A Corynebacterium glutamicum mutant with a defined deletion within the rplK gene is impaired in (p)ppGpp accumulation upon amino acid starvation

Bacteria have a common adaptational response to amino acid depletion, termed the stringent response, which is mediated by the hyperphosphorylated guanosine nucleotides (p)ppGpp. The accumulation of (p)ppGpp also occurs in response to a variety of other environmental or nutrient stresses such as heat shock, cold shock and carbon source or phosphate limitation. An elevated (p)ppGpp level leads to a complex modulation of gene expression, including the restriction of stable RNA synthesis, the stimulation of certain anabolic genes, and the induction of stationary-phase-specific genes (reviewed by Cashel et al., 1996 ).

There are two different sources of (p)ppGpp accumulation in Escherichia coli, depending on the nature of the nutritional stress imposed. The RelA protein, responsible for the (p)ppGpp biosynthesis during amino acid deprivation, is a ribosome-associated enzyme catalysing the transfer of a pyrophosphoryl group from ATP to the 3'-hydroxyl position of GTP or GDP. The enzyme is active when uncharged codon-specific tRNAs bind to the ribosomal A-site (Cochran & Byrne, 1974 ; Goldman & Jakubowski, 1990 ). Furthermore, a functional 50S ribosomal protein L11, encoded by the rplK gene, is required for the activation of the RelA protein and thus for the synthesis of (p)ppGpp (Friesen et al., 1974 ). However, relA mutant strains continue to accumulate (p)ppGpp in a spoT gene-dependent manner during energy-source exhaustion (Ryals et al., 1982 ; Murray & Bremer, 1996 ). The absence of all detectable (p)ppGpp in relA/spoT double mutants (Xiao et al., 1991 ) and the fact that RelA and SpoT have extensive amino acid sequence similarity throughout their length (Metzger et al., 1989 ) led to the proposal that SpoT is a bifunctional enzyme, responsible for (p)ppGpp degradation and for the second source of (p)ppGpp-forming activity in E. coli (Murray & Bremer, 1996 ).

In contrast to the situation described for E. coli, data from genome sequencing projects of Mycobacterium tuberculosis (Cole et al., 1998 ), Bacillus subtilis (Kunst et al., 1997 ), and Mycoplasma genitalium (Fraser et al., 1995 ) revealed the presence of only one relA/spoT homologue in each of these bacteria. Moreover, the relA/spoT homologous genes from Streptococcus equisimilis H46A and Streptomyces coelicolor A3(2) are bifunctional, encoding a (p)ppGpp synthetase and a (p)ppGpp degrading activity (Mechold et al., 1996 ; Martinez-Costa et al., 1998 ). These findings supported the idea that in some bacteria only one gene might substitute for both functions encoded by the relA and spoT genes of E. coli. However, in B. subtilis and S. coelicolor A3(2) it has been shown that, like in E. coli, a functional rplK gene product is required for (p)ppGpp accumulation upon amino acid starvation (Smith et al., 1980 ; Ochi et al., 1997 ).

The Gram-positive soil bacterium Corynebacterium glutamicum is the most important micro-organism involved in the industrial production of amino acids such as L-glutamate, L-lysine and L-isoleucine (Malumbres & Martin, 1996 ). In the past decade, molecular genetic techniques have been developed for this micro-organism and numerous genes involved in amino acid biosynthesis have been cloned and characterized. Since then, the industrial production of selected amino acids has been improved by genetic engineering and amplification of relevant structural genes. More recently, a relA/spoT homologous gene from C. glutamicum, termed rel, has been reported and shown to encode a bifunctional enzyme with (p)ppGpp synthetase and (p)ppGpp degrading activities (Wehmeier et al., 1998 ). Moreover, RNA synthesis and growth rate of Brevibacterium flavum are inversely correlated with the cellular ppGpp concentration (Ruklisha et al., 1995 ), supporting an essential role of the rel gene in the stringent response of this bacterium.

In the present study, we continue our work on the (p)ppGpp metabolism of C. glutamicum with the identification and molecular characterization of the rplK gene showing significant similarity to ribosomal large-subunit proteins L11. (p)ppGpp accumulation assays in wild-type and in rplK mutant strains strongly indicates that the rplK gene is indispensable for (p)ppGpp accumulation following amino acid deprivation in C. glutamicum.

