Random mutagenesis in Corynebacterium glutamicum ATCC 13032 using an IS6100-based transposon vector identified the last unknown gene in the histidine biosynthesis pathway

Transposon mutagenesis of Corynebacterium glutamicum ATCC 13032 with an IS6100-based transposon vector

In vivo mutagenesis of the C. glutamicum ATCC 13032 genome was performed with the IS6100- based artificial transposon vector named pAT6100. The vector is a construct resulting from cloning IS6100 into the vector pK18mob2 [31]. The pAT6100 plasmid DNA extracted from Escherichia coli DH5αMCR was transformed into electrocompetent cells of C. glutamicum RES167, a restriction-deficient strain derived from the wild-type, by electroporation. Since the vector shares no homologous sequences with the host genome and it is not able to replicate autonomously in C. glutamicum the kanamycin resistance phenotype is the result of a transposition event into the chromosome. The transposition efficiencies ranged from five to ten c.f.u. per μg of desalted plasmid DNA.

An ordered clone library was constructed comprising 10,080 independent mutant clones. With a coverage more than three-fold with respect to the 3,002 coding regions determined for C. glutamicum ATCC 13032 [10] we expect that nearly all non-essential genes contain insertions. In order to estimate the theoretical quality of the library, statistical analyses were made: Given the number of clones in the transposon library (10,080), the estimated number of non-essential genes (2,400) [32], and the average gene length (952 bp), the average probability for any given gene to be disrupted by at least one transposon insertion is 97%.

In order to perform experimental analyses of the library, first a potential target site preference of the transposon was tested. Therefore, 26 clones were randomly selected and plasmid rescue cloning was carried out by digesting chromosomal DNA of the selected clones with EcoRI or XbaI, respectively, religating of the rescue constructs, transferring to E. coli and subsequent sequencing of the genomic insert with IS6100-specific primers. The sequence data obtained was analysed for sequence homologies with BLASTN searches against the C. glutamicum ATCC 13032 whole genome sequence [10]. The BLAST results revealed the common 14-bp inverted repeat of either left and right flank of the cointegrate followed by the individual 8-bp direct repeat target site for each clone, indicating the position of the transposon integration (Table 1). As the target sites of the clones were each different, with no duplicates with the same or similar sequences, it can be assumed that these 26 transposition events occurred independently. The position of the transposon in every mutant, as well those selected randomly (Fig. 1, red bars) and those investigated in auxanographic analyses (Fig. 1, black bars) were plotted on a circular map representing the C. glutamicum ATCC 13032 chromosome. The picture indicates a random distribution throughout the genome without regions of significant accumulation of insertions. Both, intragenic and intergenic insertions were detected. The G+C content was calculated from the eight nucleotides of the target site and ten flanking nucleotides upstream and downstream. It showed a high variance ranging from 39.3% to 78.5%, whereas the medium G+C content for the chromosome of the strain ATCC 13032 is 53.8%. The two possible orientations of the cointegrate with respect to the directions of replication or transcription are present in nearly equal proportions indicating that neither orientation is favoured.

