobic regions behave as a pair of TM segments.

H5 (Amino Acids 309-327)

The presence of this sequence in the M0 vector (PY-43) resulted in a glycosylation of the Ser-Gly fusion protein. Additionally, it completely prevented glycosylation when expressed in the M1 vector (PY-54) (Fig. 7A). Therefore, the region between Ser and Gly acts as both an efficient signal anchor sequence and as a stop transfer sequence.

Figure 7: In vitro transcription/translation of vectors with insertions corresponding to the H5 (amino acids 309-327), H6(337-371), and H5+H6(309-371) sequences of the H. pylori P type ATPase. In vitro reactions were performed in the absence(-) and presence (+) of microsomal membranes to assay for insertions with membrane-spanning properties. Panel A shows the autoradiogram obtained from the S-methionine-labeled products expressed from vectors carrying the H5 insertion, Ser-Gly, in the M0 (PY-43) and M1 vectors (PY-54). B, in the lanes shown, the products using the M0 (PY-77) and the M1 vector (PY-78), which contained the Leu-Lys H6 region, are depicted. Panel C shows the products obtained when the shorter H6 sequence ending at Lys was translated as part of the M0 (PY-45) and of the M1 alpha-beta fusion protein (PY-56). D, the autoradiogram is shown found with the Leu-Ser sequence inserted into M0 (PY-44) and M1 (PY-55). Panel E shows the results obtained by in vitro transcription/translation when the sequence from Ser-Lys corresponding to the third pair of transmembrane sequences, H5+H6, was expressed in both the M0 vector (PY-79) and the M1 vector (PY-80).


H6 (Amino Acids 337-371)

When the sequence representing Leu to Lys was inserted into the M0 vector (PY-77), no glycosylation was observed. In the M1 vector (PY-78) most of glycosylation was prevented, showing that this segment was able to act as a stop transfer sequence (Fig. 7B). Truncating the C-terminal end progressively to Lys and Ser impaired this stop transfer ability, showing that the last five amino acids of the Leu-Lys H6 sequence were important for stopping translocation of the protein across the membrane (Fig. 7, C and D).

H5+H6 (Amino Acids 309-371)

Vectors containing the sequence Leu-Lys verify that this particular region of the H. pylori P type ATPase is able to act as an additional pair of membrane-spanning segments. When the M0 vector carrying the putative fifth and sixth transmembrane segments (PY-79) was translated in the presence of microsomes, no glycosylation was observed, as expected of a pair of transmembrane segments. There was residual glycosylation of the M1 vector containing this region of the ATPase (PY-80), showing perhaps that this sequence could still insert as a pair of membrane segments following the TM1 of the gastric H,K-ATPase (Fig. 7E). The fifth and sixth hydrophobic regions of this ATPase are able to insert as a pair of TM segments.

H7 (Amino Acids 637-658)

Insertion of this C-terminal sequence of the H. pylori P type ATPase into the M0 vector showed that the Ile-Gly segment acted as a signal anchor sequence when translated in the presence of microsomes (PY-49). The majority of the Ile-Gly alpha-beta fusion protein product was glycosylated. Translation of the same sequence element in the M1 vector construct (PY-60) prevented glycosylation, demonstrating that this segment functioned as an efficient stop transfer sequence for the gastric TM1 transmembrane segment (Fig. 8A).

Figure 8: In vitro transcription/translation of vectors containing the H7 (amino acids 637-658), H8(659-685), and H7+H8(637-685) sequences of the H. pylori P type ATPase. Transcription/translation was carried out in the absence (-) and presence (+) of microsomal membranes. A, products shown here were expressed from the M0 (PY-49) and the M1 vector (PY-60) carrying insertions that represent the seventh transmembrane segment (Ile-Gly). B, also shown is the result obtained by insertion of the last stop transfer sequence, Val-Arg, into both M0 (PY-50) and M1 vector (PY-61). C, the right two lanes show the autoradiograms that were obtained by using in vitro transcription/translation of vectors including H7 through H8 (Ile-Arg) inserted in M0 (PY-70) and M1 (PY-76).