Bacterial strains, plasmids and growth conditions.
All bacterial strains and plasmids used in this study are listed in Table 1. E. coli and C. glutamicum strains were routinely grown in LuriaBertani (LB) medium (Sambrook et al., 1989 ) at 37 °C and 30 °C, respectively. Antibiotics used for plasmid selection were kanamycin (50 µg ml^-1 for E. coli; 25 µg ml^-1 for C. glutamicum) and chloramphenicol (50 µg ml^-1 for E. coli; 10 µg ml^-1 for C. glutamicum).

Table 1. Bacterial strains and plasmids used in this work

DNA manipulation and DNA transfer.
Plasmid DNA from E. coli was extracted by the alkaline lysis technique (Sambrook et al., 1989 ). Chromosomal DNA of C. glutamicum was isolated as described previously (Tauch et al., 1995 ). DNA restriction, agarose gel electrophoresis, Klenow treatment and ligation were carried out as described by Sambrook et al. (1989) . Enzymes for DNA manipulation were purchased from Pharmacia or from Boehringer. DNA restriction fragments required for cloning were isolated from agarose gels using the US BiocleanMP Kit (United States Biochemicals). E. coli DH5αMCR (Grant et al., 1990 ) was used for all cloning procedures. Plasmid DNA was introduced into E. coli strains by electroporation (Tauch et al., 1994 ). Intergeneric conjugal transfer of plasmids from E. coli to C. glutamicum followed the mating procedure described by Schäfer et al. (1990) . Transconjugants were selected on LB agar containing 50 µg nalidixic acid ml^-1 and 25 µg kanamycin ml^-1.

Polymerase chain reactions.
The synthetic oligonucleotides P1-up (5'-AGGAGCAGGCTGTTGTCACC-3'), P1-down (5'-extension-CGCCGTGAGC-5'-GCCAACTGGAGGAGCAGGGT-3'), and P2-up (5'-extension-TCCAGTTGGC-5'-GCTCACGGCGTCAACATCAG-3'), P2-down (5'-GCGGATAGCTACGTCGATGG-3') were used for the amplification of two 600 bp fragments comprising the 5' and 3' regions of the C. glutamicum rplK gene, respectively, with Pfu DNA polymerase (Stratagene). The construction of the chromosomal 12 bp deletion within the rplK gene was analysed with the primers Pdel-up (5'-TGGCCCAGCACTTGGT-3') and Pdel-down (5'-GCAATAGCAACGTCGA-3'). The oligonucleotides P1-up and P2-down were used to amplify a fragment comprising the rplK wild-type gene. PCR experiments were carried out with a PCT-100 thermocycler (MJ Research). Initial denaturation was conducted at 94 °C for 2 min followed by 90 s denaturation, 90 s annealing at 58 °C and extension at 72 °C for 2 min. This cycle was repeated 30 times followed by a final extension step for 10 min at 72 °C. PCR products were purified using the PCR Purification Spin Kit (Qiagen).

Southern hybridization.
Chromosomal DNA of C. glutamicum was digested with restriction enzymes, separated on 0·8% agarose gels, and blotted onto Hybond-N nylon membranes (Amersham) using the vacuum blotter Vacu-Gene (Pharmacia). Fixation of DNA, labelling of DNA probes and hybridization were performed with the non-radioactive DIG DNA Labelling and Detection Kit from Boehringer.

DNA sequencing and sequence analysis.
During analysis of transposon insertions in C. glutamicum, the Tn5516invers insertion of the transconjugant C. glutamicum KM6 was cloned by applying the plasmid rescue technique (Tauch, 1996 ). For this purpose, total DNA of C. glutamicum KM6 was isolated, cut with SalI and religated. The ligation mixture was subsequently transformed into E. coli DH5αMCR. The resulting plasmid pKM6-6, carrying a 5·8 kb contiguous DNA fragment of the C. glutamicum chromosome, was used as a template for nucleotide sequencing by primer walking, performing the recommended procedures of the Prism Ready Reaction Dye Deoxy Termination Kit (Applied Biosystems) in conjunction with a Prism 377 DNA sequencer (Applied Biosystems). Computer-assisted compilation of the DNA sequences was conducted with the Staden software package (Staden, 1986 ). Searches for amino acid similarities were carried out with the BLAST programs (Altschul et al., 1997 ). Global alignments between protein sequences were calculated with the ALIGN computer program (Myers & Miller, 1988 ). The prediction of protein secondary structures was calculated by the neural network system PredictProtein (Rost et al., 1995 ).

