Using wobble primers in a touchdown PCR on H. elongata genomic DNA we discovered a DNA sequence with a high similarity to the translation product of Vibrio parahaemolyticus nhaD, a novel type sodium/proton antiporter 12. Using standard chromosome walking techniques and inverted PCR we acquired a sequence approximately 2800 base pairs (bp) long (full sequence available at the EMBL nucleotide sequence database, AM167899). The sequence contains at least one open reading frame (ORF) preceeded by a Shine-Dalgarno sequence (SD) 18. The ORF was designated nhaD due to the high similarity of the 412 amino acid (AA) transcription product with NhaD type antiporters of several marine γ-Proteobacteria (Fig. 1). Two putative promoter regions were found upstream of the ORF (Fig. 2). One of them is similar to E. coli σ 70 type housekeeping promoters and the other one resembles a rpoH dependent heat shock promoter (Tab. 1).

Between the putative rpoH and σ70 dependent promoters a second SD was detected. An ORF using the same reading frame would yield a putative alternate NhaD product which has an N-terminus of additional 71 AAs and would yield a total of 483 AAs.

The putative 412 AA protein has a calculated Mr of 45.7 kDa and a pI of 8.52. In the sequence 48 residues are basic, 22 are acidic and another 42 are uncharged polar, yielding a total of 112 (27%) hydrophilic and 300 (73%) hydrophobic residues. This complies with the overall composition of integral membrane proteins. The most abundant residues are 44 Leu (13%), 41 Val (10%) and 38 Ala (9.2%). Least abundant are Cys (4/1%), His (5/1.2%) and Gln (6/1.5%). Hydropathy plots show 10 to 12 possible transmembrane regions (data not shown). The data are very similar to those from V. parahaemolyticus NhaD 12 and support the concept that our protein is of the NhaD type.

A BLAST search yielded seven bacterial sodium/proton antiporters of the NhaD type from Proteobacteria: Alkalimonas amylolytica (AY962404), Shewanella oneidensis (AE015538, putative), Idiomarina loihiensis (AE017340, putative), Photobacterium profundum (CR378677, putative), Vibrio parahaemolyticus (AB006008), Vibrio cholerae (AF331042) and Vibrio vulnificus (AE016812 – all accession numbers taken from NCBI database). Several additional Na⁺/H⁺-antiporters and transport proteins for other metals show significantly lower similarities.

Multiple sequence alignment with NhaD type antiporters revealed 297 identical residues (70% based on the 412 residues of the H. elongata protein) and 311 highly conserved residues (76%) (Fig. 2). All of those organisms are γ-proteobacteria, six of marine origin and one soda lake isolate, A. amylolytica.

We also gained part of a DNA sequence of haloalkaliphilic Lake Bogoria Isolate 25B1 19, supposedly a member of the Family Halomonadaceae, which shows 95% identity to H. elongata nhaD on DNA level and therefore has equally high homology in the putative AA sequence (data not shown).

We constructed an expression vector from pUC18 where the β-lactamase is replaced in-frame by nhaD and named it pUCHelNhaD (Fig. 3). We decided to use the ORF yielding a 412AA product for two reasons: First dual control (see below) indicates that two promoters might be involved in nhaD expression, one being a stress promoter. The putative σ70 housekeeping promoter (Fig. 2) would yield this (shorter) protein. Secondly, our product is analogous to the functional V. parahaemolyticus NhaD described by Nozaki and coworkers 12. This is especially true for the ATG start codon.

E. coli mutants EP432 (nhaA and nhaB deleted in frame) and KNabc (nhaA, nhaB and chaA destroyed by insertion of omega cassettes) were transformed with pUCHelNhaD (complemented strain) and pUC18 (control). NhaD expression was under control of the pUC18 lac promoter. We decided to use both strains to assess possible differences in complementation by NhaD due to method of deletion and influence of the sodium extrusion system ChaA.

First we tested growth on solid minimal medium and compared colony size of both mutants with E. coli XL1-Blue and H. elongata (Fig. 4). In the absence of NaCl (traces from media components available) XL1blue grew well with LiCl concentrations below 38 mM. Growth was also observed at up to 150 mM, but colonies are drastically smaller. Complemented EP432 and the control both grew well only without LiCl, and did not show growth at concentrations higher than 4.8 mM. In sharp contrast to this H. elongata did not grow at all with less than 150 mM LiCl and only traces of NaCl available.

