Mutations at Several Loci Cause Increased Expression of Ribonucleotide Reductase in Escherichia coli

Abstract

Production of deoxyribonucleotides for DNA synthesis is an essential and tightly regulated process. The class Ia ribonucleotide reductase (RNR), the product of the nrdAB genes, is required for aerobic growth of Escherichia coli. In catalyzing the reduction of ribonucleotides, two of the cysteines of RNR become oxidized, forming a disulfide bond. To regenerate active RNR, the cell uses thioredoxins and glutaredoxins to reduce the disulfide bond. Strains that lack thioredoxins 1 and 2 and glutaredoxin 1 do not grow because RNR remains in its oxidized, inactive form. However, suppressor mutations that lead to RNR overproduction allow glutaredoxin 3 to reduce sufficient RNR for growth of these mutant strains. We previously described suppressor mutations in the dnaA and dnaN genes that had such effects. Here we report the isolation of new mutations that lead to increased levels of RNR. These include mutations that were not known to influence production of RNR previously, such as a mutation in the hda gene and insertions in the nrdAB promoter region of insertion elements IS1 and IS5. Bioinformatic analysis raises the possibility that IS element insertion in this region represents an adaptive mechanism in nrdAB regulation in E. coli and closely related species. We also characterize mutations altering different amino acids in DnaA and DnaN from those isolated before.

INTRODUCTION

The synthesis of deoxyribonucleotides is catalyzed by the enzyme ribonucleotide reductase (RNR). In Escherichia coli, there are three different RNRs, each of which functions under distinct growth conditions (7). However, only the class Ia RNR, encoded by the nrdAB genes, is expressed under normal aerobic growth conditions. (Throughout this paper, we will refer to the NrdAB ribonucleotide reductase as RNR.) Synthesis of deoxyribonucleotides is of necessity a tightly controlled process, as deoxynucleoside triphosphate (dNTP) imbalances have mutagenic effects and are detrimental to the cell (15). For example, overexpression of RNR is mutagenic and at high levels can be toxic (8, 29). Regulation of nrdAB expression is complex as it is controlled by a number of transcriptional regulators, among them DnaA, Fis, ArgP (IciA), Crp, and NrdR.

RNR expression is also synchronized with the cell cycle, ensuring that dNTPs are synthesized when they are needed (28). DnaA is a key factor for control of replication initiation and is active in the ATP-bound form; it is converted to the inactive, ADP-bound form following initiation of DNA replication in a process that involves the proteins Hda and DnaN (13, 22). It has been proposed that the expression of nrdAB at the time of initiation of chromosome replication is influenced by the nucleotide-bound state (ATP or ADP bound) of DnaA (8, 18). Olliver et al. have proposed that the ratio of DnaA-ATP to DnaA-ADP controls RNR expression and that DnaA affects the relative levels of RNR, but not the timing of RNR expression (18).

The ability of RNR to catalyze its essential reaction depends strictly upon the cytoplasmic thioredoxin and glutaredoxin reducing pathways which maintain cysteine thiols in the reduced state (reviewed in reference 23). Both the thioredoxin and glutaredoxin pathways use electrons ultimately derived from NADPH to reduce protein substrates, including RNR, which is the only known essential substrate of these pathways. In the thioredoxin pathway, electrons flow from NADPH to thioredoxin reductase (encoded by trxB) and to thioredoxins 1 (Trx1, encoded by trxA) and 2 (Trx2, encoded by trxC). In the glutaredoxin/glutathione pathway, also using electrons from NADPH, glutathione reductase (gor) maintains the small molecule thiol glutathione (GSH) in the reduced state. GSH in turn reduces glutaredoxins 1, 2, and 3 (Grx1, -2, and -3, encoded by grxA, -B, and -C, respectively).

RNR is reduced by Trx1, Trx2, and Grx1, both in vitro and in vivo (Fig. 1). These three proteins are the only physiological electron donors that work efficiently enough for RNR reduction, as a strain that lacks all three (trxA trxC grxA) does not grow unless one of the three missing genes is expressed from a complementing plasmid (26). Grx2 and Grx3 do not reduce RNR effectively enough to allow growth of the trxA trxC grxA strain. Although Grx1 and Grx3 are quite similar in sequence and structure, Grx3 displays much lower activity toward RNR than Grx1 does (1).

Fig 1.

The glutaredoxin/glutathione and thioredoxin pathways for the reduction of ribonucleotide reductase (RNR). The thioredoxins Trx1 and Trx2 and glutaredoxin 1 (Grx1) can reduce RNR efficiently. Glutaredoxin 3 (Grx3; gray text) reduces RNR inefficiently and is not sufficient for the growth of cells which lack Trx1, Trx2, and Grx1.