H8 (Amino Acids 659-685)

The sequence representing the adjacent putative membrane-spanning segment, from position 659 to 685, when inserted into the M0 vector, did not act as a signal anchor sequence (PY-50, Fig. 8B). However, as also shown in Fig. 8B, the Val-Arg sequence was effective as a stop transfer sequence when inserted into the M1 vector (PY-61).

H7+H8 (Amino Acids 637-685)

When the M0 vector contained a sequence comprising nearly the complete C-terminal region of the P type ATPase, Ile-Arg, only a small fraction of the product was glycosylated (PY-70, Fig. 8C). Hence the Ile-Arg segment can act as a putative fourth pair of transmembrane segments. However, expression of the Ile-Arg segment in the M1 vector did not result in glycosylation of the M1 fusion protein, showing that the C-terminal stop transfer region is unable to act as a signal anchor sequence in this construct (PY-76), as shown in Fig. 8C.

Longer C-terminal Constructs (Amino Acids 578-658 and 578-685)

The sequence between H6 and H7 contains the phosphorylation site and presumably much of the ATP binding domain. It was of interest to determine whether extending translation further on the N-terminal side of the H7- or the H7+H8-containing vectors would affect efficiency of the membrane insertion properties of the C-terminal region of the ATPase. As shown in Fig. 9, translation of the extended M0 vector PY-68 (Ala-Gly) resulted in glycosylation as predicted from translation of the vector containing a shorter sequence. However, it appears that glycosylation was more efficient than with the sequence beginning at Ile and ending at Gly (PY-49). Perhaps the preceding sequence in this region is able to improve membrane insertion of the last signal anchor sequence found in this ATPase. Insertion into the M1 vector (PY-74) also prevented glycosylation, hence the stop transfer properties previously observed for the H7 segment were retained. In the M0 vector containing the last two putative transmembrane segments with the N-terminal beginning at Ala, namely PY-69, no glycosylation was found as expected of a pair of inserted segments. Insertion of these sequences in the M1 vector, PY-75, also did not result in glycosylation, consistent with the finding that the last transmembrane segment could act only as a stop transfer sequence (Fig. 9, A and B).

Figure 9: In vitro transcription/translation of M0/M1 vector constructs containing longer C-terminal sequences (amino acids 578-658 and 578-685). Panel A shows the autoradiogram obtained when the M0 and M1 vectors with the Ala-Gly region insertion sequence, including H7, were translated in the absence(-) and presence (+) of microsomal membranes (PY-68 and PY-74). B, products obtained by expression of the Ala-Arg sequence, which includes H7+H8, inserted into M0 vector PY-69 and into M1 vector PY-75.


Translation of Vectors Carrying Inserts Corresponding to Segments H2-H5, H2-H6, and H1-H6

The above data suggest that this P type ATPase has eight transmembrane segments, based on insertion of individual or paired segments into M0 and M1 vectors. We also translated more complex constructs to show that the different transmembrane segments inserted appropriately when combined which each other.

For example, when Pro-Gly (H2-H5, PY-73) was translated in the M1 vector, the product was glycosylated as would be predicted from the signal anchor activities of H5 (Fig. 10A). Extension of the insert to Lys (PY-72) and, therefore, addition of the next putative membrane-spanning segment (Lys-Lys), which had been shown to act as an efficient stop transfer sequence did indeed inhibit glycosylation, as predicted for this construct ending with the stop transfer sequence H6 (Fig. 10B). No glycosylation was found when the Leu-Lys vector (PY-67) containing the H1-H6 segment was translated in the M0 vector (Fig. 10C) indicating that the three N-terminal pairs of transmembrane segments co-insert well during translation. In combination with the data for translation/insertion of PY-69, these constructs provide data consistent with the individual insertion results discussed above.

Figure 10: In vitro transcription/translation of the M1 vectors containing H2-H5 and H2-H6 and the M0 vector with H1-H6. S-Methionine-labeled protein products were detected by SDS-PAGE followed by autoradiography. Protein was synthesized in the absence(-) and presence (+) of microsomal membranes. A, the left two lanes show the translation product of the sequence corresponding to H2-H5, Pro-Gly, expressed in the M1 vector (PY-73). B, the protein products shown in the middle two lanes were obtained when the region representing H2-H6, Pro-Lys, was expressed as part of the M1 alpha-beta fusion protein (PY-72). Panel C shows the protein product when the M0 vector contained the sequence representing H1-H6 of the H. pylori P type ATPase (PY-67).