Screening for spontaneous and induced thiostrepton-resistant C. glutamicum mutants.
For obtaining spontaneous thiostrepton-resistant colonies, different amounts of an exponentially growing culture (10⁶10¹² cells) were spread on LB agar containing 1, 2·5, 5, or 10 µg thiostrepton ml^-1, respectively, and were incubated for 10 d at 30 °C. The screening for induced thiostrepton-resistant mutants was performed by chemical mutagenesis using ethyl methanesulfonate (EMS). An exponentially growing culture was concentrated 10-fold by centrifugation and resuspended in 0·1 M sodium phosphate buffer (pH 7·0) containing 1 mg EMS ml^-1 and incubated at 30 °C for 1 h. Cells were then washed twice with LB medium and plated directly onto LB agar supplemented with different thiostrepton quantities (see above) and incubated at 30 °C.

Construction of a plasmid-encoded rplK deletion derivative.
A 1·2 kb fragment comprising a 12 bp deletion derivative of the rplK gene was amplified from the C. glutamicum chromosome using PCR with the oligonucleotides P1-up, P1-down, P2-up and P2-down (illustrated in Fig. 3). The PCR primers were deduced from the known nucleotide sequence (P1-up, nt 32913310; P1-down, nt 39163897; P2-up nt 39293948; P2-down, nt 45344515) in such a way that P1-down and P2-up contained 10 bp 5' extensions which matched exactly the 5' end of P2-up (nt 39293938) and P1-down (nt 39163907), respectively. First, two separate 600 bp fragments were amplified from the target sequence. The first product, termed rplk-part1, was obtained with P1-up and P1-down and comprised the 5' part of the rplK gene (nt 187) together with a 10 bp extension corresponding to nt 100109 of rplK, thereby carrying a 12 bp gap within the amplified rplK portion compared to the used chromosomal template. The second fragment, named rplk-part2, was obtained with P2-up and P2-down and consisted of the rplK 3' region (nt 78435) carrying the same 12 bp gap ranging from nt 8899 within the amplified rplK part. Accordingly, these two rplK PCR products shared a 20 bp overlap region. When these intermediate products were mixed together, melted and reannealed, the top strand of rplk-part1 could anneal to the bottom strand of rplk-part2 in such a way that the two strands acted as primers on one another. Extension of this overlap and addition of the primers P1-up and P2-down created a 1·2 kb fragment which carried a 12 bp deletion derivative of the rplK gene. The purified 1·2 kb PCR product was subsequently cloned into the SmaI-digested E. coli vector pK18mobsacB, yielding pΔRPLK. Therefore, pΔRPLK comprised a mutant rplK gene, in which the 12 bp deletion resulted in the loss of the tetrapeptide Pro-Ala-Leu-Gly in the corresponding L11 protein.

(23K): Fig. 3. Strategy to obtain a PCR fragment comprising a deletion derivative of the C. glutamicum rplK gene. PCR with the primers P1-up, P1-down and P2-up, P2-down results in two products comprising the 5' (rplk-part1) and 3' parts (rplk-part2) of the C. glutamicum rplK gene, respectively. Due to the 5' extensions of P1-down and P2-up, both fragments contain the same gap within their respective rplK portion and share an overlap which enables annealing of the two intermediate products. In a further PCR reaction, extension of this overlap and the addition of primers P1-up and P2-down create the recombinant product rplk-del, consisting of an rplK deletion derivative (for further details see Methods).

