Expression analysis of the nrdHIEF operon from Escherichia coli. Conditions that trigger the transcript level in vivo.

Escherichia coli has two aerobic ribonucleotide reductases encoded by the nrdAB and nrdHIEF operons. While NrdAB is active during aerobiosis, NrdEF is considered a cryptic enzyme with no obvious function. Here, we present evidence that nrdHIEF expression might be important under certain circumstances. Basal transcript levels were dramatically enhanced (25-75-fold), depending on the growth-phase and the growth-medium composition. Likewise, a large increase of >100-fold in nrdHIEF mRNA was observed in bacteria lacking Trx1 and Grx1, the two main NrdAB reductants. Moreover, nrdHIEF expression was triggered in response to oxidative stress, particularly in mutants missing hydroperoxidase I and alkyl-hydroperoxide reductase activities (69.7-fold) and in cells treated with oxidants (up to 23.4-fold over the enhanced transcript level possessed by cells grown on minimal medium). The mechanism(s) that triggers nrdHIEF expression remains unknown, but our findings exclude putative global regulators like RpoS, Fis, cAMP, OxyR, SoxR/S, or RecA. What we have learned about nrdHIEF expression indicates strong differences between its regulation and that of the nrdAB operon and of genes coding for components of both thioredoxin/glutaredoxin pathways. We propose that E. coli might optimize the responses to different stimuli by co-evolving the expression levels for its multiple reductases and electron donors.


INTRODUCTION
Ribonucleotide reductases (RRases) 1 provide the building blocks for DNA biosynthesis in all living organisms. RRases are grouped into three major classes (class I, II and III) based on the mechanisms they use for radical generation and on their structural differences. Class I is subdivided further into two subclasses (Ia and Ib) based mainly in allosteric regulation but also on involvement of auxiliary proteins (recently reviewed in 1). Escherichia coli has the coding potential for three different RRases. The NrdAB (class Ia) is active during aerobiosis, NrdDG (class III) is strictly anaerobic and NrdEF (class Ib) is thought to be a cryptic enzyme with no obvious function (1).
Class I enzymes receive the electrons required for the reduction of ribose from small proteins, thioredoxins (Trxs) and glutaredoxins (Grxs), with two redox-active cysteine thiols, which by dithiol-disulfide interchange reduce an acceptor disulfide in the active center of RRase. Trxs and Grxs are kept in a reduced form by NADPH, which reduces the redoxin either via the flavoprotein thioredoxin reductase or via the flavoprotein glutathione reductase and the ubiquitous tripeptide glutathione (GSH) (2). E. coli contains two thioredoxins (Trx1 and Trx2), three glutaredoxins (Grx1, Grx2 and Grx3) and a novel redoxin (called NrdH) with thioredoxinlike activity but glutaredoxin-like amino acid sequence (2)(3)(4)(5). Grx1 and Trx1 are the two main hydrogen donors of the E. coli NrdAB enzyme (3,4,6,7). NrdH is a more specific hydrogen donor for the NrdEF than for the NrdAB enzyme, whereas the opposite is the case for Grx1 (5).
The nrdA and nrdB genes that code for the NrdAB class Ia reductase constitute a tightly ________________________ 1 The abbreviations used are: RRase, ribonucleotide reductase; Trx, thioredoxin; Grx, glutaredoxin; GSH, regulated transcription unit that does not include the gene for either Trx or Grx. Expression of nrdAB genes is cell cycle regulated and increases when DNA synthesis is inhibited (1).
Regulation of nrdAB expression in E. coli has shown to be very complex: multiple cis-acting positive regulatory sites identified upstream of the nrdAB promoter appear to control, independently but in concert, the cell cycle-dependent transcription and the response to inhibition of DNA synthesis (9). Recent data from our group add further complexity to the nrdAB regulation, demonstrating a tight and inverse relation between expression of nrdAB and that of genes coding for components of both glutaredoxin and thioredoxin pathways (10).
Interestingly, we have observed that induction ratios of nrdAB transcription by hydroxyurea (RRase inhibitor) are similar to the increments in nrdAB basal expression of mutants lacking both Trx1 and Grx1. Therefore, we have postulated that operation of the NrdAB enzyme in the absence of its two main reductans must lead to disturbances in deoxyribonucleotide production sensed as those caused by hydroxyurea.
The nrdE and nrdF genes that code for the NrdEF class Ib reductase form a conserved operon where the promoter is followed by four genes (11). Genes nrdH, coding for the NrdHredoxin, and nrdI, coding for a protein with a stimulatory effect on ribonucleotide reduction, are present upstream of nrdEF (5). It has been reported that the E. coli transcription unit is not expressed in sufficient amounts to support growth under normal laboratory conditions. Neither inhibitors of DNA replication nor DNA damages induce its expression. Thus far, only hydroxyurea has been shown to stimulate the expression of nrdEF genes (11).
We have quantitated the in vivo transcription of genes encoding for components of both glutaredoxin and thioredoxin pathways by means of a novel multiplex reverse transcriptionpolymerase chain reaction (RT-PCR) approach (10,12). In this protocol, all target genes, a housekeeping gene (also named control gene, reference gene, or internal standard), and one
Treatments-E. coli cells from an overnight culture in LB broth were centrifuged and diluted 100-fold into 50 ml of M9 minimal medium and incubated at 37°C and 150 rpm to reach an OD 600 of 0.2. At this stage, the bacteria were further grown in the absence or the presence of hydroxyurea, hydrogen peroxide (H 2 O 2 ), paraquat (PQ), t-butyl hydroperoxide (tBOOH), 4nitroquinoline 1-oxide (4NQO) or N-methyl-N´-nitro-N-nitrosoguanidine (MNNG), for a fixed time period. The cells were then rapidly cooled to 0°C for total RNA purification. Table I) were designed with the Oligo 6.1.1/98 (Molecular Biology Insights Inc., Plymouth, MN) program, in order to obtain the highest specificity and performance in multiplexed PCR reactions. Target genes code for the NrdH-redoxin (nrdH), the NrdI stimulatory protein (nrdI), and the R1E (nrdE) and R2F (nrdF) subunits of E. coli class Ib RRase. As described previously (10,12), the gapA gene, which codes for D-glyceraldehyde-3phosphate dehydrogenase, was used as internal standard. The internal standard normalizes variations in RNA extraction, reverse transcription and PCR amplification among samples. To evaluate the extent of putative variations in the expression of the reference gene, an external standard was included, as described (10). The external standard was an in vitro synthesized RNA fragment encoded by the CYP1A gene from Liza aurata. This external standard has no homology with the E. coli genome.

