Halotolerant and halophilic prokaryotes living in high-salt environments have developed two principal mechanisms—the salt-in-cytoplasm mechanism and the organic-osmolyte mechanism—to avoid the loss of water from the cell and achieve a cytoplasm osmotic strength similar to that of the surrounding environments [19, 22]. During adaptation to salt stress, most of eubacteria and several archaebacteria establish an osmotic equilibrium with the external environment by accumulating osmoprotective organic compounds called compatible solutes [3]. Glycine betaine, one kind of compatible solute, is accumulated by these prokaryotes in proportion to the salinity of the medium, which implies that this compound plays a major role in osmoregulation [17]. Most of the heterotrophic bacteria cannot synthesize glycine betaine de novo, but can accumulate it from the growth medium by uptake through secondary transport systems and/or ATP-binding cassette (ABC) transport systems or synthesis from its precursor choline by oxidation [3].

Halobacillus trueperi DSM10404T (hereafter H. trueperi) is a spore-forming, Gram-positive moderately halophilic bacterium, which can grow in medium with salt concentrations up to 25% [13], and accumulates glycine betaine as its major compatible solute under this condition (data not shown). A secondary glycine betaine transport system, betH gene, in H. trueperi was characterized, suggesting that the transcription of betH might be regulated by a σB-like factor [14]. The osmoregulated ABC transport systems for quaternary compounds had been found in many Gram-positive and Gram-negative bacteria, including Escherichia coli (ProU) [4], Bacillus subtilis (OpuA) [10], and Listeria monocytogenes (Gbu) [11], among others. These transporters belong to the superfamily of ATP-dependent transport systems, consisting of three different protein subunits: the ATP-binding component, the integral membrane component, and the periplasmic binding component [10]. The former two components form a membrane complex, and in Gram-negative bacteria, the substrate-binding protein resides in the periplasm as a soluble protein, whereas it is tethered to the cytoplasmic membrane in Gram-positive bacteria [10]. They constitute the main system to protect these organisms against hyperosmotic stress. Interestingly, no ABC transport system for quaternary ammonium compounds in the moderately halophilic bacteria has been reported until now. To gain a deeper insight into compatible solute uptake in halophilic bacteria, which accumulate glycine betaine as their main osmolyte, the genes encoding an ABC glycine betaine transporter in H. trueperi were cloned and characterized in this study.

Materials and Methods

Bacterial Strains and Culture Conditions

Halobacillus trueperi DSM10404T from Deutsche Sammlung von Mikroorgamismen und Zellkulturen GmbH (DSMZ) was grown at 37°C in SWYE medium and E. coli DH5α was grown at 37°C in Luria–Bertani medium. The antibiotics used were 50 μg/mL ampicillin and 50 μg/mL kanamycine. pGEM-T easy vector (Promega, Madison, WI, USA) was used as a general vector for cloning and sequencing.

DNA Manipulation and Materials

The genomic DNA from H. trueperi was extracted as previously described by Spring et al. [20]. The general recombinant DNA manipulation was performed by standard procedures [18]. Restriction endonucleases, DNA Blunting Kit, TaKaRa RNA PCR kit (AMV, ver. 3.0) (TaKaRa), T4 DNA ligase, and Pyrobest DNA polymerase were purchased from TaKaRa Dalian Co. and New England Biolabs Inc. The synthesis of oligonucleotide primers and DNA sequencing were conducted at Sangon Biotech. Carnitine, glycine betaine, and ectoine were purchased from Sigma (St. Louis, MO). All other chemicals were of the highest purity commercially available.

Cloning and Sequencing

To isolate a fragment of the glycine betaine ABC transporter genes from H. trueperi, a polymerase chain reaction (PCR) strategy with degenerate primers was used. Two stretches of amino acids, GEIFVIMGL (N-terminus) and HDLDEALRL (C-terminus), based on the well-conserved regions of ATPase were chosen to synthesize the degenerate primers. A 539-bp amplicon with primers AA-up (5’-ggwgarathttygtbatwatgggdct-3’) and AA-down (5’-awvcg harbgcytcrtccaartcatg-3’) was obtained and sequenced. Similarity between the deduced amino acid sequence of this fragment and that present in databases showed that it was homologous to the opuAA gene of B. subtilis. To obtain the nucleotide sequence flanking the DNA fragment, inverse PCR (IPCR) was performed. The genomic DNA was completely digested by EcoRI, purified and self-ligated with T4 DNA ligase at 16°C for 48 h. Then the circular fragments were used as templates and two sets of primers [up-144 (5’-acttctggcgacggacttcc-3’) and down345 (5’-gatgcaacagcgtgttggtc-3’), up-111 (5’-gctgtccatcttag ccaaatcc-3’) and down370 (5’-cgagcccttgcgaacgatcc-3’)] derived from the nucleotide sequence of the PCR products were used for the first and nested IPCR, respectively.