Measurement of (p)ppGpp accumulation.
E. coli and C. glutamicum strains were screened for patterns of (p)ppGpp accumulation following the rapid screening method described previously (Cashel, 1994 ). Non-uniformly ³²P_i-labelled E. coli and C. glutamicum cells were obtained by growing the strains overnight on LB plates, followed by dispersion of single colonies in phosphate-free MOPS minimal medium and incubation with MOPS medium containing 10 µCi ³²P_i ml^-1 (3·7x10⁵ Bq ml^-1) for 30 min at 37 °C. For amino acid starvation responses, the colonies were resuspended in amino-acid-free, phosphate-free MOPS medium supplemented with DL-serine hydroxamate (1 mg ml^-1) and L-valine (0·5 mg ml^-1). The samples were mixed with 13 M formic acid and three freezethaw cycles were performed. Afterwards, the samples were centrifuged and 5 µl samples of supernatants were subjected to one-dimensional polyethyleneimine (PEI) TLC developed with 1·5 M potassium phosphate (pH 3·4). Nucleotide abundance was determined by autoradiography after overnight exposure of the PEI chromatograms at -70 °C to Cronex 4 films (Sterling). Analysis of the autoradiograms was performed with the help of a HP Scanjet 6100 C/T optical scanner (Hewlett Packard). The spot intensities were quantified densitometrically using the WinCam computer program (Cybertech). Accumulation data concerning the residual (p)ppGpp level are means of two independent PEI TLC assays.

Transcripitional assay.
RNA of C. glutamicum was isolated by means of the Ribolyser instrument and the Hybaid Ribolyser RNA isolation kit. Blotting and hybridization experiments were performed as described by Sambrook et al. (1989) . Hybridization signals were detected with the chemoluminescence substrate CSPD (Roche) and quantified densitometrically using the LabImage computer program (Labsoft). Gene probes of rplK and rplA were generated by PCR amplification using the Pwo DNA polymerase and the oligonucleotides rplK1 (nt 37583777), rplK2 (nt 42734292), rplA1 (nt 44904509) and rplA2 (nt 49764995), respectively. Subsequently, the PCR products were cloned into the vector pZEro-2 (Invitrogen). In vitro labelling of the cloned DNA fragments was performed with the DIG DNA labelling kit (SP6/T7) from Roche.

The C. glutamicum rplK gene encoding the large ribosomal subunit protein L11 is part of a DNA region also comprising secE, nusG and rplA
During analysis of transposon insertion sites in C. glutamicum, a 5·8 kb SalI DNA fragment was rescued from the chromosome of C. glutamicum KM6 (Tauch, 1996 ). The resulting plasmid, termed pKM6-6, carried the chromosomal 5'- and 3'-flanking regions of the transposon insertion which are approximately 0·6 kb and 5·2 kb in length (Fig. 1). Subsequently, pKM6-6 was cut with HindIII to construct the deletion derivative pKM6-67 (Tauch, 1996 ). DNA single-read sequence analysis of pKM6-67 using the universal sequencing primer (Fig. 1) revealed significant similarity to internal fragments of rplK genes from Gram-positive and Gram-negative bacteria (data not shown). Since the E. coli rplK gene is known to be involved in (p)ppGpp accumulation during amino acid starvation (Friesen et al., 1974 ), we decided to sequence the isolated rplK region.

(12K): Fig. 1. Restriction map of the contiguous 5·8 kb rplK region from C. glutamicum ATCC13032. The positions of the open reading frames (grcC, secE, nusG, rplK, rplA) and the three tRNA genes (tRNAThr copyA, tRNAThr copyB, tRNATrp) are indicated by arrows. The locations of the secE promoter fragment (P37; Patek et al., 1995 ) and the transcriptional terminators (Ω) of grcC, secE, and downstream of rplA are marked. The insertion site of Tn5516invers, consisting of two IS1249 elements flanking pK18mob, is indicated. The HindIII restriction sites within Tn5516invers and the chromosomal fragment, used to construct a deletion derivative of the originally rescued plasmid, are typed in boldface. The position of the initial DNA single-read analysis with the universal sequencing primer (uni) is marked by an arrow.