E. coli wild type is known to grow on minimal medium containing up to to 3% NaCl (w/v) (500 mM) and, in the presence of solutes, to tolerate salinities up to 5% (w/v) (666 mM). On MM63 supplemented with 0.1% (w/v) yeast extract we observed growth for XL1-Blue but it was very poor at 3 and 4% NaCl (w/v) (500 and 666 mM). At 1% and 2% (w/v) (166 and 333 mM) the complemented EP432 strain and the control did grow, but colonies were smaller than those of XL1-Blue with the control colonies being even slightly smaller. At 3% (w/v) (500 mM) the complemented strain showed weak growth where the control did not.

When grown in liquid medium (Fig. 5 and 6, Tab. 2 and 3) enhanced growth was observed for complemented EP432 pUCHelNhaD and KNabc pUCHelNhaD compared to the controls carrying pUC18. We observed higher growth rates, shorter lag phases, higher OD maxima or a combination of those factors. Growth enhancement was more drastic in minimal medium (Fig. 5B). Interestingly, in complex medium (Fig. 5A) growth was comparably poor. We attribute this to the high content of Na⁺-co-transported substrates (see discussion). Even in the presence of only traces of NaCl, growth of the transport deficient EP432 was reduced compared to the complemented strain. The observed worse growth in KNabc compared to EP432 was to be expected since the latter strain still has the ChaA Na⁺-extrusion system.

LiCl tolerance was also tested in shaking cultures. According to the data on solid medium no growth was observed at more than 4.8 mM LiCl. Below this concentration no significant differences in growth were observed between EP432 pUCHelNhaD and EP432 pUC18 or KNabc pUCHelNhaD and KNabc pUC18; this is true for pH values ranging from 7 to 9 (data not shown). We also tried to complement the NhaA deficient E. coli strain NM81 but results were poorly reproducible (not shown).

H. elongata is able to tolerate Li⁺-concentrations of up to 3 M depending on pH (Tab. 4). Therefore, E. coli DH5α where the full set of Na⁺-extrusion systems including NhaA, NhaB and ChaA is present was transformed with pUCHelNhaD and pUC18 (control). Both strains were grown at 500 mM (3% (w/v)) NaCl and at 300 mM NaCl/200 mM LiCl to test for enhanced Li⁺ tolerance. As can be seen in Fig. 7 and Tab. 5, in the absence of LiCl there appeared to be only little difference in growth between the two strains and in the presence of LiCl growth of the control is inhibited. To our surprise expression of H. elongata NhaD did not increase LiCl tolerance in an E. coli having a functional set of Na⁺/H⁺ antiporters. In contrast to our expectations next to no growth was observed in E. coli DH5α pUCHelNhaD, thus clearly indicating that NhaD is able to import Li⁺-ions into the cells. The same observations were made at a pH of 8.5 (in contrast to pH 7.3 of MM63) which is close to the average marine pH (data not shown). As can be seen in Tab. 6 lithium sensitivity is higher for the NhaD producing DH5a throughout the whole spectrum of pH and Li⁺ concentrations used. Regrettably growth rates are not well reproducible since the cells tended to aggregate at low phosphate and higher Li⁺ concentrations.

Analysis of mRNA shows an enhanced level of expression for H. elongata nhaD under conditions of a hyperosmotic shock (Fig. 8). Growth rates are similar for cultures under hypo-, iso- and hypersaline shock conditions but levels of nhaD mRNA rise only when the organism is subjected to a higher salinity.

Sodium proton antiporters are ubiquitous not only in Bacteria but also in Archaea and Eucaria 10 with Clostridium fervidum being the only known exception so far 9. Therefore finding a Na⁺/H⁺-antiporter in H. elongata was not too surprising. Sequence comparison indicates strongly that the antiporter belongs to the NhaD family (Fig. 1). Highly conserved and proposedly crucial residues Asp(369) and Thr(370), which correspond to positions 288 and 289 in 20, and other conserved polar amino acids 21 are present in the H. elongata protein.