Other Possible TM Segments (Amino Acids 493-513, 541-563, and 578-601)

There are other regions of the protein that have some hydrophobicity, and these were also analyzed to determine whether they have any capacity for membrane insertion. These particular regions were selected because the hydropathy plot for the Cd-ATPase shows a possible additional pair of membrane segments in these regions. As shown in Fig. 11, none of these regions showed any signal anchor properties. However, all three had very weak stop transfer activity. Since they are not preceded by signal anchor sequences, their weak stop transfer activity is not likely to have significance for the membrane domain of this P type pump. A hydrophobic sequence in cyclo-oxygenase that is not membrane-inserted can act as a stop transfer sequence in the M1 vector (48, 49) but not as a signal anchor sequence. It may be concluded that a stop transfer sequence on its own does not imply membrane insertion unless it is preceded by a signal anchor sequence.

Figure 11: In vitro transcription/translation of M0/M1 vectors carrying insertions of other possible TM segments (amino acids 493-513, 541-563, and 578-601). Reactions were performed in the absence(-) and presence (+) of membranes. A, the left two lanes show the results of translation of the M0 vector PY-46 and the M1 vector PY-57, both carrying the sequence starting at amino acid Thr and ending at position Ile. B, the autoradiogram shown was obtained by using the M0 vector PY-47 and the M1 vector PY-58, both containing the insertion corresponding to amino acids Ser-Lys. Panel C shows the protein products obtained when the sequence Ala-Gly was translated as part of the M0 (PY-48) and the M1 fusion protein (PY-59).


Table 1summarizes the sequences that were analyzed in the two vectors and the effects of the inserts on induction of glycosylation in M0 and prevention of glycosylation in M1 along with the figures in which the data are presented. This method, therefore, reveals the presence of eight regions of the protein with membrane insertion properties.

DISCUSSION

Using the conserved DKTGT(L/I)T signature sequence of phosphorylation of bacterial P type ATPases we have isolated a clone from an H. pylori gene library predicting a 75-kDa protein. In terms of homology, hydrophobicity, and conserved sequence elements, there are several features that show it to be a member of the family of cation-pumping P type ATPases. The predicted amino acid sequence, 686 residues in length, contains an aspartate residue at position 388 as part of the DKTGTLT phosphorylation signature sequence, which is homologous with the phosphorylation sequence of another P type ATPase, hpCopA, recently cloned from H. pylori(20) . A GDGINDAP sequence is found in the cytoplasmic loop at positions 582-589 of the novel protein. This sequence matches the GDGXNDXP sequence element of other P type ATPases postulated to be within the region of ATP binding(12) .

This protein of H. pylori shows a strong overall sequence homology to the bacterial efflux protein CadA (21) and the bacterial and eukaryotic copper-transporting ATPases(19, 20, 24, 25, 26, 27, 28) . Two Cu P type ATPases have been cloned from E. hirae, CopA and CopB. While the CopA gene product was postulated to be a copper-importing protein, the CopB protein is responsible for copper export. The main sequence difference between these two related pumps resides in their first 100 N-terminal amino acids(19) . Compared with these pumps, the N-terminal amino acid region of the H. pylori P type ATPase analyzed here exhibits more homology to the N terminus of the E. hirae CopA ion pump as compared with the N terminus of the CopB protein. This might suggest an uptake function for this new ATPase. The N-terminal region of the cloned H. pylori pump, which contains a CXXC ion-binding motif, as found in staphylococcal CadA and enterococcal CopA, is also related to a cation binding protein, hpCopP, operon-associated with the hpCopA P type ATPase gene of H. pylori(20) . This gene is predicted to have only six transmembrane segments, but it should be pointed out that the original sequence for the CopA pump of E. hirae was later found to be truncated at the N-terminal end, and the full-length cDNA contained two additional predicted transmembrane segments(19) .