Multiplex RT-PCR for In Vivo Quantification of nrdHIEF Transcription-RNA
purification and cDNA synthesis were as described (10). At least two independent RNA preparations were isolated for each experimental condition; each RNA sample being retrotranscribed on three separated occasions. PCR amplification of cDNA was carried out using all the primer pairs listed in Table I, or using just those for the nrdH and nrdE target genes, the gapA reference gene and the external standard. In this last case, the multiplex PCR amplification

RESULTS
Quantification of the nrdHIEF Operon Expression-In a prior study (11), expression of nrdEF genes was quantified by a competitive RT-PCR, in which the competitor had an internal deletion of 99 bp and the reaction products were resolved in ethidium bromide agarose gels and analyzed by densitometry. Several points at which errors can occur by using such a methodology include differences in amplification efficiencies because of the great size difference between the competitor and the target, possible heteroduplex formation, and the need to compensate differences in fragment label incorporation. When measuring rare species of mRNA, which requires high number of PCR cycles, the control of those problems are of maximal importance, even for relative quantification (e.g. see [22][23][24]. In this work, by taking advantage of the high sensitivity, accuracy and reproducibility of our multiplex RT-PCR procedure (13), we monitored variations in basal levels of nrdHIEF expression, and in response to several stress conditions. To achieve this objective, we first confirm that transcription from the nrdHIEF promoter normally takes place in wild-type cells, and increases upon hydroxyurea exposure (11).
UC5710 cells were, thus, exposed to increasing concentrations of hydroxyurea (varying from 1 to 30 mM) for 5 min. Thereafter, the expression of all the genes that constitute the nrdHIEF operon was examined by using the primer pair set in Table I. Basal levels of nrdH, nrdI, nrdE and nrdF gene expression were readily detected and increased upon hydroxyurea treatment (data not shown). As average, an induction level of 12.4 ± 2.5-fold (relative to untreated bacteria) was observed at the minimal dose assayed of 1 mM. It is worth noting that in the prior study by Jordan et al (11), such an induction level was observed under much more acute treatment conditions (bacteria grown in the presence of 50 mM hydroxyurea). Since we do not detect differences in the number of times that expression of each nrd gene was increased by

Regulation of E. coli NrdHIEF Expression 9
hydroxyurea, the multiplex PCR was simplified in further experiments by amplifying just two (nrdH and nrdE) of the four genes included in the nrdHIEF transcription unit (in addition to the internal and external standards).