The interesting fact is that most antiporters of this family known so far are either found in marine bacteria belonging to the γ-subgroup of Proteobacteria. Haloalkaliphilic A. amylolytica and Lake Bogoria Isolate 25B1 might be regarded as exceptions, but we have to keep in mind that marine habitats have a pH of about 8.2 and are therefore also slightly alkaliphilic. Thus, one might speculate whether the NhaD-type antiporter is specific for marine bacteria and possibly represents a special mechanism to adapt to moderate salinities of marine habitats. Complementation and expression studies, which are discussed in detail later, do not support this. We do not observe an enhanced salt tolerance for E. coli wildtype (MM63 with 4% NaCl (w/v) (666 mM) and 37°C, data not shown) and we observed enhanced expression of NhaD only after a hyperosmotic shock to salinities well above those of sea water (Fig. 8). Nevertheless, this is no contradiction to NhaD being specific for marine bacteria. We have to take into consideration coastal areas which can dry out periodically, thus resulting in higher salinities than the usual 3% (w/v) (500 mM) for seawater. In such situations it could prove useful to express a special system to extrude Na⁺ from the cytoplasm to support adaptation to salinities above 3%.

Further physiological data support the theory that NhaD is typical for marine or haloalkaliphilic habitats but indicate a different function.

As stated above, we chose to express the protein starting with the ATG at position 803 rather than 589 (Fig. 2) for two reasons. Firstly expression under control of the putative housekeeping promoter would yield this product and secondly the amino acid sequence reported for V. parahaemolyticus 12 and Fig. 1), is an equivalent to the shorter protein and completely functional. A similar observation was made by Ostroumov and her group 20 for V. cholerae NhaD who also successfully expressed a shorter ORF but gained a functional protein.

We also preferred using the pUC18 lac promoter for expression control rather than the putative promoter sequences upstream of the ORF because the promoter seems to be only induced at a reasonable level when the organism undergoes a hyperosmotic shock (Fig. 8) and also because we did observe that not all H. elongata promoters do work in E. coli (unpublished).

Complementation with H. elongata NhaD did not restore the lithium and sodium tolerance to the level of a wild type E. coli (Fig. 4). This might be due to incompatibility of this system to the host or simply due to lower numbers of antiporters in comparison to the wild type. The fact is that under given conditions NhaD does only poorly replace the NhaA/NhaB system, but nevertheless, we observed significantly enhanced growth when the antiporter was expressed in NhaA and NhaB deficient strains EP432 (Fig. 5, Tab. 2) and KNabc (Fig. 6, Tab. 3). Especially KNabc does not show any growth at 3% salinity (w/v) (500 mM) in mineral medium MM63 but has a reasonable growth rate when complemented (Fig. 6, Tab. 3). Effects are similar for EP432 in MM63 (Fig. 5A). This demonstrates that Halomonas elongata NhaD can replace the E. coli sodium/proton antiporters at elevated salinities.

To our surprise growth was equally poor in complex medium (Fig. 5A, Tab. 2) for both the control and the complemented EP432 at no (0% (w/v)) and elevated (3%) salinity (500 mM). We attribute this to the high amount of Na⁺-co-transported substarates in LB medium. A less efficient or incompatible export system for Na⁺ might consume more energy, which could otherwise be used for growth, or might result in a high intracellular sodium concentration which damages the organism. Again, this shows that NhaD is only a poor replacement for the A/B system.

Quantification of mRNA indicates that we have a certain housekeeping level of nhaD expression at all salinities. This basic level is enhanced by a factor of four within a few minutes after a hyperosmotic shock (Fig. 8). Considering that we worked with a starting salinity above that of sea water and ended up doubling this salinity, as discussed above this indicates that NhaD is not needed in large amounts at marine salinities. The expression level seems to drop after one hour in a fresh culture medium, even below the starting level and not depending on shock conditions, but since we had to use 16S rRNA as a standard, this is easily explained with variations in rRNA levels at those later time points 22.