Clone notation a	Integration position b	Locus tag c	Gene	Gene product	Target site duplication (TSD) d	G+C content [%] e	Transposon orientation f
51G06	141,831	cg0165		ABC-2 type transporter	GGCGCGCG	78.5	-
01A01	603,250	cg0683		Permease	AGTGAACC	53.5	+
52D11	737,930	cg0822		Conserved hypothetical protein	CCCAACGG	60.7	+
51B11	853,795	cg0924		ABC-type cobalamin/Fe3+-siderophores transport system periplasmic components	CTCAACGG	53.6	+
31D11	899,182	cg0963_0964			CTCTTTTT	39.3	+
51H01	1,097,334	cg1185	tnp10b	Transposase – fragment ISCg10a	AAAAATAC	57.1	+
51D02	1,099,353	cg1192		Aldo/keto reductase	CCCTAGCG	46.5	+
147A01	1,129,669	cg1228		ABC-type cobalt transport system, ATPase component	CCTACGTT	42.9	+
51B03	1,137,471	cg1237		Putative membrane protein	CCTTTGAG	42.9	+
51E04	1,161,219	cg1265		Conserved hypothetical protein	TAAGGAAG	39.2	-
51E02	1,217,257	cg1310	tfdF	Maleylacetate reductase	CGATTACG	50.0	-
51G12	1,246,055	cg1338	thrB	Homoserine Kinase	CCTATTAC	46.4	-
57D08	1,408,925	cg1516		Hypothetical protein	GCGATATC	46.4	-
121D05	1,613,084	cg1724		Putative protein kinase ArgK or related GTPase of G3E family	AGGTGAAG	60.7	+
43C10	1,717,632	cg1825	efp	Translation elongation factor P	CGGGTGTC	60.7	+
157H05	1,944,316	cg2053_2054		upstream putative membrane protein (cg2054)	CTTGATTC	42.9	+
29F07	2,180,143	cg2296	hisI	Probable phosphoribosyl-AMP cyclohydrolase	CGCCAAAG	57.1	-
17B06	2,234,105	cg2349		ATPase components of ABC transporters with duplicated ATPase domains	TTCTGGAA	64.3	-
75E09	2,418,388	cg2537	brnQ	Branched-chain amino acid uptake carrier	GTTTCATT	42.8	+
51C02	2,537,800	cg2661		Putative dithiol-disulfide isomerase involved in polyketide biosynthesis	ACTGGACT	53.6	+
51A01	2,772,437	cg2913		ABC-type Mn2+/Zn2+ transport system, permease component	TTGCTGAT	57.2	+
97B06	2,913,169	cg3049_3050		upstream acyltransferase (cg3050)	CTCGACTG	57.1	+
02A09	2,958,982	cg3100	dnaK	Heat shock protein Hsp70	TTCTCAGC	53.5	+
51A02	3,132,229	cg3271_3272			GTCAGCCG	60.7	-
35E03	3,166,513	cg3319_3320		upstream uncharacterized enzyme related to sulfurtransferases (cg3319)	CTTCAGTC	46.4	+
51F01	3,222,657	cg3373		Bacterial regulatory protein, ArsR family	CTTCCGCG	60.8	+

^a Indicates the position of the clone in the transposon library.

^b Absolute position of the cointegrate in the C. glutamicum ATCC 13032 genome [GenBank:BX927147].

^c Integration of a transposon within or between genes (separated by underscoring)

^d Sequence of the 8-bp direct repeat.

^e G+C content was calculated from 28 nucleotides around integration (TSD and 10 nt down-/upstream).

^f Orientation of the integrated transposon with respect to the genome in clockwise (-) and counterclockwise (+) direction

An additional test for target site preferences was carried out by performing sequence pattern searches with the TEIRESIAS algorithm [33]. For this purpose a total of 172 target sites acquired from the random selected clones and from mutants of the auxanographic analyses (see below) were used. No sequence pattern, palindromic sequences or regions of nucleotide symmetry could be detected with this bioinformatic approach (data not shown). Additionally, the occurrence of the bases at a distinct position of the 8-bp target sequence was calculated (data not shown). The values and therefore the appearance of bases appear to be distributed evenly. Neither the pattern searches nor the nucleotide distributions in the TSD indicate a significant target site preference for insertion.

The cointegration of the transposon vector forms long direct repeats in the form of IS6100 copies at the site of insertion. Such direct repeats might cause instability by replication slippage or by homologous recombination, thereby losing the integrated vector and one or both of the IS6100 elements in the absence of antibiotic selection. To assess this, five randomly selected individual clones were subcultured in antibiotic-free liquid complex medium in twelve rounds of 48 h each. The last culture was plated on solid complex medium. From this culture, 900 single colonies from each clone were tested on kanamycin-containing plates. No loss of the resistance phenotype and therefore the integrated transposon could be observed. It has to be concluded that the integrations are stably maintained in C. glutamicum even in the absence of selective pressure. Additionally, isolated DNA from randomly selected clones which had not been grown under selective growth conditions was used to perform PCR experiments. In all cases the presence of the cointegrate was verified (data not shown).