Effects of Growth Conditions on Basal Levels of nrdHIEF ExpressionThe expression
profiles of nrdH and nrdE genes throughout the growth curve of wild-type UC5710 cells in rich LB medium is shown in Fig. 1. Maximal expression levels were observed at the initial stages of exponential growth (OD 600 d 0.2); then, gene expression decreased rapidly reaching 25-fold lower levels as the culture continued growing from mid-exponential to stationary phase (OD 600 e 0.4). Fis and RpoS (also named σ S or σ 38 ) are two regulatory proteins with acute growth phase-dependent expression in LB medium (25,26). To determine the hypothetical influence of Notable variations in gene expression were observed also with respect to the composition of the growth medium ( Fig. 1). Thus, bacteria growing on M9 minimal medium with glucose had from 2-fold (OD 600 = 0.2) to 75-fold (OD 600 = 0.7) higher levels of nrdH and nrdE transcripts than did bacteria growing on rich LB medium alone. Since carbon sources control intracellular cAMP levels (27), we examined next the effects on nrdH and nrdE expression, of addition of glucose to the LB nutrient medium (to 2 g/l) and of replacement of glucose by lactose as the sole carbon source in the M9 minimal medium. No dependence on carbon source was observed (data not shown).
As indicated above, fold variations were identical for both nrdH and nrdE genes; nrdH data

Regulation of E. coli NrdHIEF Expression 10
will henceforth be considered representative of the entire nrdHIEF transcription unit.

Effects of Deficiencies in Thioredoxin and Glutaredoxin/GSH Pathways or in Antioxidant
Enzymes on Basal Levels of nrdHIEF Expression-In addition to the classic function of acting as reductants for RRase, both the Trx and Grx/GSH pathways are required to maintain the low thiol-disulfide redox potential of the bacterial cytoplasm (20,28). On the other hand, catalase, alkyl hydroperoxide reductase and SOD activities maintain the steady-state concentrations of peroxides and superoxide beneath their respective toxicity thresholds (29). The influence of missing various components of these systems on basal levels of nrdHIEF expression is studied in Table II Since OxyR can be activated either by challenge with an oxidant such as H 2 O 2 or directly by a change in the cellular thiol-disulfide status caused by the inactivation of the two Trx and Grx/GSH reductive pathways (20,28), it seemed possible that the transcriptional induction of nrdHIEF that we report in Table II was due to activation of OxyR. This possibility was investigated by studying the effects of oxyR mutant alleles on basal levels of gene expression in UC1358. As shown in Table II, expression of nrdHIEF genes was modulated by mutations in the oxyR regulatory locus, but it followed a pattern opposite to that exhibited by most OxyRregulated genes (10,29). Therefore, the high basal level of nrdHIEF message in UC1358 (32.7fold) was further elevated (not diminished) in its ∆oxyR::kan null mutant derivative (UC1363), where a 54.4-fold increase in the amount of nrdHIEF transcript was quantified. On the contrary,

Regulation of E. coli NrdHIEF Expression 11
a decrease (not an increase) in the steady-state level of UC1358 was observed in the strain (UC1395) that carries the oxyR2 constitutive mutation. These results indicated that OxyR is not directly involved in the nrdHIEF up-regulation reported in Table II. The higher increment in basal level of nrdHIEF expression caused by the ∆oxyR::kan null allele might, thus, be attributed to the underexpression in UC1363 of OxyR-regulated genes involved in antioxidant defense, like those coding for catalase HPI and alkyl hydroperoxide reductase (10,18,29). A similar (while opposite) argumentation might explain the lower increment quantified in bacteria with the oxyR2 constitutive mutation. Likewise, the difference between UC827 and UC1358 with respect to nrdHIEF expression can be explained by differences in the expression levels of OxyR-regulated genes. The strain (UC1358) with higher levels of antioxidant defenses (10) displayed the lower increment in nrdHIEF/gapA ratio (Table II).