The N-terminal His/Cys-containing sequence element, HIHNLDCPDC, found at positions 5-13 in this P type ATPase, contains the CXXC sequence found in the N-terminal region in members of the P type ATPase family, which is thought to be involved in cation binding(12) . Strikingly, the CPDC sequence of the cloned enzyme is preceded by a HIHNLD sequence revealing a His/Cys-rich motif. Such motifs are thought to be part of an ion-binding motif for divalent ions, in particular Zn and Ni(50) . The N-terminal CXXC sequence element is also present in both the Cd-ATPase of S. aureus(21) and the CopA ATPase of E. hirae(19) and is missing in the 611-amino acid CopA gene product of H. pylori, but, as discussed above, its absence in this cDNA might reflect a cloning artifact. As part of the bicistronic CopA/P operon the CopA message of H. pylori is co-expressed with the CopP gene product, the latter containing a CXXC element(20) . In contrast to the published sequence of H. pylori CopA/P ATPase, the H. pylori pump cloned in our laboratory is a single polypeptide chain as was found for the E. hirae ATPase, which has been purified and reconstituted(17) .

A further motif of the H. pylori enzyme, as found in other pumps of this class is the membrane-associated CPC sequence, as predicted from the hydrophobicity plot, located in positions 344-346. It was postulated that this sequence motif, a proline residue flanked by two cysteines, is conserved in heavy metal transporting pumps(12) . However, the Fix1 gene product, which represents a P type ATPase of unidentified function of the nitrogen-fixing genus R. meliloti, also contains an intramembranous CPC sequence (44) perhaps indicative, with respect to the nitrogen assimilation machinery of R. meliloti, of a more extended substrate specificity of these pumps.

Additionally, the cloned P type ATPase contains several histidine and cysteine residues in the region around the conserved sites of phosphorylation and ATP binding, which may reflect the presence of Cu or Ni binding sites in this cytoplasmic region(45, 50) . The presence of putative cation binding sites in this domain could reflect regulation by divalent cations or transport of the cation itself. However, expression of the protein in E. coli did not alter Ni uptake, whereas synthetic peptides containing the histidine/cysteine-rich region of the N-terminal domain did appear to bind Ni or Cu (data not shown).

All P type ATPases are polytopic membrane proteins. Whereas the mammalian and fungal P type ATPases have similar hydropathy profiles, these are distinct from the bacterial type. In the case of this bacterial P type ATPase, various algorithms predict some transmembrane sequences in common and some different transmembrane sequences based on the deduced amino acid sequence. As compiled in Table 2, we used the algorithms of Rao & Argos(35) , Klein et al.(37) , and Eisenberg et al.(38) for prediction of putative membrane-spanning sequences of the H. pylori P type ATPase to guide selection of sequences to be inserted in the fusion vectors. The membrane insertion properties of the predicted segments were assayed using inserts into the M0 and M1 vectors for in vitro transcription/translation. As shown in this study, in vitro translation of the various M0/M1 vectors carrying various possible transmembrane sequences of the H. pylori enzyme led to the identification of eight transmembrane sequences along the polypeptide chain. The even number of transmembrane segments and the location of the phosphorylation and ATP binding region places both the N- and C-terminal amino acids in the cytoplasmic domain.

The experimental data presented above show that the regions corresponding to H1 and H2 acted as signal anchor and stop transfer sequences, corroborating the predictions of the different algorithms for this region of the enzyme. The region corresponding to H3 acted as signal anchor sequence, whereas the region of H4 acted only as a stop transfer sequence, consistent with the predictions derived from the Rao/Argos algorithm. The vectors containing the H5 and H6 regions acted as signal anchor and stop transfer sequences as predicted by all the algorithms. The first half of the enzyme behaves as if it has four segments. The vectors containing the putative H7 and H8 regions beginning at Ile and ending at Arg again were characterized by signal anchor and stop transfer properties, respectively. The seventh and eighth sequences predicted by the Eisenberg algorithm (PTMS X and Y; Table 2) were not found to have significant membrane insertion properties. The agreement between the experimental results found here is best with the Rao/Argos algorithm predicting six transmembrane segments and a seventh longer sequence that splits into two domains, giving the segments discovered here.