Auxanographic screening and targeted gene identification using PCR techniques

The initial step of performing genome-scale auxanographic analyses was the plating of the entire transposon library on minimal medium plates. After 32 h of incubation this screening approach delivered 342 mutants which exhibit a clear growth-deficient phenotype. The appearance of potential auxotrophic mutants which are unable to form colonies under such growth conditions was used as an initial criterion for efficient mutagenesis. These mutants were further characterized by the rapid method for identification of bacterial biochemical requirements according to Holliday [34]. Therefore, 12 plates of minimal medium were each supplemented with different combinations of six growth factors. Mutants with a single biochemical requirement would grow on two plates. The double auxotrophic mutants could be determined by growth on only a single plate and an alternative requirement by growth on more than two plates. For the latter two cases the exact nutritional requirements have to be determined by further tests combining individual growth factors.

With this method, a total of 36 different growth factors, which are adapted to corynebacterial needs by omitting biotin, choline, and thiosulfate, and adding cobalamine, glutamine, as well as asparagine, could be tested simultaneously. These growth factors comprise 21 amino acids, six nucleotides and nine vitamins, including some of their precursors. Each candidate auxotrophic clone was inoculated on all twelve plates and incubated for 32 h. Forty-seven out of the 342 clones tested formed slow growing colonies on every plate, suggesting that these are generally growth-deficient but not auxotrophic mutants. With the redefined number of 295 clones circa 2.9% of the transposon library are genuine auxotrophic mutants. This roughly correlates to prior studies with different transposons in which auxotrophic mutants were obtained with frequencies of 1.3% in C. glutamicum [24], 2.5% in Rhodococcus spp. [35] and 2% in Rhodococcus fascians [36].

Out of these 295 clones 167 display a requirement for a single growth factor (57%). Among these, 141 exhibit an auxotrophy for one of eleven different amino acids, 24 for one of four different nucleotides and two for different vitamins. Additionally, we have found twelve clones that show a double and six with an alternative requirement. Altogether, the auxotrophic requirements of nearly two thirds of the mutants could be identified.

In the majority of clones the observed auxotrophic phenotype is well explained by the existing annotation of the disrupted gene. For example amino acid biosynthesis genes like argG (arginine), thrB (threonine), proC (proline), serA (serine), leuD (leucine) or lysA (lysine) were subject to intensive studies in C. glutamicum before, and the transposon insertions in these genes resulted in the expected auxotrophic phenotypes. This was also the case for insertions into genes that show high similarities to genes with a known function in closely related species, e.g. guaA and guaB2, known to be involved in the guanosine monophosphate biosynthesis in Corynebacterium ammoniagenes [37], or to genes in more remote organisms, e.g. purA and purB which are involved in adenosine monophosphate biosynthesis in E. coli [38]. Another example are the carAB genes: even though these genes are not characterized in C. glutamicum they are well known to cause a double requirement for both arginine and uracil in E. coli since they encode for the subunits of a protein that provides the essential carbamoylphosphate for arginine and pyrimidine biosynthesis [39]. The ilvC gene is well studied in C. glutamicum and is known to cause a double auxotrophy for isoleucine and valine when disrupted [40]. The successful supplementation of a metE transposon mutant strain with either methionine or cobalamin (vitamin B12) is explained by the fact that a cobalamin-dependent (MetH) and a cobalamin-independent enzyme (MetE) perform the same step in methionine biosynthesis in C. glutamicum [41]. The loss of the MetE enzyme, the preferred route during aerobic growth, can be compensated by supporting the MetH function with appropriate amounts of cobalamin.