Gene Expression Induction by Hydrogen Peroxide and Superoxide-To gain further
information for the regulation of nrdHIEF expression under oxidative stress conditions, wildtype bacteria were exposed to increasing concentrations of H 2 O 2 or paraquat (a superoxidegenerating compound). Since OxyR and SoxR together with SoxS are key regulators of the adaptive responses to H 2 O 2 and superoxide radicals, respectively (recently reviewed in 29), strains with mutations that eliminate either OxyR (UC1342) or both SoxR and SoxS (UC1333) were used in conjunction with wild-type bacteria (UC5710). As shown in Fig. 2, induction of nrdHIEF mRNA was readily seen in response to both oxidants. Therefore, increments of 23.4fold and 4.3-fold were quantified shortly (5 min) upon exposure of wild-type cells to 100 µM H 2 O 2 or 500 µM PQ concentration, respectively. Inductions were preserved in the oxyR and soxR/S mutant strains, indicating that neither OxyR (in agreement with the results above) nor

Regulation of E. coli NrdHIEF Expression 12
SoxR/S are involved in the oxidative stress stimulation of nrdHIEF expression. In fact, as expected from data in Table II, Table II), we examined the effect of tBOOH treatments on expression of nrdHIEF operon (Fig. 3). In this experiment, we used a set of primers (13) that allows to compare the response of nrdHIEF genes with that of genes encoding for the main aerobic RRase (nrdAB operon), besides that of well-known components of the E. coli oxidative stress responses (like oxyS and grxA genes) (10). While the amount of nrdAB transcript remains basically unchanged, the nrdHIEF transcript level was readily induced by tBOOH. As shown in Fig. 3, the oxidative-stress responsive genes, oxyS and grxA, gave also a clear positive response to tBOOH treatments.
4NQO is a widely studied model mutagen and carcinogen, which derives its activity from the induction of both bulky adducts (for that is often referred to as UV radiation-like) and

Regulation of E. coli NrdHIEF Expression 13
oxidative damages in cellular DNA (33). As in the case of tBOOH, treatments with 4NQO elevated the nrdHIEF and the oxyS and grxA transcript levels, without affecting the nrdAB expression. Since 4NQO is known to trigger the SOS response (33), we study next if this compound increased nrdHIEF expression in the absence of RecA protein. No difference between RecA + and RecAcells was observed (data not shown), indicating that this effect is not SOS dependent.
In contrast to chemical oxidants, MNNG, which is a strong monofunctional alkylating agent that methylates cellular DNA resulting in multiple types of primary lesions (33), did not affect nrdHIEF expression in wild-type E. coli (data not shown).

DISCUSSION
Previous studies have reported that suppression by the nrdHIEF genes of the inviability of E. coli strains defective in either the NrdAB reductase or three of its reductants (Grx1, Trx1 and Trx2), requires a second gene copy placed in either the chromosome or a cloning plasmid (11,34). Based on these genetic evidences, it is commonly accepted that the nrdHIEF operon is underexpressed in bacteria grown at standard aerobic conditions, thus somewhat questioning its physiological significance. The main goal of this study was the accurate quantification by multiplex RT-PCR of variations in nrdHIEF transcript levels, in order to elucidate the growth conditions and stress circumstances, under which expression of nrdHIEF genes might become important to E. coli.
Results presented in this work confirm that transcription of the nrdHIEF operon normally takes place in E. coli cells grown in LB medium to mid exponential phase (11). Nevertheless, we present the first indication that this low basal level of nrdHIEF mRNA can be dramatically enhanced in wild-type bacteria as a function of the growth phase and the composition of the

Regulation of E. coli NrdHIEF Expression 14
growth medium: a pronounced increase (from 25-to 75-fold) in nrdHIEF transcript could be monitored at the initial stages of exponential growth in LB and in cells cultured in M9 minimal medium. Of note is the additional observation of no significant differences among the relative transcript levels for the genes of the nrdHIEF operon, indicating for the first time that the expression of the genes encoding the NrdEF ribonucleotide reductase are tightly co-regulated with that of genes encoding accessory proteins, like its specific NrdH-redoxin hydrogen donor.
We hypothesize that the strong growth phase-and growth medium-dependent regulation of nrdHIEF transcription might have a functional significance for wild-type cells. It is well-known that the physiology of a bacterial cell shifts between the phases of a culture and with the quality of the growth medium, and that many of these changes are realized at the level of gene expression (35). Therefore, while the rich medium contained preformed building blocks of macromolecule synthesis, in the minimal medium, the carbon backbone of the glucose molecule is rearranged through biosynthetic pathways to generate each of the building blocks de novo. The higher expression of nrdHIEF genes in minimal medium might thus be indicative of the need to generate the building blocks for DNA biosynthesis de novo from glucose. Accordingly, a recent single experiment that used DNA arrays of the entire set of E. coli genes to discover genomic expression patterns, has revealed that those genes with a pivotal role in central metabolism tend to be expressed at higher levels in minimal medium than in rich LB medium (36). Here, we verified this tendency for the expression of the nrdHIEF operon. Many genes having a growth phase-and growth medium-dependent regulation are under global regulatory mechanisms like those mediated by RpoS, Fis or the intracellular levels of cAMP (35,36). Our data indicate, however, that these global regulators are not responsible for the nrdHIEF expression pattern, thus making a difference with the nrdAB operon which is known to be under the positive regulation of Fis protein (37).