However, we have found that certain mutations can alter the cell so that Grx3 is capable of reducing ribonucleotide reductase. These include mutations in the gene for Grx3 (grxC) that alter amino acid methionine 43 to a smaller hydrophobic acid, among them grxC(M43V). These mutations make Grx3 a better reductant of RNR (5). Alternatively, mutations that cause increased expression of the nrdAB operon overcome the weak affinity of Grx3 for RNR by mass action, thus generating sufficient reduction of ribonucleotides to permit bacterial growth. These mutations map to the dnaA and dnaN genes that encode components of the replication initiation machinery. Alteration of the ATP-bound state of DnaA by these mutations, either by defects in ATP binding or by rapid hydrolysis of ATP to ADP, results in derepression of RNR (7).

In this report, we set out to obtain mutations that would further increase the activity of Grx3(M43V), by selecting for increased growth of a strain carrying the mutation in a background lacking Trx1, Trx2, and Grx1. Such mutations could provide insight into structure-function relationships in the glutaredoxins (5, 6). However, instead of mutations that enhanced Grx3 activity, we obtained three classes of mutations that directly altered the expression of the nrdAB operon. These are (i) new mutations in the dnaA and dnaN genes, (ii) a mutation in the hda gene which encodes another protein involved in initiation of DNA replication, and (iii) insertion mutations in the nrdAB promoter region that increase its expression. The frequency of these latter insertion mutations and bioinformatic analysis of the nrdAB promoter region in organisms closely related to E. coli raise the possibility that the insertions upstream of this operon represent an adaptive mechanism to respond to oxidative stress conditions.

MATERIALS AND METHODS

Media and growth conditions.

All experiments were performed on rich (NZ amine) media (9) unless otherwise specified. Antibiotics were added at the following final concentrations: ampicillin, 200 μg/ml; chloramphenicol, 10 μg/ml; kanamycin, 40 μg/ml; tetracycline, 15 μg/ml; and streptomycin, 1.5 mg/ml. Arabinose or glucose was added at a final concentration of 0.2% for induction or repression from pBAD vectors.

Strain construction.

Strains and plasmids used in this work are described in Table 1. All strains are derivatives of DHB4 (3) unless otherwise specified. Strains were constructed using standard techniques for P1 transduction and electroporation (16). Strains which lacked multiple redoxin genes (e.g., RO36 [trxA trxC grxA nrdH/pBAD18-trxC]) were grown in the presence of arabinose for induction of the complementing gene from the plasmid, and all transductions were also plated on media with arabinose present.

Table 1.

Strains and plasmids used in this work^a

Strain or plasmid	Relevant genotype	Source or reference
Strains
DHB4	Δ(ara-leu)7697 araD139 ΔlacX74 galE galK rpsL phoR Δ(phoA)PvuII ΔmalF3 thi	3
FA173	trxA trxC nrdH::Sp/pBAD18-trxC	19
RO36	trxA trxC grxA::Km nrdH::Sp/pBAD18-trxC	19
NK21	trxA trxC grxA::Km grxC(M43V) nrdH::Sp/pBAD18-trxC	This work
MAF173	RO36 dnaA(A345G)	This work
MAF174	RO36 dnaA(S203T)	This work
MAF175	RO36 dnaA(N346D)	This work
MAF176	RO36 dnaA(S331P)	This work
MAF177	RO36 hda(F85V)	This work
NK111	NK21 dnaN(Q156S)	This work
NK113	NK21 dnaA(A362P)	This work
CAG18469	hda...Tn10(Tc)	25
CAG18499	dnaAN...Tn10(Tc)	25
SMG216	dnaAN...Tn10(Cm)	Lab collection
MAF657	DHB4 hda(F85V)...Tn10(Tc)	This work
MAF658	DHB4 hda+...Tn10(Tc)	This work
MAF715	DHB4 dnaAN+...Tn10(Cm)	19
MAF716	DHB4 dnaA(S203T)...Tn10(Cm)	This work
MAF717	DHB4 dnaA(N346D)...Tn10(Cm)	This work
MAF718	DHB4 dnaA(N156S)...Tn10(Cm)	This work
MAF719	DHB4 dnaA(A362P)...Tn10(Cm)	This work
MAF747	FA173/pBAD33-trxA	This work
MAF750	MAF747 grxA::Km	This work
MAF778	MAF750 hda(F85V)...Tn10(Tc)	This work
MAF779	MAF750 hda+...Tn10(Tc)	This work
MAF784	MAF750 dnaA(S203T)...Tn10(Tc)	This work
MAF785	MAF750 dnaAN+...Tn10(Tc)	This work
MAF786	MAF750 dnaA(N346D)...Tn10(Tc)	This work
CSH100	F′ lac proA+B+(lacIqlacPL8)/araD(gpt-lac)5	30
FW102	F−/araD(gpt-lac)5 rpsL	30
Plasmids
pBAD18-trxC		26
pBAD33-trxA		Lab collection
pFW11		30
F′ tacp-lacZ		This work
F′ nrdABp-lacZ		This work
F′ nrdABp(IS1)-lacZ		This work
F′ nrdABp(IS5)-lacZ		This work

^aRelevant genotypes are listed. All strains are derivatives of DHB4 unless otherwise specified.