In some cases the correlation of observed phenotype and the gene targeted by transposon integration are not so obvious. An example is the guanine auxotrophy of the nadA insertion. The nadA gene is located together with nadB and nadC and encodes a protein which is apparently involved in quinolinate and nicotinamide adenine dinucleotide biosynthesis. It is currently unclear why guanine supplements growth of this mutant. Also in the case of the integration in gene thiO the connection between gene function and auxotrophy is not clear yet. The gene as part of the thiEOSGF cluster encodes a putative D-amino acid oxidase flavoprotein oxidoreductase that is known to be involved in thiamine biosynthesis [42]. Although a polar effect on the downstream genes thiSGF is not experimentally verified, such an effect would furthermore affect a more extensive part of the thiamine synthesis. The mutant is supplementable by histidine and not, as expected, by thiamine.

Since the de novo purine and pyrimidine nucleotide synthesis pathways are poorly characterized in C. glutamicum especially the functions of genes like purF or pyrB require further studies as well. Mutants with transposon insertions in the C. glutamicum genes purC, purE, and purK, known as purine biosynthesis genes in E. coli [38], were found among those which grew on more than three plates (data not shown). These mutants appear interesting to be investigated in further studies.

The genes of most biosynthetic pathways in C. glutamicum are scattered in different genetic loci. A special case where the distribution of transposon insertions in a limited genomic region could be tested is the trp operon containing six genes which encode all functions of tryptophan biosynthesis in C. glutamicum [43]. For this, the transposon insertion sites of 43 tryptophan-auxotrophic clones were determined. The insertion sites were all different and located within every gene from trpE to trpA (Fig. 2). Although the trp operon contains more insertions than statistically expected, this example gives additional evidence that the transposon library is random and that the transposon system can be used to mutagenise the C. glutamicum chromosome up to a high density.

To find a gene of interest that is mutated by a transposon is sometimes not possible by phenotypic screening. We were interested to map additional insertions upstream from the trp operon. For this, a PCR-based screening strategy was chosen and the workflow by Hobom et al. [44] was adapted so that the entire library could be screened for insertion events by iterative rounds of three-step PCR experiments simultaneously. This strategy delivered another transposon insertion which was mapped 35 bp upstream from the predicted trpP translation start. This mutant did not exhibit a tryptophan-auxotrophic phenotype in subsequent growth tests which coincides with the assumption that the trpP gene encodes a permease necessary for tryptophan- and 5-methyltryptophan uptake in C. glutamicum [45].

Identification of a novel C. glutamicum gene closing the last gap in histidine biosynthesis

The whole genome sequence of C. glutamicum ATCC 13032 is known and the genes are annotated based on sequence similarity and experimental data [10]. Due to the fact that a large number of amino acid biosynthesis genes have been characterized before, only few pathways remained incomplete. With the help of sequence similarities and subsequent genetic analyses further genes have been identified [46] leaving only few gaps. These "missing" genes mostly encode transaminases which are notorious for their involvement in more then one pathway and were also partly characterized by bioinformatic and biochemical studies [47, 48]. Beside this, only a single step in histidine biosynthesis remained, where no gene was known or a candidate gene could be identified by sequence similarity.

The metabolic pathway of the histidine biosynthesis, reactions, enzymes, as well as the organization of the corresponding genes display a high degree of conservation in bacteria [49]. According to the KEGG PATHWAY database [50] the histidinol-phosphate phosphatase (HolPase) is an exception which was not characterized at a genetic level in C. glutamicum and closely related Actinobacteria. This enzyme (L-histidinol-phosphate phosphohydrolase; EC 3.1.3.15) catalyses the penultimate step of the biosynthesis, the dephosphorylation of histidinol phosphate to histidinol, the direct precursor of histidine. In the reference organisms E. coli and Salmonella typhimurium the HolPase activity is associated with the N-terminal domain of the HisB bifunctional enzyme. The C-terminal domain exhibit similarity to an imidazoleglycerol-phosphate dehydratase (EC 4.2.1.19), which catalyses the sixth step of the histidine synthesis. However, amino acid sequence homology searches revealed that in C. glutamicum, like in most bacteria, the imidazoleglycerol-phosphate dehydratase is encoded as a monofunctional enzyme. Additional similarity searches for the E. coli homologs of the HolPase in C. glutamicum and other organisms, including the entire class of Actinobacteria, does neither provide any significant sequence homologies to ORFs in the C. glutamicum genome nor to well-conserved domains in other Actinobacteria.