Another degenerate primer AC-down (5’-accwgcmccwggytcrath ccwgtrat-3’), deduced from the well-conserved region (ITGIEPEAG) of glycine betaine-binding proteins for glycine betaine ABC transporter was designed. Primers AA-up and AC-down were used for PCR with genomic DNA of H. trueperi as a template. In the following IPCR experiment, AflII was chosen to digest the genome of H. trueperi completely, and IPCR was performed as described earlier to obtain the nucleotide sequence flanking the DNA fragment. Primers [R1-up (5’-tggacttacccctgctgatgctcg-3’) and R1-down (5’-cggtatggtgccaggggtgtt tgc-3’), R2-up (5’-gaagtggcagtcgtccgcaaatgg-3’) and R2-down (5’-tcggaactggtttcgttgcaggtc-3’)] were chosen for the first and nested IPCR, respectively. The 3.2-kb IPCR product was cloned and sequenced. To confirm that the assembled sequence was the same as on the chromosome of H. trueperi, PCR analysis with primers ABC-up (5’-ttcccgggttccgcctgttgcatttc ctct-3’, SmaI site is underlined) and ABC-down (5’-aacccgggaagtggcagtcgtccgcaaatg-3’, SmaI site is underlined) using genomic DNA from H. trueperi as template amplified a 5525-bp fragment, which was digested by SmaI and cloned into the SmaI site of the pACYC177 vector to generate pAY55 and sequenced.

Probe Labeling and Southern Blot Analysis

A 2.1-kb DNA fragment from H. trueperi was amplified by PCR with degenerate primers AA-up and AC-down, purified on agarose gels, and labeled with digoxigenin-dUTP using Dig high prime DNA labeling and the detection starter kit I (Roche, Mannheim, Germany). For Southern blot analysis, H. trueperi genomic DNA was completely digested with six different restriction enzymes (AflII, ApaI, BglII, EcoRV, MluI, and SphI), respectively. After electrophoresis, it was transferred onto a positive-charged membrane (Amersham) under vacuum and hybridized with the probe. The hybridization was carried out at 38°C overnight. Detection was carried out by color detection using an anti-digoxigenin-alkaline phosphatase conjugate and nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP) according to the manufacturer’s instructions.

Functional Complementation of the Mutant E. coli MKH13

To determine which part of the 5.5-kb DNA fragment from pAY55 is coding for the potential transporter mediating the uptake of glycine betaine in E. coli MKH13 [5], a number of deletion derivatives of pAY55 (Fig. 5) were constructed. Plasmid pAY55 was hydrolyzed by restriction enzymes KpnI, BclI, XcmI, BsrBI and ApaI, respectively, and relegated after the creation of blunt ends with the DNA Blunting Kit (TaKaRa). Plasmids pAY52 and pAY49 derived from pAY55 were constructed by SmaI digestion, followed by either ApaI (pAY52) or BsrBI (pAY49). Each of these plasmids was introduced into E. coli MKH13 by transformation. The transformants were selected on selective M9 minimal medium plates containing 100 μg/mL ampicillin. The ability of the qatA encoding transporter to promote growth of E. coli MKH13 at elevated salt concentrations was tested on M9 mineral salt medium containing 0.8 M NaCl and 5 mM glycine betaine.

Extraction of Intracellular Solutes and 13C NMR Analysis

Escherichia coli MKH13 transformed with pAY41 was grown at 37°C on the selective M9 minimal medium. Cells were harvested by centrifugation and extracted twice with 80% ethanol [16]. Freeze-dried extracts were dissolved in 0.5 mL D2O and analyzed with a Bruker Advance DPX 300MHz [14].

RNA Isolation and RT-PCR

Exponentially growing cells of H. trueperi were harvested by centrifugation (4°C, at 10,000g for 6 min), and the pellet was stored in liquid nitrogen. Total RNA was prepared with the Trizol reagent (Invitrogen) according to the manufacturer’s instruction, and RNA was dissolved in diethypyrocarbonate (DEPC)-treated H2O. Residual DNA was removed from the RNA preparation by DNase I (Promega, China) digestion, followed by a phenol-acid extraction (at 65°C). A TaKaRa RNA PCR kit was used for reverse transcription (RT)-PCR. Primers Rta1(5’-aatcgacactcgtccggttg-3’) and Rta2 (5’-ctcggcacatccgtacgaag-3’), Rtb1 (5’-ttgccaaccaggagttatggg-3’) and Rtb2 (5’-agcacgctgaagtccggtaa-3’), Rtc1 (5’-ggtgatgagaccgcttc tgaa-3’) and Rtc2 (5’-ccattccgtcagcttcacca-3’) were designed from the qatAA, qatAB, and qatAC sequences. The absence of contaminating genomic DNA was controlled by non-RT-PCR performed under the same conditions, except that the avian myeloblastosis virus reverse transcriptase was replaced with DEPC-treated H2O.