The nucleotide sequence of a 5245 bp SalIBglII chromosomal fragment present on pKM6-6 was determined by a primer walking strategy. The overall G+C content of the sequenced region is 52·1 mol%, which is in the range known from C. glutamicum chromosomal DNA (Malumbres et al., 1993 ). A computer-assisted analysis of the coding probability (Staden, 1986 ) revealed the presence of five ORFs with protein-coding character (Fig. 1). One of these ORFs, termed rplK (nt 38304267), encodes a protein of 145 amino acids with a calculated molecular mass of 15329 Da and a deduced isoelectric point of 9·78. A potential ribosome-binding site (AAGGAA) is located 11 nucleotides upstream of the putative ATG start codon. Within the upstream region, neither E. coli consensus promoter sequences (Hawley & McClure, 1983 ) nor any homology with conserved -35 and -10 regions of C. glutamicum (Patek et al., 1995 ) could be identified. Sequence comparison of the deduced RplK protein revealed significant similarity to ribosomal large-subunit proteins L11 (RplK) from other bacteria, including M. tuberculosis (EMBL Z92772), B. subtilis (EMBL Z99104), Streptomyces griseus (Küster et al., 1998 ) and E. coli (EMBL U00006). The highest similarity was found to be with the M. tuberculosis RplK protein, sharing 85% amino acid identity and 88% similarity, by performing a global protein alignment (Fig. 2). PROSITE database searches with the deduced protein sequence identified a ribosomal protein L11 signature (PS00359) at the C-terminal end of the C. glutamicum RplK protein ranging from amino acid 130 to 145. Furthermore, the neural network system PredictProtein (Rost et al., 1995 ) indicated three α-helical structures within the C-terminal half of the protein. This part of the RplK protein has been shown to exist in an extended and loosely folded conformation and is likely to be buried within the ribosomal structure in E. coli and Bacillus stearothermophilus (Choli, 1989 ; Xing & Draper, 1996 ).

(16K): Fig. 2. Global amino acid sequence alignment of the deduced RplK protein from C. glutamicum (C. glu) with RplK from M. tuberculosis (EMBL Z92772) (M. tub). Identical amino acid residues are typed in bold and marked by asterisks; dots represent positions of conservative substitutions. Gaps (-) were introduced by the ALIGN computer program to optimize the protein alignment. Numbers on the right correspond to amino acid positions relative to the start of each protein. The position of the tetrapeptide Pro-Ala-Leu-Gly, which is deleted in the C. glutamicum RplK mutant protein, is indicated (-Δ-). The C-terminal α-helical domains IIII and the amino acid residues of the ribosomal protein L11 signature (boxed) are marked.

The downstream ORF, termed rplA, revealed significant amino acid sequence similarity to the ribosomal large-subunit protein L1 (RplA) from a variety of micro-organisms (Table 2). Two additional complete ORFs are located upstream of rplK (Fig. 1 and Table 2). The nusG gene is likely to encode a transcriptional antiterminator protein. The protein deduced from the secE gene displayed similarity to SecE proteins from several bacteria (Table 2). SecE proteins play an essential role in the bacterial protein translocation system.

Table 2. Coding regions in the C. glutamicum rplK gene region

The wild-type rplK gene of C. glutamicum was replaced by an allele carrying a 12 bp in-frame deletion in a highly conserved N-terminal region
E. coli rplK mutants are unable to synthesize (p)ppGpp in response to amino acid starvation (Friesen et al., 1974 ). Several attempts to disrupt the rplK coding region of C. glutamicum by integration of a chloramphenicol resistance gene or to delete larger parts of the rplK gene failed (data not shown), indicating the essential nature of the RplK protein in this species. Therefore, we created a C. glutamicum mutant strain carrying a 12 bp in-frame deletion within a highly conserved N-terminal region of rplK which results in the loss of the tetrapeptide Pro-Ala-Leu-Gly from the L11 protein (Figs 2 and 3).

A DNA fragment comprising the rplK deletion derivative was obtained by a PCR strategy and was cloned into pK18mobsacB, yielding pΔRPLK. A detailed experimental procedure of this amplification is outlined in Methods and in Fig. 3. Subsequently, pΔRPLK was transferred to E. coli S17-1 and integrated into the chromosome of C. glutamicum RES167 by conjugation resulting in C. glutamicum LW45. Excision of the plasmid can be selected for by growing the cells on LB agar containing 10% sucrose. Cells able to grow on this medium have lost the plasmid due to a second crossover event that either restores the wild-type gene arrangement or leads to a mutant strain carrying a defined 12 bp deletion in the chromosome. However, all 97 clones tested contained the wild-type gene arrangement.