Since the loss of the HolPase activity is supposed to exhibit a histidine-auxotrophic phenotype the corresponding unidentified gene in C. glutamicum was presumed to be among the 16 histidine-auxotrophic transposon mutants. Determination of the transposon integration sites in these mutants revealed insertions in seven of the nine histidine synthesis genes known so far, hisGE, hisHAFI and hisDCB, and another one in the gene cg0910 (Table 2). The integration in this gene, which comprises 783 bp, is located 125 bp downstream from the predicted cg0910 translation start, suggesting that the gene product is non-functional. The cg0910 gene is preceded by the gene cg0911 and both might be transcribed together (Fig. 3). The cg0911cg0910 locus is located far from known histidine biosynthesis genes. BLASTP searches with the deduced protein sequence of both genes revealed significant similarities to each other and to inositol-1(or4)-monophosphatases (IMP; EC 3.1.3.25) which hydrolyse the ester bond of inositol phosphate to generate inositol.

Validation of cg0910 gene function by deletion and homologous complementation

In order to confirm the phenotype of the transposon mutation, a cg0910 deletion mutant was constructed using the "gene splicing by overlap extension" (gene SOEing) technique [51]. Therefore, a fusion product of chromosomal DNA regions of 622 bp and 541 bp directly up- and downstream, respectively, of the cg0910 target locus was created, which was subsequently cloned into the pK18mobsacB vector system [5]. The resulting vector pK18mobsacB-Δcg0910 was finally used for targeted gene deletion. The deleted internal cg0910 fragment was 702 bp of size (Fig. 3).

The initially observed phenotype of the transposon mutant with the integration in cg0910 (clone 99G01) could be reproduced with the resulting deletion mutant, termed CG281, which was tested negative for its ability to grow on minimal medium without and positive with histidine supplementation. This indicated that possible secondary effects in the transposon mutant could be ruled out and that the loss of the cg0910 gene product is responsible for histidine auxotrophy. Both the mutants 99G01 and CG281 also grew on minimal medium supplemented with histidinol, the end product of the histidinol phosphatase reaction, comparable to the wild-type strain. These results suggest that cg0910, assigned with the gene name hisN, encodes the hitherto unknown but essential HolPase activity in C. glutamicum, and is the functional homolog of the E. coli N-terminal HisB protein domain (Table 2).

To provide further evidence that the cg0910 gene encodes the HisN function, homologous complementation was carried out in both mutants 99G01 and CG281. For this, the complete cg0910 coding sequence was amplified by PCR with primers providing an optimized ribosome binding site and additional restriction enzyme recognition sites. The amplificate was subsequently cloned into the IPTG-inducible shuttle expression vector pEC-XT99A [52]. The resulting plasmid pEC-XT99A-cg0910 ex (Fig. 3) was subsequently transferred to the corresponding mutant strains which were then tested for histidine prototrophy. In both cases the extrachromosomally located cg0910 gene product complemented the loss of the corresponding chromosomal region in trans (data not shown).

Phylogenetic analysis of Cg0910 and related inositol monophosphatases (IMP)

BLAST searches revealed four IMP paralogs in the C. glutamicum ATCC 13032 genome with significant similarity to cg0910, by name cg0911, cg0967 (cysQ), cg2090 (suhB), and cg2298 (impA). For a functional classification of the inositol monophosphatase family-like proteins, phylogenetic analysis in comparison to sequence-related proteins from other Actinobacteria and the model organism E. coli was conducted. For this, similarity searches with the deduced amino acid sequences of the C. glutamicum IMPs in non-redundant databases were carried out, comprising, beside E. coli K-12, the genomes of Actinobacteria including completed as well as draft status genomes. Proteins with reliable similarity scores were used to perform a multiple alignment. Based on this alignment a phylogenetic tree was created by the neighbour-joining method and, subsequently, the tree was evaluated by bootstrap analysis (Fig. 4).