Computer Analysis

The sequences were assembled and analyzed using the DNAman version 3.0 software (Lynnon BioSoft). Predicted amino acid sequences homology searches were carried out at the National Center for Biotechnology Information by using the BLAST programs (http://www.ncbi.nlm.nih.gov/BLAST) [1]. Protein alignments were performed by using the Clustal W program (http://www.ch.embnet.org/software/ClustalW.html#). Potential transmembrane spanning fragments were identified by using the SOSUI program (http://sosui.proteome.bio.tuat.ac.jp/sosuiframe0.html) [8]. Free energies of the stem–loop structures were calculated using the Mfold program (http://www.bioinfo.rpi.edu/applications/mfold).

Nucleotide Sequence Accession Number

The nucleotide sequences of H. trueperi qatAA, qatAB, and qatAC and its flanking sequences have been deposited in GenBank under No. AY786324.

Results and Discussion

Cloning Genes Encoding Glycine Betaine ABC Transporter from H. trueperi 10404T

By using degenerated primers AA-up and AA-down, a 539-bp DNA fragment encoding for a putative ATPase showing amino acid similarity with OpuAA from B. subtilis was obtained. A 2144-bp amplicon flanking the fragment was obtained by IPCR. After cloning and sequencing, the amino acid sequence deduced from the open reading frame (ORF) present in the fragment showed amino acid homology with the N-terminal and C-terminal sequences of OpuAA and with the N-terminal sequence of OpuAB from B. subtilis. Using primers AA-up and AC-down generated a second, somewhat larger PCR fragment of 2196 bp. The deduced amino acid sequence of the 2196-bp fragment showed significant similarity to the C-terminal sequence of OpuAA, the whole OpuAB, and the N-terminal sequence of OpuAC from B. subtilis. To obtain the complete sequence of genes encoding the putative glycine betaine ABC transporter QatA, a Southern hybridization experiment using the 2196-bp fragment as probe and another IPCR experiment were performed. Southern analysis indicated that the 3′ end of the genes was located on a 5.3-kb AflII fragment (Fig. 1). In all of the samples digested with restriction enzymes, only one strong band was observed, indicating that the chromosome of H. trueperi harbors only one copy of the qat gene cluster. The still missing sequences of the qat gene cluster were amplified by IPCR. The resulting 3.2-kb IPCR product was analyzed and the deduced amino acid sequence showed significant similarities to the N-terminal sequence of OpuAB and the C-terminal sequence of OpuAC from B. subtilis. An overview of the cloning strategy is shown in Fig. 2.

Fig. 1.
figure 1

Southern hybridization of a DIG-labeled 2196-bp PCR fragment to AflII (1), ApaI (2), BglII (3), EcoRV (4), MluI (5), and SphI (6) digests of genomic DNA from H. trueperi genomic. M: λDNA/EcoRI+HindIII.

Full size image
Fig. 2.
figure 2

Schematic diagram of the 5.5-kb cloned genomic DNA fragment containing the qatA operon in H. trueperi. The locations of the ORFs and their transcription orientation are shown as arrows. The positions for the potential transcriptional terminator are indicated (TT). The potential σA-dependent promoter is indicated by an angled arrow. Primers used: 1, AA-up; 2, AA-down; 3, up-144; 4, down345; 5, up-111; 6, down370; 7, AC-down; 8, R1-up; 9, R1-down; 10, R2-up; 11, R2-down; 12, ABC-up; 13, ABC-down.