These findings comply with the two putative promoter structures upstream of the open reading frame (Tab. 1). One shows significant similarity to sigma70 dependent housekeeping promoters 23 which are also reported for H. elongata (24 and recent unpublished data). The other one is reasonably similar to rpoH dependent stress promoters. For V. parahaemolyticus also two putative promoters are given 12, but here only two sigma70-like structures were found. So far no further expression studies for NhaD type sodium/proton-antiporters are reported, this definitely needs deeper investigation.

Several publications reporting Na⁺ export activity (e.g. 2117) and our own observations notwithstanding, we suppose that NhaD is rather a sodium import system than an export system. This agrees with an idea of active Na⁺ import 16, though we disagree with the hypothesis that NhaD is a pathogenicity marker. We do not see concomitance of pathogenicity and NhaD, especially with regard to H. elongata and the haloalkaliphiles.

In accordance with others (e.g. 12) we find NhaD can partially compensate for the NhaA/NhaB system in a deficient organism, but as can be seen from the physiological data in this study, it is only a poor replacement. NhaD does also not confer enhanced salt tolerance on a non-halotolerant organism. In addition to this, by expressing NhaD in an E. coli which still has both antiporters A and B, Li⁺ sensitivity is increased (Fig. 7, Tab. 5, Tab. 6). One might argue that this is caused by Li⁺ leaking through NhaD, but leakage would also occur through NhaA, NhaB and other channels and porines. Thus, leakage through one additional antiporter would not explain the drastic effect observed.

This sensitivity under conditions of heterologous expression complies with observations from Vibrio NhaD deletion mutants 16, which are less susceptible to lithium under alkaline conditions. (This referring to unpublished results only, no supporting data are given and no further data are published yet.) Both findings indicate import of small monovalent cations via NhaD.

Active Na⁺ import would make sense on a regulatory level, but this is highly speculative especially when we take riboswich elements into consideration (Reviewed in 25). Here we touch a field of regulatory mechanismns on a level of interactions of nucleic acids with small metabolites which we only begin to understand in detail.

Organisms are shown in Tab. 7 with genotype and references included. If not stated otherwise Halomonas elongata DSM2581^T was cultivated at 37°C and pH 7.3 in complex medium K 26 and modified minimal medium MM63 27, using glycerol as carbon source. E. coli strains were grown in modified Luria-Bertani (LB) 28 and MM63, also at 37°C and pH 7.3. Salinity and LiCl content of media were modified as necessary. For experiments concerning the range and pH dependency of Li⁺-tolerance a Tris/HCl-buffered version of MM63 with low (25 mM) phosphate content was used to prevent lithium-phosphate precipitation at high Li⁺-concentrations.

To pUC18 and pUCHelNhaD bearing strains we added 50 μg/ml ampicillin and 100 mM IPTG to induce nhaD expression (or as control). Cultivation was done in 250 ml shaking flasks containing 50 ml medium at 180 rpm or on solid medium containing 1.5% (w/v) agar. Growth was tracked in flasks equipped for optical density (OD) measurements. OD was determined at 600 nm and corrected according to the method of Dalgaard et al. 29.

Part of the sequence was found using wobble primers (5'-GAC GTC GCC GTC GSC GTS ATC ATH-3' and 5'-GGC GAA GGC GAT GAT SAG GAA RTY-3', with H = A, C or T; R= A or G; S= G or C; Y= C or T) in a touchdown PCR on H. elongata genomic DNA. Sequence information was completed using standard chromosome walking techniques and inverse PCR. Sequencing reactions were carried out by GATC (Konstanz, Germany).

H. elongata nhaD was amplified by PCR using Pwo DNA polymerase (Roche) and primers HeNxpF1 (5'-GAAACTG

CAtATG

GACgtc

RNA was extracted from H. elongata culture samples with the E.Z.N.A.™ Bacterial RNA kit (Peqlab). cDNA was constructed according to the protocol of Bertioli and Burrows 30. nhaD cDNA was quantified by real-time PCR in a GeneAmp^® 5700 System (Applied Biosystems) measuring SYBR^®-Green fluorescence.

Data base research was done using the NCBI Blast interface 31. Multiple sequence alignment was carried out with MULTALIN 32 and CLUSTHALW 33. Promoter analysis was done using the SEARCHLAUNCHER interface 34. Hydropathy plots were also done via SEARCHLAUNCHER according to the method of Eisenberg et al. 35.