The phylogenetic tree separates the five C. glutamicum IMP paralogs (Fig. 4, bold) into distinct classes which all include apparent orthologs from other Actinobacteria. The denotation of the respective class was derived from members for which a physical function has been described (Fig. 4, box): CysQ in E. coli [53], ImpA in Mycobacterium smegmatis [54], SuhB in M. tuberculosis [55] and HisN (Cg0910) in C. glutamicum.

The putative orthologs of Cg0911 form a separate class. None of its members, comprising only corynebacterial and one Nocardioides species, was functionally characterized up to now.

Bacterial strains, plasmids, oligonucleotides and culture conditions

DNA isolation, manipulation and transfer

Plasmid DNA was extracted from E. coli and C. glutamicum by an alkaline lysis technique using the QIAprep Spin Miniprep Kit (Qiagen, Hilden, Germany) with a preliminary incubation for 2 h at 37°C in resuspension buffer P1 containing 50 mg ml^-1 lysozyme (185,000 U mg^-1; Serva, Heidelberg, Germany) for C. glutamicum. DNA modification, analyses by gel-electrophoresis and ligation were carried out by standard procedures [77]. Restriction endonucleases and T4 DNA ligase, together with enzyme buffers, were purchased from Fermentas (St. Leon-Rot, Germany) and used according to the manufacturer's instructions. DNA restriction fragments required for cloning were recovered from agarose gels by using QIAEX II gel extraction kit (Qiagen). E. coli and corynebacterial cells were transformed by electroporation [78] using the Bio-Rad Gene Pulser system (Bio-Rad, Munich, Germany).

Transformation of C. glutamicum

The artificial transposon vector pAT6100 was electrotransferred into the restriction-deficient strain C. glutamicum RES167 by means of a method previously described [79]. To force transposition, the cells were additionally incubated by shaking at 200 rpm for 50 min at 37°C directly after electroporation procedure before pelleting and spreading them on solid complex media plates containing 25 μg ml^-1 kanamycin to select transformants. Kanamycin-resistant colonies were picked with sterile sticks, transferred to separate wells of 96-well microtiter plates (Greiner, Solingen, Germany) containing 200 μl selective complex medium and inoculated for 48 h at 30°C. Long-term storage was carried out in 96-well microtiter plates containing 10% Hogness modified freezing medium (HMFM), composed of 87% (v/v) glycerol, 1.7 mM tri-sodium citrate, 6.8 mM (NH₄)₂SO₄, 0.4 mM MgSO₄ · 4 H₂O, 36 mM Na₂HPO₄ · 2 H₂O, 13.2 mM KH₂PO₄, and 90% TSB at -80°C.

Determination of transposon integration sites

Genomic DNA from C. glutamicum transposon mutants was extracted using the GenElute Bacterial Genomic DNA Kit (Sigma-Aldrich, Taufkirchen, Germany). Cloning and sequencing of insertion sites were carried out by plasmid rescue techniques as described before [20] with the exception that the chromosomal DNA was digested with EcoRI and XbaI to clone either side of the insertion. Sequencing of the resulting plasmids was performed with the synthetic oligonucleotide primers s6100e and s6100x for the EcoRI and the XbaI sites, respectively, by IIT Biotech (Bielefeld, Germany).