In a further approach, a DNA fragment comprising the wild-type rplK gene was obtained by PCR and was cloned into the SmaI-digested E. coliC. glutamicum shuttle vector pECM3. The resulting plasmid, pRPLK, was subsequently transferred to C. glutamicum LW45 by conjugation. With the wild-type rplK copy provided in trans, we were able to select a second crossover and to replace the chromosomal rplK gene which was verified by sequencing of a chromosomal rplK PCR fragment originating from a sucrose-resistant clone (data not shown). The elimination of the plasmid-encoded rplK wild-type copy from the cell was subsequently achieved by heat-curing of pRPLK (Schäfer et al., 1994a ). The complete loss of pRPLK was verified by Southern blot experiments using labelled pRPLK as a probe (data not shown). Finally, the resulting strain C. glutamiucm RES167 ΔrplK carried a chromosomal 12 bp deletion within the rplK gene corresponding to the tetrapeptide Pro-Ala-Leu-Gly from the L11 protein.

The deletion within the rplK gene leads to a tolerance of C. glutamicum against thiostrepton
The RplK protein of E. coli forms part of the target site for the antibiotic thiostrepton, which binds the same domain of the 23S rRNA (Ryan et al., 1991 ). Therefore, mutations in the RplK protein render ribosomes thiostrepton resistant (Friesen et al., 1974 ). Since our former attempts to isolate rplK mutants of C. glutamicum by screening for spontaneous or induced thiostrepton-resistant colonies failed, we have analysed whether the constructed deletion within the rplK gene confers thiostrepton resistance in C. glutamicum. The growth of C. glutamicum RES167 was inhibited in the presence of 60 ng ml^-1 thiostrepton in liquid growth medium expressed as the 50% value of the OD₅₈₀, whereas the rplK mutant strain RES167 ΔrplK grew in the presence of thiostrepton concentrations of up to 150 ng ml^-1. Thus, the C. glutamicum rplK mutant exhibited a 2·5-fold increased tolerance of thiostrepton.

The C. glutamicum rplK mutant is significantly impaired in (p)ppGpp accumulation upon amino acid starvation
To assess the effect of the mutant rplK gene on (p)ppGpp accumulation, C. glutamicum RES167 ΔrplK was subjected to amino acid starvation. Conditions of amino acid deprivation were induced by supplementation with 1 mg DL-serine hydroxamate ml^-1, which inhibits tRNA^Ser aminoacylation, and with 0·5 mg L-valine ml^-1, which resulted in isoleucine starvation (Cashel et al., 1996 ). Under such stress conditions, we detected (p)ppGpp in nucleotide extracts of C. glutamicum RES167 which co-migrates with (p)ppGpp synthesized by the E. coli wild-type (Fig. 4, lanes 1 and 3). This starvation response was completely abolished in E. coli CF1652 (ΔrelA) and C. glutamicum RES167 Δrel, a mutant lacking a functional (p)ppGpp synthetase (Wehmeier et al., 1998 ) (Fig. 4, lanes 2 and 4). The C. glutamicum rplK mutant is significantly impaired in its ability to form (p)ppGpp following nutritional shift-down (Fig. 4, lane 5). A computer-assisted analysis of the spot intensities, using the WinCam program, revealed only 14±5% (p)ppGpp compared to the (p)ppGpp level detected in the C. glutamicum wild-type. The (p)ppGpp accumulation assignment was verified by comparing the nucleotide activities of a limited series of sample dilutions (data not shown). The ability of C. glutamicum RES167 ΔrplK to produce (p)ppGpp could be completely restored by introducing a plasmid-encoded rplK wild-type allele into the mutant strain (Fig. 4, lane 6), indicating that the rplK gene is required for (p)ppGpp accumulation upon amino acid deprivation.

(121K): Fig. 4. (p)ppGpp accumulation of C. glutamicum wild-type and the rplK mutant in response to amino acid starvation. Lanes: 1, E. coli CF1648, wild-type; 2, CF1652, ΔrelA; 3, C. glutamicum RES167, restriction-deficient wild-type derivative; 4, Δrel derivative of RES167; 5, ΔrplK RES167 strain; 6, ΔrplK host of RES167, carrying plasmid pRPLK. Cultures were non-uniformly labelled with 32Pi and amino acid starved after growth on LB plates using DL-serine hydroxamate (1 mg ml-1) and L-valine (0·5 mg ml-1) as described in Methods. The positions of standards after PEI-TLC development in 1·5 M KH2PO4 are indicated at the right. C. glu, C. glutamicum.