The grxC(M43V) mutation was placed at the native chromosomal locus for grxC using the thyA selection/counterselection outlined in reference 31. Using the lambda red system, the grxC gene was first replaced by thyA in a strain that otherwise lacked thyA, allowing us to select for thymine prototrophy. In a second step, the grxC::thyA construct was replaced by the grxC(M43V) allele (amplified by PCR from pRO2) (19), using trimethoprim to select for thyA mutant transformants. The grxC gene was sequenced from the thymine auxotrophic, trimethoprim-resistant transformants to confirm the presence of grxC(M43V) on the chromosome. The grxC(M43V) allele was then moved into other strains by taking advantage of the fact that it is near cysE, and so it can be transduced into cysE mutant strains by selecting for cysteine prototrophy.

The Tn10 (Tet^r) transposon linked to hda disrupts the guaA gene, causing a growth defect on minimal media. The transposon was removed by P1 transduction of the guaA⁺ allele into the strain (MAF657) containing the region from hda(F85V) to Tn10 [hda(F85V)… Tn10] and selecting for guanine prototrophy. The removal of Tn10 was necessary for constructing nrdAB-lacZ reporter strains because the nrdAB promoter-lacZ fusion is in an F′ plasmid carrying a Tet^r selective marker.

Suppressor analysis.

Suppressors of RO36 (trxA trxC grxA nrdH/pBAD18-trxC) and NK21 [trxA trxC grxA nrdH grxC(M43V)/pBAD18-trxC] were isolated as described previously (19). These strains were first plated on media containing glucose (to repress the complementing trxC gene), and colonies (candidate suppressors) were purified by restreaking on media that also contained glucose. In order to show that the suppressor mutations were not linked to the complementing pBAD18-trxC plasmid, we sought to isolate derivatives of the suppressor strains that had spontaneously lost the plasmid. The suppressor strains were restreaked several times on media that lacked ampicillin, and single colonies were restreaked on media containing ampicillin in order to identify ampicillin-sensitive derivatives of the suppressor strains.

Sequencing to identify mutations in dnaA, dnaN, or hda was done using the primers described in Table 2. The presence of the hda(F85V) mutation in hda suppressor…Tn10 transductants was verified using either sequencing or restriction fragment length polymorphism (RFLP) analysis based on the fact that the F85V mutation had introduced a new BsaJI site in the gene. PCR using primers in the nrdAB promoter region was used to identify the presence of IS elements in nrdABp.

Table 2.

Oligonucleotides used in this work

Name	Sequence
dnaA-1	5′ TTGTTCGAGTGGAGTCCGCC
dnaA-2	5′ TCAAACACACGTTTGATAAC
dnaA-3	5′ TCGACGATATTCAGTTTTTTG
dnaA-4	5′ GTGAGCTGGAAGGGGCGCTG
dnaA-5	5′ CTACGGTAAATTTCATAGGT
dnaN-1	5′ AACATTGTCATCGTAAACCT
dnaN-2	5′ CGATGAAGCGTCTGATTGAAG
dnaN-3	5′ TGCTCGCGCGGCGATTCTCTC
dnaN-4	5′ CAACAAGCGGGTGAGGGACA
nrdAB-Fwd	5′ GCGTCATAATTCAAGTTAAT
nrdAB-Rev	5′ TTCTGATTCATGTATGTCGT
Grx3-Fwd	5′ ATGGCCAATGTTGAAATCTA
Grx3-Rev	5′ TCAATGATGATGATGATGATG

Analysis of RNR levels.

Protein samples were harvested using trichloroacetic acid (TCA) precipitation and were separated on a 10% SDS-PAGE gel before being transferred to a polyvinylidene fluoride (PVDF) membrane. Levels of RNR (NrdB) were measured by Western blotting with anti-NrdB antibody, using the Amersham ECL Plus Western blotting detection system (19).

nrdABp-lacZ fusions and assays.

nrdABp-lacZ fusions were constructed using the method of Whipple (30). Briefly, the desired nrdAB promoters were cloned into pFW11 using an EcoRI blunt/KpnI digest, and the resulting constructs were integrated into the F′ episome as described previously (30). First, the nrdAB promoter region was amplified by PCR from DHB4 (wild-type), RO36 (IS5), and NK21 (IS1) using the PnrdAB1 and PnrdAB2-KpnI primers. The PCR products were digested with KpnI and ligated into the plasmid pFW11, which had been cut with EcoRI, blunted, and cut with KpnI. The resulting constructs were verified by sequencing and transformed into strain CSH100 in order to allow the plasmid to recombine with the F factor. The constructs were then transferred into strain FW102 by conjugation (selecting for Tet^r Str^r exconjugants). The constructs were then mated into MAF715, MAF716, and NK356 by selecting for Tet^r Cm^r exconjugants, or mated into RO36, MAF174 and MAF177 by selecting for Tet^r Amp^r exconjugants. β-Galactosidase activity was measured qualitatively (on solid media containing 0.07 mg/ml X-Gal [5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside]) or quantitatively in liquid culture (16).