Sequence data/accession numbers

Nucleotide homology searches were applied against the annotated C. glutamicum ATCC 13032 genome sequence [GenBank:BX927147] [10, 80] obtained from GenBank [81]. Amino acid sequences of predicted inositol monophosphatase proteins were obtained from the Swissprot/TrEMBL databases [82]: Cg0910 [Swiss-Prot:Q8NS80], Cg0911 [Swiss-Prot:Q6M6Y2], CysQ [Swiss-Prot:Q8NS37], SuhB [Swiss-Prot:Q6M4B2] and ImpA [Swiss-Prot:Q8NNT8]. Further protein sequence information used for comparative analyses were retrieved from the following GenBank genome entries: [GenBank:AE014295] Bifidobacterium longum NCC2705, [GenBank:BA000035] C. efficiens YS-314, [GenBank:BX248353] C. diphtheriae NCTC13129, [GenBank:CR931997] C. jeikeium K411, [GenBank:U00096] Escherichia coli K-12, [GenBank:CP000249] Frankia sp. CcI3, [GenBank:AE016822] Leifsonia xyli subsp. xyli str. CTCB07, [GenBank:AE016958] Mycobacterium avium subsp. paratuberculosis K-10, [GenBank:AL450380] M. leprae TN, [GenBank:AL123456] M. tuberculosis H37Rv, [GenBank:AP006618] Nocardia farcinica IFM 10152, [GenBank:AE017283] Propionibacterium acnes KPA171202, [GenBank:BA000030] Streptomyces avermitilis MA-4680, [GenBank:AL645882] S. coelicolor A3(2), [GenBank:CP000088] Thermobifida fusca YX, [GenBank:AAHG00000000] Arthrobacter sp. FB24, [GenBank:AAGP00000000] Brevibacterium linens BL2, [GenBank:AAII00000000] Frankia sp. EAN1pec, [GenBank:AAMN00000000] Janibacter sp. HTCC2649, [GenBank:AAEF00000000] Kineococcus radiotolerans SRS30216, [GenBank:AAJB00000000] Nocardioides sp. JS614, and [GenBank:AAEB00000000] Rubrobacter xylanophilus DSM 9941.

Bioinformatic analyses and tools for interpretation of C. glutamicum sequences

Database searches and sequence similarity-based searches with nucleotide and amino acid sequences were performed with the BLASTN- and BLASTP- algorithms [83], respectively. The multiple amino acid sequence alignment of the inositol monophosphatases-family proteins was performed with the DIALIGN2 software [84]. The phylogenic tree was calculated using the neighbour-joining method [85] integrated in the CLUSTALX software package [86] and visualized as a radial tree with the interactive phylogenetic tree plotting program TreeTool [87]. Sequence pattern searches were performed with the TEIRESIAS algorithm [33] provided by the IBM Bioinformatics Group.

Auxotrophy screening and supplementation of auxotrophic phenotypes

The identification of auxotrophic mutants was performed by parallel plating on minimal media MM1 (MMYE without yeast extract [88]) and TSB complex media. The auxanographic characterizations were carried out by picking the colonies on MM1 plates supplemented with growth factors which were added as sterile filtered stock solutions by means of the purchaser's instructions, following a plating pattern described by Holliday [34], and 25 μg ml^-1 kanamycin. The list of growth factors comprises 20 biogenic L-amino acids (alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, valine; final concentrations 1 mM each), nucleic acid bases (adenine, cytosine, guanine, thymine, uracil; 1 mM final cc. each), vitamins and precursors (thiamine (vitamin B1; 500 μg l^-1), riboflavin (B2; 200 μg l^-1), nicotinic acid (B3; 400 μg l^-1), pantothenic acid (B5; 400 μg l^-1), pyridoxine (B6; 400 μg l^-1), cobalamin (B12; 10 μg l^-1), folic acid (20 μg l^-1), p-aminobenzoic acid (200 μg l^-1), inositol (20 μg l^-1), ornithine (1 mM), and hypoxanthine (1 mM)). The chemicals were purchased from VWR (Darmstadt, Germany) and Merck.