The rplK and rplA genes of C. glutamicum form a bicistronic operon
The results obtained by complementation analysis indicated that the rplK gene is involved in (p)ppGpp accumulation under amino acid starvation conditions. Nevertheless, another gene encoding a ribosomal protein (rplA) is located in close proximity to rplK (Fig. 1), suggesting a coordinate expression of these genes. In order to study the organization of the rplK and rplA genes of C. glutamicum, Northern blot hybridization with total RNA of RES167 and RES167 ΔrplK and antisense RNA probes covering rplA and rplK was performed (Fig. 5). Two transcripts of 1·5 kb and 0·6 kb were observed on hybridization to the rplK-specific probe (Fig. 5a). On hybridization to the rplA-specific probe only the 1·5 kb signal was detected (Fig. 5b). Considering the size of the rplK and rplA genes, these results indicate that a bicistronic mRNA covering rplK and rplA is formed and that additionally a monocistronic mRNA covering only rplK is synthesized. The length of the transcripts coincides well with the genetic organization of the rplKrplA cluster (Fig. 5c). The same transcriptional patterns of the rplK and rplA genes were observed in C. glutamicum RES167 and in the defined mutant strain C. glutamicum RES167 ΔrplK with the exception that the amount of their mRNAs slightly increased in RES167 ΔrplK (Fig. 5).

(21K): Fig. 5. Transcriptional analysis of the rplKrplA cluster. (a) Hybridization of RNA of C. glutamicum strains with an rplK probe. Amino acid starvation was induced (ind) in C. glutamicum RES167 and C. glutamicum RES167ΔrplK as described in Methods. Total RNA (4 µg) was loaded in each lane. Size marker I (Roche) was used for length determination. The length of the transcripts was determined with the LabImage software (Labsoft). (b) Hybridization of RNA of C. glutamicum strains with an rplA probe. (c) Genetic organization of the gene cluster. The position of probes used for hybridization is marked. The two transcripts (mRNA1 and mRNA2) are shown by wavy lines. The presence of a putative transcriptional terminator (Ω) is indicated. C.g., C. glutamicum.

In E. coli, the rplK gene, encoding the large ribosomal subunit protein L11, was shown to be indispensable for the activation of the RelA protein and consequently for synthesis of the stringent response mediator (p)ppGpp upon amino acid starvation (Friesen et al., 1974 ). A chromosomal region comprising the entire rplK gene of C. glutamicum was cloned during analysis of transposon insertions by a plasmid rescue technique. The deduced RplK protein showed strong similarity to all known ribosomal L11 proteins. In addition, the gene arrangement of the isolated region from C. glutamicum matched exactly the organization found in high and low G+C Gram-positive bacteria including M. tuberculosis (EMBL Z92772), Str. coelicolor A3(2) (Puttikhunt et al., 1995 ), B. subtilis (EMBL Z99104), and Staphylococcus aureus (EMBL U96619). In all these organisms, rplK is preceded by two ORFs that are similar to nusG, encoding a transcription antiterminator protein, and secE, acting as a component of the protein export apparatus. The rplK gene is followed by an ORF homologous to rplA encoding the 50S ribosomal protein L1. Moreover, the corresponding DNA regions in several Gram-negative bacteria, for example E. coli (EMBL U00006) and Helicobacter pylori (Tomb et al., 1997 ), also appear extensively conserved.

E. coli rplK mutants are strongly impaired in their ability to accumulate (p)ppGpp following a nutritional shift-down and exhibit, like relA mutants, a relaxed control of RNA synthesis. Therefore, the rplK locus was originally termed relC (Cashel et al., 1996 ). Almost identical results have been obtained in several other bacteria including B. subtilis (Smith et al., 1980 ), Bacillus megaterium (Stark & Cundliffe, 1979 ), S. coelicolor A3(2) (Ochi et al., 1997 ), and S. griseus (Kawamoto et al., 1997 ), where mutations within the respective rplK/relC gene lead to an abolished stringent response. To determine the role of the C. glutamicum rplK gene in (p)ppGpp accumulation, we constructed and analysed an rplK mutant strain. All rplK mutants described so far have been isolated either by chemical mutagenesis or due to their property to confer thiostrepton resistance (Ochi, 1990 ). In contrast to these observations, thiostrepton-resistant C. glutamicum colonies could not be obtained under a variety of different screening procedures. Moreover, attempts to disrupt the rplK coding region, or to delete larger internal parts of rplK using the sacB system, which facilitates the detection of an allelic exchange by homologous recombination (Schäfer et al., 1994b ) were unsuccessful, providing evidence that the RplK protein might be essential in C. glutamicum. Finally, the strain C. glutamicum RES167 ΔrplK was constructed which carries a 12 bp in-frame deletion within the rplK gene, resulting in the loss of the tetrapeptide Pro-Ala-Leu-Gly from the L11 protein. The deleted sequence represents the most highly conserved region of protein L11 in Gram-positive and Gram-negative bacteria and was shown to nearly prevent (p)ppGpp accumulation under amino acid starvation conditions in S. griseus (Kawamoto et al., 1997 ) and S. coelicolor A3(2) (Ochi et al., 1997 ).