The hydropathy profile of this P type ATPase is similar to the various Cu P type ATPases including the mammalian types and the Cd-ATPase of S. aureus. The phosphorylation site, as shown in Fig. 4, is predicted to be between the sixth and seventh transmembrane segments in these pumps. In contrast, the mammalian and fungal ATPases have their phosphorylation site between the fourth and fifth transmembrane segments. Alignment of these various pumps suggests that the H. pylori pump and other homologous bacterial ATPases have an additional pair of membrane sequences preceding the first pair of membrane segments that are present in the mammalian and yeast enzymes as well as in the Mg-ATPase of S. typhimurium(32) . The H. pylori pump and other similar bacterial P type ATPases, on the other hand, do not have sequences following the eighth transmembrane segment. Hence P type ATPases fall into two groups, one containing eight transmembrane segments acting as heavy metal cation transporters and the other containing 10 transmembrane segments, which transport monovalent cations as well as Ca and Mg. The last six transmembrane segments of the heavy metal transporters and the first six transmembrane segments of the monovalent cation transporters could therefore form a central core, flanked by one or two pairs of additional transmembrane segments depending on the nature of the pump.

Although bacterial membrane-inserted proteins are often inserted post-translationally and bacteria lack the endoplasmic reticular system that characterizes eukaryotic cells, the translation/transcription system that had been applied to the mammalian gastric H,K-ATPase and SERCA 2 Ca-ATPase (33, 49) was successful in finding eight transmembrane segments in a bacterial P type ATPase as shown in this study. All signal anchor sequences had stop transfer properties, even though the sequence when used for stop transfer is usually of opposite orientation as compared with the assembled protein. It is also of interest to note that when a transmembrane segment showed only stop transfer properties, this property was retained in longer constructs, and signal anchor properties did not appear. Therefore, this method appears to be generally applicable to defining membrane-spanning alpha-helices in ATPases. Varying the length of the insert also provided information on the number of amino acids required for efficient membrane insertion. For example, truncating H4 or H6 prevented their acting as stop transfer sequences.

The membrane sequences predicted for this P type ATPase vary considerably in their content of charged or hydrophilic residues. The TM2 segment is predicted to have a lysine, a glutamic acid, and an aspartic acid; TM3 a glutamic and aspartic acid; TM4 three glutamic acids and a lysine; TM7 a lysine; and TM8 an aspartic acid residue. The presence of membrane-embedded carboxylic acids is frequent in mammalian P type ATPase that have been analyzed, but lysines are not often present in the membrane domain although an arginine is predicted in the membrane domain of the SERCA Ca-ATPase(51) . There are enough carboxylic acids present in the predicted membrane domain of the H. pylori ATPase to form ion pairs with the lysines. The extracytoplasmic domain has several carboxylic acids, and there is a CPC sequence in TM6. Whereas the composition of the transmembrane segments does not allow any firm conclusion as to the nature of the ion(s) transported by this enzyme, it seems highly likely that the enzyme is a cation transporter. The number of histidines and the CPC motif suggest either a role for divalent cation in transport or in fact transport itself of a divalent cation.

FOOTNOTES

*
This work was supported in part by United States Veterans Administration (to S. M. I.) and National Institutes of Health Grants DK46917, DK40615, and DK41301. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore by hereby marked ``advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank(TM)/EMBL Data Bank with accession number(s) L46864[GenBank].

§
To whom correspondence may be addressed: c/o Byk Gulden Pharmaceuticals, P.O. Box 100310, D-78462 Konstanz, Germany. Fax: 49 7531 843360; melchers@byk.de.

**
To whom correspondence may be addressed: Bldg. 113, Room 326, Wadsworth VA Hospital, Los Angeles, CA 90073. Fax: 310 312 9478.

(^1)
The abbreviations used are: H1, H2, etc., putative transmembrane segment from hydropathy; PCR, polymerase chain reaction; SSC, standard saline citrate; PAGE, polyacrylamide gel electrophoresis; ORF, open reading frame; LB, Luria-Bertani; M0, M1, fusion protein vectors used to detect membrane insertion; PY, fusion protein vector containing sequences from H. pylori; TM, transmembrane; ORF, open reading frame.