Full size image

Analysis of the 5.5-kb Fragment from H. trueperi 10404T

Inspection of the sequenced 5.5-kb DNA fragment revealed the presence of four complete ORFs and one incomplete ORF. ORF2, ORF3, and ORF4 are each preceded at an appropriate distance by a putative ribosome-binding site and are in the same orientation, whereas ORF1 and ORF5 are divergent from the others (Fig. 2). The deduced amino acid sequences ORF2, ORF3, and ORF4 were subjected to BLASTP searches and showed similarity to the amino acid sequence of two known glycine betaine ABC transporters: OpuA from B. subtilis [10] and ProU from E. coli [4]. Both are glycine betaine transport systems belonging to the quaternary amine uptake transporter (QAT) family (3.A.1.12) of the ABC superfamily [15]. Therefore, we named the newly discovered ORFs as qatAA, qatAB, and qatAC (quaternary amine transporter), respectively. The end of the qatAA gene overlaps the beginning of the qatAB gene by 17 bp, and the intergenic distance between qatAB and qatAC is 70 bp. Interestingly, proU from E. coli [4], gbu from L. monocytogenes [11], and opuA from B. subtilis [12] also show similar overlaps between their first and second ORFs. The intergenic region (247 bp) between qatAA and ORF1 is highly A·T rich, which is also found in gbsAB genes of B. subtilis [2]. Inspection of the DNA sequence upstream of the qatAA gene revealed the presence of a potential ribosome-binding site (AGGTGG) and the putative –35 and –10 sequences (TTGAAC[17nt]TTACAT), which showed considerable similarity to the B. subtilis σA-dependent consensus promoter [6]. In contrast, the promoters of opuA from B. subtilis, gbu from L. monocytogenes, and proU from E. coli do not contain consensus sequences recognized by any known sigma factor from Gram-positive and Gram-negative bacteria. Furthermore, an inverted repeat from 9 to 45 bp downstream of the qatAC stop codon TAA that potentially could form a stem–loop structure [ΔG(25°C) = −97.1 kJ/mol] was found, which might function as a rho-independent transcription termination signal, followed by a stretch of five thymidine nucleotides. All of these suggested that the three genes might genetically be closely arranged in an operon.

Divergently oriented with respect to the qatA operon is an incomplete ORF (ORF1), which is highly similar (39% identity) to the transcriptional regulator from Desulfotomaculum reducens MI-1 (GI:88945602). Another ORF (ORF5) was found downstream of qatAC, encoding a potential 135-residue protein, which showed 53% and 41% amino acid identities with the protein of unknown function from O. iheyensis (GI:22776833) and B. subtilis (GI:2633336), respectively. It was preceded by a potential ribosome-binding site and was followed by a possible stem–loop structure, which was likely to function as a transcriptional terminator (data no shown).

Features of the qatA Genes

The ORF qatAA is predicted to encode a highly hydrophilic protein with a predicted molecular mass of 46 kDa, consisting of 414 amino acid residues. Inspection of the deduced amino acid sequence of opuAA revealed 63% and 53% identities to OpuAA of B. subtilis and ProV of E. coli, respectively. Both proteins have been proposed to form the ATPase subunit of OpuA and ProU, respectively. Most of the ATPase of glycine betaine ABC transporters usually contain two conserved sequences: the Walker A motif and the Walker B motif. The Walker A motif is also known as the P-loop and has a consensus sequence of GxxGxGKST, in which x represents any amino acid. The Walker B motif has four aliphatic residues followed by two negatively charged residues, generally aspartate followed by glutamate [21]. The putative QatAA amino acid sequence contained the Walker A and Walker B motifs, the highly conserved sequences of the ATPase subunit of these transporters, providing further evidence that QatAA is the ATPase subunit (Fig. 3). A highly conserved glycine-glutamine-rich sequence L-S-G-G-Q-Q-Q, named the linker peptide, was located between the membrane-spanning helical domain and the nucleotide-binding pocket in ABC transporters [9].

Fig. 3.
figure 3

Conserved sequence motifs of H. trueperi QatAA are underlined and arrowed: WA (the Walker A motif), Q (Q-loop), C (signature or C motif), WB (the Walker B motif), D (D-loop), and switch (the switch region).

Full size image

The ORF qatAB encodes a 281-residue hydrophobic protein with a predicted molecular mass of 30.78 kDa. The deduced amino acid sequence is similar to OpuAB (48% identity) from B. subtilis and ProW (48% identity) from E. coli, which are the transmembrane protein components of their respective transporters. Analysis of the hydropathic characteristics of the amino acid sequence indicated that the protein contains six segments with sufficient hydrophobicity and length to span the membrane. The qatAB gene product was therefore proposed to form a transmembrane channel. The permeases of three bacterial betaine ABC transport systems that demonstrate the highest similarity to QatAB of H. trueperi are aligned in Fig. 4.

Fig. 4.
figure 4

Alignment of the predicted amino acid sequence of H. trueperi QatAB with those of glycine betaine ABC transporter permease subunits from Bacillus subtilis, O. iheyensis HTE831, and L. monocytogenes EGD-e. Identical amino acids in all aligned proteins are indicated by asterisks and similar amino acids are indicated by dots. The predicted transmembrane-spanning domains for each sequence are underlined. Gaps introduced to maximize alignment are indicated by dashes.