RESULTS

Construction of a trxA trxC grxA nrdH grxC(M43V) mutant strain and suppressor analysis.

In order to select for mutations that would improve the ability of glutaredoxin 3 to reduce RNR over that seen with the grxC(M43V) mutation, we constructed a strain (NK21) that lacked trxA, trxC, grxA, and nrdH and also had a chromosomal copy of the grxC(M43V) mutant allele. (The nrdH gene encodes a protein that can reduce RNR, but which is not expressed normally under aerobic conditions. The nrdH null mutation was included in the strain to avoid suppressor mutations that activate nrdH gene expression [19].) While grxC(M43V) suppresses the growth defect of the trxA trxC grxA nrdH strain when it is overexpressed from a plasmid, NK21 did not grow unless arabinose was added to induce expression of trxC from the complementing plasmid (Table 3). This finding indicates that Grx3(M43V), when expressed at native levels, is not sufficient for growth.

Table 3.

Growth of trxA, trxC, grxA, and nrdH mutant strains^a

Strain	Genotype	Complementing plasmid(s)	Growth on medium:
			+ara	+glu
FA173	trxA trxC nrdH	pBAD18-trxC	+	+
RO36	trxA trxC grxA nrdH	pBAD18-trxC	+	0
NK21	trxA trxC grxA nrdH grxC(M43V)	pBAD18-trxC	+	0
MAF747	trxA trxC nrdH	pBAD18-trxC, pBAD33-trxA	+	+
MAF750	trxA trxC grxA nrdH	pBAD18-trxC, pBAD33-trxA	+	0

^aThe indicated strains were grown on medium with arabinose (+ara) to induce expression from the complementing plasmid(s) or with glucose (+glu) to repress expression from the complementing plasmid(s). + indicates that a strain was able to grow, and 0 indicates no growth.

We carried out selections for suppressor mutants that would restore growth to NK21 and to the original strain used for suppressor selections, RO36. Out of seven suppressor strains isolated, none had mutations in grxC, as determined by PCR amplification and sequencing of the grxC gene. Instead, all of the suppressor strains had increased levels of NrdAB, as assessed by Western blotting (Fig. 2A). Therefore, we looked in these strains for mutations in chromosomal loci that were candidates for ones involved in NrdAB regulation.

Fig 2.

Effect of trxA trxC grxA nrdH suppressor mutations on the expression of ribonucleotide reductase. (A) Western blot showing the expression of one of the subunits of RNR (NrdB) in suppressor strains (left) and wild-type strains carrying the suppressor mutations (right). Equal amounts of protein were loaded in each lane. The α-subunit of RNA polymerase (RNAPα) was used as a loading control. α-NrdB, anti-NrdB; α-RNAPα, anti-RNAPα. (B) Western blot showing the expression of RNR (NrdB) in FA173 (trxA trxC nrdH/pBAD18-trxC), RO36 (FA173 grxA), NK21 [FA173 grxA grxC(M43V)], RO36 dnaA(A345G), and RO36 hda(F85V). Arabinose was added to induce expression of trxC from the complementing plasmid where indicated (+), or glucose was added to repress trxC (−). Equal amounts of protein were loaded in each lane, and anti-RNA polymerase α subunit was used as a loading control. (C) β-Galactosidase assays measuring expression from the nrdABp (wild-type) reporter construct in RO36, RO36 dnaA(A345G), and RO36 hda(F85V) with repression (in dark gray) and induction (in light gray) of trxC from the complementing plasmid. Error bars represent standard deviations based on cultures grown in triplicate.

Sequencing of dnaA and dnaN in the suppressor strains revealed mutations in these genes that had not been obtained previously with RO36 and NK21 (Table 4). We also sequenced the gene hda in the suppressor strains since it encodes the protein which mediates the interaction between DnaA and DnaN (13, 27). Suppressor strain MAF177 contained an altered residue in Hda changing a phenylalanine to a valine (F85V).

Table 4.