ACKNOWLEDGEMENTS

We thank Denis Bayle and David Weeks for much helpful advice in the analysis of the topology of the ATPase.

REFERENCES

  1. Pedersen, P. L., and Carafoli, E. (1987) Trends Biochem. Sci. 12, 146-150 [CrossRef]
  2. Glynn, I. M., and Karlish, S. J. D. (1990) Annu. Rev. Biochem. 59, 171-205 [CrossRef][Medline] [Order article via Infotrieve]
  3. Serrano, R. (1988) Biochim. Biophys. Acta 947, 1-28 [Medline] [Order article via Infotrieve]
  4. Epstein, W. (1990) in Bacterial Energetics, Vol. 12 (Krulwich, A., ed) pp. 87-110, Academic Press, Inc., San Diego
  5. Warren, J. R., and Marshall, B. (1983) Lancet i, 1273-1275
  6. Lee, A., Fox, J., and Hazell, S. L. (1993) Infect. Immun. 61, 1601-1610 [Free Full Text]
  7. Hazell, S. L., and Lee, A. (1986) J. Infect. Diseases 153, 658-663 [Medline] [Order article via Infotrieve]
  8. Malfertheiner, P., and Bode, G. (1993) Eur. J. Gastroenterol. 5, Suppl. 1, S1-S8
  9. Graham, D. Y., and Go, M. F. (1993) Gastroenterology 105, 279-282 [Medline] [Order article via Infotrieve]
  10. Cover, T. L., and Blazer, M. J. (1995) ASM News 61, 21-26
  11. Dick, J. D. (1990) Annu. Rev. Microbiol. 44, 249-269 [CrossRef][Medline] [Order article via Infotrieve]
  12. Silver, S., Nucifora, G., Chu, L., and Misra, T. K. (1989) Trends Biochem. Sci. 14, 76-80 [CrossRef][Medline] [Order article via Infotrieve]
  13. Laimins, L. A., Rhoads, D. B., and Epstein, W. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 464-468 [Abstract/Free Full Text]
  14. Epstein, W. (1985) Curr. Top. Membr. Transp. 25, 153-175
  15. Hugentobler, G., Heid, I., and Solioz, M. (1983) J. Biol. Chem. 258, 7611 [Abstract/Free Full Text]
  16. Fürst, P., and Solioz, M. (1986) J. Biol. Chem. 261, 4302-4208 [Abstract/Free Full Text]
  17. Apell, H. J., and Solioz, M. (1990) Biochim. Biophys. Acta 1017, 221-228 [Medline] [Order article via Infotrieve]
  18. Solioz, M., Mathews, S., and Fürst, P. (1987) J. Biol. Chem. 262, 7358 [Abstract/Free Full Text]
  19. Odermatt, A., Suter, H., Krapf, R., and Solioz, M. (1993) J. Biol. Chem. 268, 12775-12779 [Abstract/Free Full Text]
  20. Ge, Z., Hiratsuka, K., and Taylor, D. E. (1995) Mol. Microbiol. 15, 97-106 [CrossRef][Medline] [Order article via Infotrieve]
  21. Nucifora, G., Chu, L., Misra, T. K., and Silver, S. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 3544-3548 [Abstract/Free Full Text]
  22. Bull, P. C., and Cox, D. W. (1994) Trends Genet. 10, 246-252 [CrossRef][Medline] [Order article via Infotrieve]
  23. Bull, P. C., Thomas, G. R., Rommens, J. M., Forbes, J. R., and Cox, D. W. (1993) Nature Genet. 5, 327-336 [CrossRef][Medline] [Order article via Infotrieve]
  24. Silver, S., Nucifora, G., and Phung, L. T. (1993) Mol. Microbiol. 10, 7-12 [CrossRef][Medline] [Order article via Infotrieve]
  25. Phung, L. T., Ajlani, G., and Haselkorn, R. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 9651-9654 [Abstract/Free Full Text]