Full size image

The final ORF qatAC is predicted to encode a relatively hydrophilic protein with a calculated molecular mass of 33.7 kDa, consisting of 311 amino acid residues, which are presumably from the substrate-binding protein. It is noteworthy that a 19-amino-acid region at the N-terminus of QatAC contains a high proportion of hydrophobic residues. Given the fact that a signal peptide consensus sequence (Leu-Val-Ala-Ala-Gly-Cys), existing in about three-fourths of the bacterial lipoprotein signal sequences [7], was detected in QatAC from H. trueperi, it is likely that the signal peptide is removed from the preprotein by a type II SPase (prolipoprotein signal peptidase) with the aid of Sec machinery to form the mature lipoprotein. The cysteine in this lipobox is lipid-modified by a diacylglyceryl transferase and attaches to the extracellular membrane after the proteolytic cleavage. Further tests also need to prove such a hypothesis.

Functional Complementation of Mutant E. coli MKH13

To verify whether qatA from H. trueperi encodes a glycine betaine transport system, complemen- tation experiments were carried out with E. coli MKH13, which is defective for glycine betaine synthesis, lacking the compatible solute transport systems ProP and ProU, and cannot grow in high-osmolarity M9 media containing glycine betaine. At 0.8 M NaCl and 5 mM glycine betaine, growth was observed in transformants with pAY55, pAY52, pAY49, and pAY41, but the strains transformed with pACYC177, pAY50, pAY45, or pAY39 cannot survive under the same condition. In the absence of glycine betaine, none of the above strains grew. Subclone experiments indicated that a 4183-bp DNA fragment was sufficient to confer osmoprotection to E. coli MKH13 (pAY41) in the presence of glycine betaine (Fig. 5). These findings implied that only qatAA, qatAB, and qatAC were required to restore salt tolerance in E.coli MKH13.

Fig. 5.
figure 5

Genetic map of the 5525-bp DNA fragment from H. trueperi (pAY55) and its deletion derivatives. The extent of the DNA segments present in the various deletion derivatives of plasmid pAY55 are indicated by the lines. Numbers indicate the terminal positions (bp) of the subclones relative to pAY55. Osmoprotection by glycine betaine was assayed by monitoring the growth of E. coli mutant strain MKH13 with various pAY55-derived plasmids on high-osmolarity minimal plates. Growth of the strains was scored after 3 days of incubation at 37°C (+: growth; −: no growth).

Full size image

In addition, to test whether QatA accumulates other compatible solutes as substrates, growth experiments with E. coli MKH13 (pAY55) and MKH13 (pAY41) were carried out in the presence of 2 mM glycine betaine, ectoine, or carnitine, with strain E. coli MKH13 (pACYC177) as the negative control. The strains were spread out on the selective M9 medium containing 0.8 M NaCl with or without compatible solutes and incubated at 37°C. After 3 days, clones appeared only on plates with glycine betaine, whereas plates with ectoine or carnitine did not support growth as well as the negative control. 13C NMR (nuclear magnetic resonance) analysis (Fig. 6) showed that glycine betaine was accumulated in cytoplasm of E. coli MKH13 harboring the qat gene cluster on plasmid pAY41. These results proved that qatAA, qatAB, and qatAC together encode a functional uptake system for glycine betaine.

Fig. 6.
figure 6

13C NMR spectrum of E. coli MKH13 (pAY41) ethanolic cell extracts. Resonances from glycine betaine (B) and the standard dimethyl sulfoxide (*) were indicated.

Full size image

RT-PCR

We had intended to detect qatA transcripts by Northern blots, but it was unsuccessful. This might be due to their instability or the low-level transcription in vivo. We thus tried to detect these transcripts by reverse PCR. To test whether the qatA genes were expressed in vivo, total RNA from H. trueperi was subjected to RT-PCR. Using primers as described earlier, three of the expected size DNA fragments (bands of 840 bp, 496 bp, and 666 bp for the qatAA, qatAB, and qatAC transcripts, respectively) were amplified (Fig. 7), which proved that the qatA genes are transcribed in cells of H. trueperi.

Fig. 7.
figure 7

RT-PCR fragments amplified from total RNA of H. trueperi for the qatAA transcript with the primers Rta1 and Rta2 (lanes 1 and 2, respectively); the qatAB transcript with the primers Rtb1 and Rtb2 (lanes 3 and 4, respectively), and the qatAC transcript with the primers Rtc1 and Rtc2 (lanes 5 and 6, respectively). Negative control was performed without reverse transcriptase (lanes 2, 4, and 6). M: DNA molecular marker.

Full size image