Regulation of E. coli NrdHIEF Expression 15
The striking up-regulation of the nrdHIEF operon in wild-type cells under certain growth conditions raises an intriguing question: why is the inactivation of the nrdAB operon lethal to E. coli cells in the presence of oxygen? The straight answer to this question is that this upregulation is conceivably insufficient to complement the lack of the first NrdAB RRase, or three of its reductants (Grx1, Trx1 and Trx2), unless a second extra copy of the nrdHIEF operon is placed on the bacterial chromosome. In this context, quantification at the protein level would be of maximal interest, as it has been postulated that translation of nrdE message might be low because the start codon is TTG (11).
We have been also able to show that basal level of nrdHIEF mRNA is dramatically increased (> 100-fold) in bacteria (UC827) simultaneously lacking Trx1 and Grx1, the two main reductants of the NrdAB reductase. We speculate that this enormous increment in nrdHIEF expression might be physiologically relevant for the viability of UC827 (38). This trxA grxA double mutant would maintain the balanced supply of deoxyribonucleotides required for DNA synthesis, by triggering the transcription of the operons (nrdAB and nrdHIEF) (12,this work) that code for both aerobic RRases and for the NrdH reductant. To this respect, its worth to note that (i) NrdH, the specific reductant of the NrdEF enzyme, is also a functional hydrogen donor for the NrdAB reductase (5), and (ii) although the other two known reductants for NrdAB remain in trxA grxA mutant cells, one (Grx3) is highly inefficient (4) and the other (Trx2) has a low level of expression under normal aerobic conditions (3,34).
Oxidative stress is an unavoidable by-product of aerobic life. Oxidative stress is caused by exposure to H 2 O 2 , superoxide anion and hydroxyl radical, which in turn damage proteins, nucleic acids, and cell membranes producing detrimental molecules like alkyl hydroperoxides (29,32). In this paper, we present evidences that nrdHIEF expression is triggered in E. coli when What might an enhanced nrdHIEF expression do in the E. coli oxidative stress response? An answer to this question could be to increase the free-radical scavenging capacity of cells by increasing the NrdH protein level. In this respect, it is worth noting that Trx is a highly efficient antioxidant with a role in protecting E. coli against oxidative stress (39)(40)(41); NrdH with a redox potential of -248.5 mV is as potent a reductant as Trx (5). Furthermore, an enhanced ribonucleotide reduction capacity should be advantageous under oxidative stress conditions since ROS escaping from antioxidant defenses, can inflict many oxidative damages on DNA (42).

Regulation of E. coli NrdHIEF Expression
The mechanism by which nrdHIEF expression is triggered under oxidative stress conditions remains elusive, but data reported indicate that the presence of ROS must be sensed by regulators that are distinct from both OxyR and SoxR/S. Contrarily to what we have learned in this work about nrdHIEF expression, the expression of nrdAB operon that codes for the main class I reductase was not induced by oxidative stress, in agreement with previous results (10).
Interestingly, however, genes that code for two (Grx1 and Trx2) out of the five known NrdAB reductants, together with those that code for the enzymes (glutathione reductase and thioredoxin

Regulation of E. coli NrdHIEF Expression 17
reductase) that regenerate their reduced forms, are part of the OxyR regulon. These findings suggest that E. coli might optimize the responses to different stress situations by co-evolving the expression levels for multiple RRases and reductants. In this respect, it is worth noting that while DNA-damaging agents that induce the SOS response produce in general an overexpression of nrdAB genes, these agents have no effect on nrdHIEF expression (11,this work).
In short, this report strongly suggests that nrdHIEF expression might be important under specific physiological circumstances. Findings presented here open numerous ways for future studies. One challenge will be the construction of a mutant lacking the entire nrdHIEF transcription unit in order to elucidate further compensations among the expression of both aerobic RRases and their reductants, under either normal or stressed conditions. The multiplex RT-PCR approach will be of relevance in these next coming experiments. Nevertheless, quantifications at the protein level will be also necessary in order to unravel the relationships between mRNA production and protein synthesis.