(p)ppGpp accumulation assays in C. glutamicum wild-type and the rplK mutant strain established that the rplK gene is required for (p)ppGpp accumulation. Synthesis of (p)ppGpp invoked by amino acid starvation was almost abolished in the rplK mutants but not in the wild-type strain. The residual (p)ppGpp level corresponding to approximately 14±5% in the rplK mutant compared with the C. glutamicum wild-type agrees well with the (p)ppGpp accumulation level of about 10% remaining in rplK mutant strains of E. coli (Friesen et al., 1974 ), S. griseus (Kawamoto et al., 1997 ) and S. coelicolor A3(2) (Ochi et al., 1997 ). The deletion within the C. glutamicum rplK gene leads only to a 2·5-fold increase in the tolerance against thiostrepton, which is much lower than the 14100-fold resistance known from the rplK mutant strains of S. griseus and S. coelicolor A3(2), respectively. However, the C. glutamicum wild-type is about 1327-fold more sensitive to thiostrepton than both Streptomyces parental strains. Western blotting analysis revealed that the ribosomes from the S. griseus and S. coelicolor A3(2) rplK mutants incorporate the corresponding mutant L11 protein normally, although the level of incorporation is only approximately one-third to one-eighth of that observed with the wild-type L11 protein (Kawamoto et al., 1997 ; Ochi et al., 1997 ). Since C. glutamicum RES167 ΔrplK, like the S. griseus rplK mutant, lacks the highly conserved internal 12 bp region, it seems likely that ribosomes from the C. glutamicum mutant strain also contain apparently lower amounts of mutant L11 protein, resulting in the strong impairment in (p)ppGpp accumulation. Despite the result of complementation in trans of the rplK mutant with the wild-type gene, we could not finally exclude the possibility that the defined 12 bp in-frame deletion within rplK may cause a polar effect on rplA which itself may eventually contribute to the defect in (p)ppGpp accumulation. Northern analysis revealed that the rplK and rplA genes of C. glutamicum form a bicistronic operon since a transcript covering both genes was detected. However, the same transcription patterns with slightly higher levels of the rplKrplA transcript were observed in the defined rplK mutant strain, suggesting that effects on the stability of the messenger can be neglected.

Recently, we isolated a relA/spoT homologous gene from C. glutamicum, designated rel, which probably encodes a bifunctional enzyme with (p)ppGpp synthetase and (p)ppGppase activities. A rel mutant strain failed to synthesize (p)ppGpp in response to amino acid starvation (Wehmeier et al., 1998 ). In the present study, we established that a functional 50S ribosomal protein L11 in C. glutamicum is indispensable for (p)ppGpp accumulation upon nutritional deprivation. Therefore, it appears possible that the C. glutamicum Rel protein, like the homologues of E. coli (Cochran & Byrne, 1974 ) and S. coelicolor A3(2) (Martinez-Costa et al., 1998 ), is ribosome-associated, requiring the RplK protein for activation of its (p)ppGpp synthetase function. With the molecular characterization of the rel and rplK genes from C. glutamicum we have identified two genes involved in (p)ppGpp synthesis in this industrially important micro-organism. Both genes are therefore involved in a global regulatory network response which might also influence amino acid production. This aspect is under current investigation in our laboratory.

We thank M. Cashel (Bethesda, USA) for kindly providing the E. coli strains CF1648 and CF1652. This work was supported by contract No. Bio-4-CT-960145 from the European Commission.