Mutations that suppress the trxA trxC grxA nrdH strains^a

Strain	Gene in which mutation was found	Amino acid change
RO36 suppressors
MAF173	dnaA	A345G
MAF174	dnaA	S203T
MAF175	dnaA	N346D
MAF176	dnaA	S331P
MAF177	hda	F85V
RO34	dnaA	A345S
RO52	dnaA	T174P
RO51	dnaN	G157C
NK21 suppressors
NK111	dnaN	N156S
NK113	dnaA	A362P

^aSuppressor mutations in RO36 (trxA trxC grxA nrdH) and NK21 [trxA trxC grxA nrdH grxC(M43V)] were identified as described in Materials and Methods. The mutations in RO34, RO52, and RO51 were identified previously by Ortenberg et al. (19).

To confirm that the hda mutation suppressed the growth defect of the trxA trxC grxA nrdH strain, we used a linked Tn10 marker to P1 transduce a wild-type hda allele into strain MAF177, which carried a complementing copy of trxC. We performed the transduction in the presence of arabinose, allowing for expression of trxC from pBAD18-trxC and allowing all transductants to grow (whether they had the suppressor mutation or not). The transductants that received the wild-type hda⁺ allele were unable to grow on media containing glucose (when the complementing trxC was repressed), while transductants that had retained the hda(F85V) allele did grow in the absence of trxC expression. Additionally, the transduction of hda(F85V)…Tn10 into the parental strain, RO36, yielded the same results.

We asked whether the mutations in hda, dnaA, and dnaN led to the upregulation of RNR as had the previously characterized dnaA and dnaN suppressor mutations. When the mutations were transduced into an otherwise wild-type strain, the mutations in dnaA and dnaN caused increased expression of RNR expression, as assessed by the levels of one of its subunits, NrdB (Fig. 2). However, the hda suppressor mutation in the wild-type background did not detectably increase NrdB expression (Fig. 2 and 3). We checked this result by assessing the activity of the nrdAB promoter again in the wild-type background with and without the hda mutation, this time using an nrdAB-lacZ fusion carried on an F′ plasmid. The β-galactosidase activities confirmed the Western blot results; the hda mutation did not increase the β-galactosidase activity in a wild-type background.

Fig 3.

Insertion sequences in the nrdAB promoter region lead to increased expression of the operon. (A) Expression of a lacZ reporter from nrdABp on solid medium containing bromo-chloro-indolyl-galactopyranoside (X-Gal) as an indicator. The reporter construct is expressed from an F′ episome in wild-type cells (MAF715). The lacZ mutant strain (DHB4 F⁻) serves as a negative control and the trcp-lacZ construct as a positive control. The nrdABp(IS1)-lacZ strain is present in duplicate. (B) β-Galactosidase assays measuring expression from the nrdABp reporter constructs: nrdABp (wild type [WT]) is green, nrdABp(IS1) is blue, and nrdABp (IS5) is orange. The reporter constructs were expressed in dnaA⁺ (MAF715), dnaA(S203T) (MAF716), and hda(F85V) (NK356) cells. Error bars represent standard deviations based on cultures grown in triplicate.

Since the hda mutation restores growth to RO36 under conditions in which the cells are depleted of reduced thioredoxins and glutaredoxin 1, we considered the possibility that the hda(F85V) mutation caused increased expression only under this stress condition. Therefore, we introduced the F′ plasmid carrying the nrdAB promoter fused to lacZ into RO36, RO36 dnaA(S203T), and RO36 hda(F85V) and assayed cultures with TrxC expressed and not expressed (excepting RO36 in the latter case, as this strain cannot grow without TrxC expression). We found that the hda mutation caused a 2-fold increase in expression of the nrdAB operon over that seen in RO36 when the trxC gene is fully expressed in the two strains, while the dnaA mutation caused a 3-fold increase. (Results described below indicate that full induction of TrxC from this plasmid is not sufficient to entirely replace the missing thioredoxin 1 and glutaredoxin 1.) Moreover, under conditions in which TrxC is depleted, the presence of either the hda or dnaA mutation causes a 10-fold increase in expression over that seen when TrxC is fully induced from the plasmid.

Insertion of IS1 and IS5 insertion elements in nrdABp in the quadruple mutant strains.

One class of suppressor mutations that we thought we might find were those that alter the promoter of nrdAB so as to give increased RNR levels. To test this possibility, we amplified by PCR the nrdAB promoter region from the RO36 and NK21 suppressor strains as well as that of DHB4, the starting strain that is wild type for all of the genes in the thiol-redox pathways. In all of the suppressor strains, the PCR products were larger in size than the PCR product which was amplified from DHB4. Sequencing revealed IS insertion elements in the nrdABp locus in these strains. Although we initially thought that the IS insertions were contributing to suppression, to our surprise, PCR and sequencing of the nrdABp locus from the NK21 and RO36 parental strains revealed that the IS inserted in the nrdAB promoter region was already present in these strains as well (Table 5). Nevertheless, to ask whether the IS element insertion in nrdABp, on its own, suppressed the growth defect of these two parent strains, we reexamined their growth properties. As we had found previously, neither strain could grow without the addition of arabinose to induce expression of trxC from the complementing plasmid (Table 3). The insertion sequences, by themselves, were not suppressor mutations.