PCR techniques and conditions

DNA fragments for deletion and complementation experiments were amplified from chromosomal DNA of C. glutamicum ATCC 13032 using Pwo DNA polymerase (Invitrogen), and Taq DNA polymerase (Peqlab, Erlangen, Germany) for the verification of chromosomal deletions. PCR reaction conditions were as follows: initial denaturation at 94°C for 2 min followed by 35 cycles of denaturation for 30 s, annealing for 1 min at a primer-dependent temperature, extension at 72°C for 2 min and a final extension for 4 min at 72°C. PCR products were purified using the QIAquick PCR purification kit (Qiagen).

PCR mutant screening was carried out with Eppendorf Taq DNA polymerase (Eppendorf) with pooled transposon mutants as templates, IS6100-specific primers IS6100x and IS6100y for both possible orientations of the transposon and a position-specific search primer gs_trpP (Table 3). The basic procedure was adapted from a method published by Hobom et al. [44]. Iterative rounds of PCR experiments were performed in order to identify the clone of interest in the library. In the first round clone-pools were examined containing all mutants in one microtiter plate. Sequential PCR rounds were carried out with the mutant-pools representing the clones of those individual plates that were tested positively in prior rounds until the exact position of the clone in the library could be determined. The bacterial cells were directly set in the PCR reaction and lysed with an initial PCR step of 94°C for 10 min to release the chromosomal DNA. PCR reaction conditions were as follows: initial denaturation at 94°C for 10 min followed by 40 cycles of denaturation for 30 s, annealing for 45 s at a primer-dependent temperature, extension at 72°C for 2 min and a final extension for 4 min at 72°C. PCR experiments were performed with the delivered buffers according to the manufacturer's instructions and with the use of a DNA Engine DYAD thermocycler from MJ Research (Watertown, MA, USA).

Deletion and genetic complementation of cg0910

Defined chromosomal deletions within the cg0910 gene were constructed with the pK18mobsacB vector system which allows to identify an allelic exchange by homologous recombination [5]. The deletion was introduced into the target gene by "gene SOEing" [51]. Therefore, upstream and downstream regions of the gene to be deleted were amplified by two different PCR reactions using primer pairs cg0910_del1+2 and cg0910_del3+4 (Table 3), respectively. After purification the resulting amplificates were used as templates for the second round of PCR. The final products were digested with the restriction enzymes BamHI and XhoI, corresponding to the cleavage sites introduced via the PCR primers, and subsequently cloned into appropriately digested pK18mobsacB vector. The ligation mixture was used to transform E. coli DH5αMCR and the transformants were selected on TSB plates containing 50 μg ml^-1 kanamycin and 40 mg l^-1 X-Gal (5-bromo-4-chloro-indolyl-β-D-galactopyranoside). The resulting deletion plasmid pK18mobsacB-Δcg0910 was extracted and transformed into C. glutamicum ATCC 13032. Integration of the introduced plasmid into the chromosome by single-crossover was selected on TSB plates supplemented with 25 μg ml^-1 kanamycin. The kanamycin-resistant clones were grown overnight in liquid media and spread on TSB plates containing 10% (w/v) sucrose. Colonies from this plates were tested for the desired kanamycin-sensitive and sucrose-resistant phenotype by parallel picking. PCR experiments were used to verify the deletion in the C. glutamicum chromosome. Construction of plasmid pEC-XT99A-cg0910 ex for in trans- expression of the Cg0910 protein was initiated with the amplification of the corresponding gene by PCR using the primer pair cg0910_ex1 and cg0910_ex2 (Table 3). Primer cg0910_ex1 was provided with a sequence of an optimal ribosome binding site upstream from the cg0910 start codon. The resulting 821-bp DNA fragment was purified and cleaved using MunI and BamHI. Appropriate restriction sites were added with the 5'-extension of the PCR primers. The fragment was subsequently ligated into the pEC-XT99A vector, which was prior cut with the corresponding enzymes. The ligation mixture was used for transformation of E. coli DH5αMCR, the transformants were selected on TSB plates containing 5 μg ml^-1 tetracycline.