Table 5.

Insertion sequences found in the nrdAB promoter region in E. coli (wild-type and mutant) and Shigella (wild-type) strains^a

Strain	IS found upstream of nrdAB	Mutations in glutaredoxin and thioredoxin genes
Shigella flexneri 5 strain 8401	IS2	None
Shigella boydii Sb227	IS1	None
Shigella sonnei ss046	IS2	None
Shigella flexneri 2a strain 301	IS2	None
Shigella flexneri 2a strain 2457T	IS2	None
Shigella dysenteriae Sd197	None	None
Shigella boydii BS512	IS1, IS2	None
E. coli MG1655 (wild type)	None	None
E. coli MG1655 (RO36)	IS5	trxA trxC grxA nrdH
E. coli MG1655 (NK21)	IS1	trxA trxC grxA nrdH grxC(M43V)

^aThe nrdAB promoter region was analyzed using MicrobesOnline (4). Glutaredoxin and thioredoxin homologs were identified using BLAST searches with the E. coli proteins against the indicated genomes.

The two parent strains from which suppressors had been obtained contained different IS sequences in the nrdAB promoter. Sequencing showed that NK21, as well as its derivative suppressor strains, had an IS1 element inserted 358 nucleotides upstream of the transcription start site, while RO36 and its derivative suppressor strains had an IS5 insertion element 161 nucleotides upstream of the transcription start site. However, the parent strain of both RO36 and NK21, FA173 (trxA trxC nrdH/pBAD18-trxC), had no insertion sequence in nrdABp. Therefore, at some point in the construction of RO36 and NK21, derivatives were obtained in which the IS elements had inserted into the nrdAB promoter region. It seemed possible in both cases that when the grxA::Kan allele was introduced into the strains, eliminating the RNR reducing protein Grx1, selection for this insertion occurred. The effects of the insertions presumably gave growth advantages to strains missing Grx1 at some stage of their construction, even though their growth was maintained by the expression of trxC.

Upregulation of RNR by IS elements in nrdABp.

In some instances, IS elements are known to activate transcription from nearby promoters: for example, expression of the cryptic bgl operon in E. coli can be activated by insertion of a number of different IS elements (including IS1 and IS5) (10). We asked whether the IS1 and IS5 elements that we identified in nrdABp affected expression of RNR and found that the levels of NrdB in both RO36 and NK21 were elevated in comparison to the parent strain, FA173 (Fig. 2B). While this result suggested that the IS insertions were causing upregulation of RNR, the mutational disruption of the thioredoxin and glutaredoxin pathways in RO36 and NK21 should induce an oxidative stress response that might lead to nrdAB upregulation (17).

To test further whether IS element insertion into nrdABp could on its own upregulate RNR, we compared the amounts of LacZ expression from the wild-type nrdAB promoter fused to lacZ and those from the nrdAB promoters from RO36 (IS5 insertion) and NK21 (IS1 insertion) in a wild-type strain with no defects in the reductive pathways. Notably, using the X-Gal indicator, we saw a clear increase in LacZ activity when the IS-containing reporter constructs were grown on solid media, the condition under which we may have inadvertently selected for the nrdABp mutations (Fig. 3A). When we quantified the increase in transcription by growing the strains in broth culture and assaying β-galactosidase activity, we observed that both of the IS element-containing promoters showed an approximately 2-fold increase in LacZ activity over that measured with the wild-type promoter (Fig. 3B).

We also tested whether the IS elements inserted into the nrdAB promoter region increased transcription in strains carrying a suppressor mutation in dnaA [dnaA(S203T)] or in hda [hda(F85V)]. The IS elements are located upstream of the DnaA boxes in the nrdAB promoter. In a wild-type strain carrying only the wild-type nrdABp reporter construct and the dnaA(S203T) mutation, there was an approximately 3-fold-higher level of β-galactosidase activity, compared to that seen in the dnaA⁺ strain. When the IS1 or IS5 element in the nrdAB promoter was introduced into this strain, the expression of LacZ was further increased (Fig. 3B). The results indicate that the IS elements act additively with the dnaA suppressor mutation to increase RNR expression.

Conservation of IS element insertion in the nrdAB promoter region.

Given that we had twice found different IS elements inserted upstream of the nrdAB promoter in our mutant strains, we wondered whether these insertions might reflect a general mechanism for upregulating RNR expression, comparable to that seen with the bgl operon mentioned above. If so, we reasoned that we might be able to find genomic evidence of IS insertion in the nrdAB promoter in some species. We therefore examined the region upstream of the nrdAB homologs in various species and noticed that several Shigella species had IS elements in the region upstream of nrdAB (Table 5), while others did not. These IS elements were in approximately the same position (300 to 400 nucleotides upstream) relative to the start of nrdA as those found in RO36 and NK21.

We sought, using genomic analysis, to identify differences between the Shigella species with and without the IS elements in the nrdAB promoter region that might explain the presence of these insertion elements. These differences might include defects in the glutaredoxin/glutathione and thioredoxin pathways (similar to the case in the mutant strains in which we isolated IS insertions). We also looked for the presence of homologs of the various known regulators of nrdAB transcription (Fis, Crp, IciA, and NrdR) (Table 5). We found no differences between the Shigella species and E. coli K-12 that would explain the acquisition of the IS elements in some Shigella species. There may be more subtle regulatory differences or environmental factors that affect the requirement for RNR in Shigella, or the insertions may have negligible effects on nrdAB expression in these bacteria.

Suppression of the trxA trxC grxA nrdH strain lacking the IS sequences in the nrdAB promoter.

Since we had twice constructed strains which lacked trxA, trxC, grxA, and nrdH by transducing a grxA deletion into a trxA trxC nrdH strain and obtained insertion elements in nrdABp, we wondered whether the IS element insertion, and its consequent upregulation of RNR, was necessary for the growth of the quadruple mutant strain. We again attempted to construct strains that lacked trxA, trxC, grxA, and nrdH. Transduction of the grxA::Km allele into FA173, a trxA trxC nrdH/pBAD18-trxC strain, expressing thioredoxin 2 from the plasmid (the same approach that was taken to construct RO36 and NK21) yielded two classes of transductants. One class of transductants had IS elements in nrdABp (IS1 or IS5, as in RO36 and NK21). The second class of transductants did not have IS insertions in that region. However, these colonies appeared sickly (forming colonies after 2 days of incubation at 37°C on rich media containing arabinose) and were difficult to maintain without selection for faster-growing variants taking place.

One explanation for the appearance of these IS insertions during the P1 transduction step in strain construction is that there was insufficient expression of TrxC from the complementing plasmid for good growth of the strain being constructed (trxA trxC grxA nrdH). Mutations (e.g., IS insertions) that would upregulate RNR might have been selected for, allowing expression from pBAD18-trxC to suffice for growth. Therefore, we repeated construction of the starting strain for suppressor selection, this time introducing a second complementing plasmid (pBAD33-trxA) into FA173 to provide more reductive power for RNR reduction, thus avoiding the selective pressure for increased RNR expression during the grxA transduction step.

The resulting strain obtained by this construction method, MAF750 (trxA trxC grxA nrdH/pBAD18-trxC, pBAD33-trxA), required arabinose for growth, as do RO36 and NK21, the quadruple mutant strains we had previously constructed (Table 3). However, in contrast to the strains which were made with only the pBAD18-trxC complementing plasmid, all of the MAF750 transductants grew well and did not contain the IS insertions in nrdABp.

With the quadruple mutant strain MAF750, we could now ask whether the suppressor mutations isolated in RO36 and NK21 suppressed the growth defect of this strain in the absence of IS sequence. We transduced three suppressor mutations (two dnaA mutations, S203T and Q346D, and the hda mutation F85V) into MAF750. All three suppressor mutations tested [including hda(F85V), which appears to be the weakest suppressor isolated] suppressed the growth defect of MAF750 (Table 6). None of the transductants had acquired IS elements in nrdABp. Therefore, the insertion sequences in nrdABp are not required for restoration of growth to the trxA trxC grxA nrdH mutant strain by these suppressor mutations.

Table 6.

Suppression of the trxA trxC grxA nrdH mutant strain does not require the presence of an IS insertion element in the nrdAB promoter region^a

Transductant	Growth on medium:
	+ara	+glu
MAF750 hda+...Tn10	+	0
MAF750 hda(F85V)...Tn10	+	+
MAF750 dnaA+...Tn10	+	0
MAF750 dnaA(S203T)...Tn10	+	+
MAF750 dnaA(Q346D)...Tn10	+	+

^aTransductants were tested for their ability to grow on media containing arabinose (+ara), which induces expression of the complementing trxA and trxC, and on media containing glucose (+glu), which represses expression from the complementing plasmids. + indicates that a strain was able to grow (formed large colonies in 1 to 2 days at 37°C) under the indicated condition, while 0 indicates that a strain was unable to grow. The nrdABp allele was checked by PCR, as described in Materials and Methods, and was wild type in each case.

DISCUSSION

Our selection for suppressors of an E. coli strain defective in two major reductive pathways has yielded mutations in genes for three different proteins that play roles in the initiation of DNA replication, DnaN, Hda, and DnaA. In the very construction of the defective suppressor strains themselves, we inadvertently generated conditions under which an additional class of mutations, insertion sequences in the nrdAB promoter, was selected for. All of these mutations resulted in an increase in expression of ribonucleotide reductase (RNR). RNR is an enzyme that must be tightly controlled because of its key role in the synthesis of dNTPs. Either over- or underproduction of NTPs can have deleterious effects. Neither the mechanisms which underlie the cell cycle control of RNR nor the mechanisms which balance the levels of RNR with levels of the proteins which reduce it are completely understood (11, 17, 28).

The characterization of suppressor mutants we had isolated previously gave new insights into RNR regulation. Studies of the mutant proteins suggested that DnaA was a repressor of the nrdAB operon and that its repressor activity depended on the degree to which it was bound to ATP (8, 19). Here we isolated additional mutations in dnaA and dnaN. The new mutation in dnaN is adjacent to the previously isolated mutation (N156S and G157C, respectively) and may alter the activity of DnaN in a similar fashion. The dnaN(G157C) mutation appears to exert its effects both by affecting the ability of DnaA to bind ATP and by inducing the SOS response, which leads to nrdAB upregulation (12). We also isolated two new mutations in dnaA (S203T and A362P). These mutations, like the previously isolated suppressor mutations, are in the ATP-binding domain of DnaA; however, they are in different regions of this domain (Fig. 4), and it remains to be seen what the effect of these mutations are on ATP binding and hydrolysis or perhaps on DNA binding.

Fig 4.

Structure of DnaA showing the residues found mutated in trxA trxC grxA nrdH grxC(M43V) suppressor strains. The structure of the ATP-binding domain of DnaA from Thermotoga maritima (PDB accession no. 2Z4S) is shown, with the residues found mutated in the suppressor strains indicated in pale gray. The protein is in the ADP-bound form.

The mechanism by which the hda(F85V) mutation suppresses the growth defect of the trxA trxC grxA nrdH mutant strain has not been determined. The upregulation of RNR by the hda(F85V) mutation is only manifested when the thiol redox pathways provide less than optimal reductive power to RNR. Since the Hda protein mediates the interaction between DnaN and DnaA, which leads to DnaA-bound ATP hydrolysis and RNR induction, it seems reasonable to conclude that the hda(F85V) mutation influences that interaction and results in increased ATP hydrolysis (8, 27). This ATP hydrolysis may not be sufficient to increase RNR expression in a wild-type background. Rather, it appears that there is a synergistic effect between the hda mutation and the upregulation induced by the diminished reductive capacity of the thioredoxin/glutaredoxin pathways in the mutant strain background (even when TrxC is expressed). Since there are multiple factors influencing RNR expression, including self-regulation by RNR, it is difficult to speculate on how this synergism may work (20).

While we have favored the explanation for the hda mutation's effect on DnaA as being the reason for its acting as a suppressor, the mutant Hda protein may be influencing nrdAB expression in some other way. If this is the case, it would seem that each of three components of the replication initiation complex altered in our suppressor strains may be affecting different regulatory mechanisms (12).

It seems likely that our suppressor selection will continue to yield new mutations in dnaA and dnaN as we have not yet isolated the same mutation twice. This is the case even though one residue, A345, was found mutated in two suppressor strains but to two different amino acids (A345S in RO34 and A345G in MAF173). Therefore, continued use of this selection procedure could provide a valuable collection of mutations affecting these essential proteins as well as the protein Hda. They could be used to further elucidate the regulatory activity of DnaA, including its binding to regulatory sites, as well as the interactions between DnaN, Hda and DnaA.

Our results suggest possible additional layers of complexity to the regulation of RNR. We find that the insertion of IS1 or IS5 elements into the promoter region of nrdAB leads to the upregulation of RNR, based on the following lines of evidence. (i) The RO36 and NK21 strains in which we isolated the insertion sequences (IS5 and IS1, respectively) show elevated levels of RNR (Fig. 2). (ii) We observe higher levels of activity of a β-galactosidase reporter gene cloned under the control of the nrdAB promoter containing either IS element, relative to the wild-type nrdAB promoter alone (Fig. 3).

The discovery of activation by insertion elements of the bgl operon of E. coli, which encodes a β-glucosidase, has been suggested to represent an evolutionarily adaptive mechanism for bacteria similar to phase variation (14, 21, 24). Similarly, activation of the flagellar regulator FlhD by IS element insertion has been proposed as an adaptive mechanism for regulation of motility (2). The fact that the insertions we detected in the nrdAB promoter appeared repeatedly and frequently during our strain constructions raises the possibility that they also represent a means for members of the bacterial population to survive better under conditions in which the activity of RNR is compromised. Such a problem for the bacteria may arise when there is an absence of sufficient reducing power—for example, via depletion of NADPH. A second suggestion that this phenomenon may have some significance comes from our finding that insertion elements upstream of the nrdAB promoter are present in several Shigella species.