Sustaining N₂-Dependent Growth in the Presence of CO

Abstract

Low levels of carbon monoxide inhibit the N₂-dependent growth of Rhodospirillum rubrum unless the ∼100-residue CowN protein is expressed. Expression requires the CO-responsive regulator RcoM and is maximal in cells grown in the presence of CO and a poor nitrogen source, consistent with the role of CowN in N₂ fixation.

The uniquely prokaryotic process of nitrogen fixation depends upon a highly regulated and elaborate enzyme complex to effect the energy-intensive reduction of N₂ to NH₃. Aside from the nifHDK structural genes that encode the dinitrogenase catalytic complex, additional factors execute the assembly of its unique metal centers and enable protein maturation and stabilization, while other nif-coregulated gene products remain functionally undefined (6, 19, 20). As this elaborate synthesis and function are inhibited by the prevalent environmental gases O₂ and CO (4, 7), it is reasonable that some accessory factors might prevent enzyme damage or effect repairs should it occur. Indeed, the “Shethna” protein has been shown to associate with dinitrogenase, forming an inactive but O₂-tolerant complex (5). No factors are known to specifically protect against inactivation by CO.

We noted that some genomes bear a conserved open reading frame (cORF) immediately downstream of genes encoding RcoM, a newly discovered CO-responsive regulator (8). Here we report detection of this cORF in a configuration similar to that of genes for another CO-responsive regulator (CooA) or within nif operons, and we show that in Rhodospirillum rubrum the encoded protein, hereby designated CowN (CO Weal-Nitrogenase), protects this organism's ability to fix N₂ in the presence of CO.

Database searches employing the PSI-Search function (SSEARCH ver. 35.04 [http://www.ebi.ac.uk/] [16, 22]) with CowN from R. rubrum as a probe indicated 33 similar proteins, averaging 98 residues in length (current as of 30 March 2010; see Table SA1 in the supplemental material). A representative alignment of CowN proteins from diverse organisms is presented in Fig. 1, and the complete alignment is shown in a supplement (see Fig. SA1 in the supplemental material). All organisms bearing cowN also contained nifHDK genes and are known or presumed to be capable of N₂ fixation. In nine instances cowN was preceded by cooA and a probable CooA binding site (tGTCg-[6n]-tGACa; lowercase letters designate less-conserved residues), in six instances it was preceded by rcoM and a probable RcoM-binding sequence (a triple direct repeat of cngTTnnnG-[11-12n]), and in four instances cowN appeared to be transcriptionally linked to possible nif genes. In 14 other instances cowN is genomically isolated, unlinked to nif genes or nearby genes encoding DNA-binding regulators; in fact, most of these organisms appear to lack the genes for CooA and RcoM (Table SA1). We hypothesized that N₂ fixation in the presence of CO would depend upon cowN expression because of (i) the genomic concurrence of cowN and nif genes, (ii) the indication of occasional transcriptional linkage of nif and cowN genes, (iii) the frequent juxtaposition of genes encoding the CO-responsive regulators RcoM or CooA with cowN, and (iv) the known inhibition of nitrogenase by CO.

FIG. 1.

Conservation of CowN residues (representative subset). This alignment of CowN proteins from alphaproteobacterial, betaproteobacterial, and deltaproteobacterial representatives (R. rubrum, Desulfuromonas acetoxidans DSM 684, and Dechloromonas aromatica RCB, respectively) is based upon a full alignment of 33 proteins (see Fig. SA1 in the supplemental material) generated by the ClustalW2 facility (Clustal version 2.0.12 [http://www.ebi.ac.uk/Tools/clustalw2/] [13]) using the Gonnet 250 matrix. Shading (Boxshade 3.21 [http://www.ch.embnet.org/software/BOX_form.html]) is based upon the full alignment with a 0.7 consensus.

To test this hypothesis, we evaluated the growth of R. rubrum in a defined medium containing (per liter of purified water): 8.4 g morpholinopropanesulfonic acid (adjusted to pH 6.8), 0.1 mg biotin, 0.2 mg p-aminobenzoic acid, 250 mg MgSO₄·7H₂O, 100 mg CaCl₂·2H₂O, 200 μM EDTA, 50 μM H₃BO₃, 10 μM FeSO₄·7H₂O, 2 μM Na₂MoO₄·2H₂O, 1 μM CoCl₂·6H₂O, and 2 μM MnCl₂·4H₂O. Note that this formulation omits nickel to diminish CO utilization by the Ni-dependent coo system (10). Sterile, pH-adjusted (7.0) additions of potassium phosphate (to 10 mM) and dl-sodium malate (to 25 mM) were added to the autoclaved medium, along with a source of nitrogen: addition of NH₄Cl (to 15 mM) or N₂ present in the GasPak jar (O₂ was removed using GasPak Plus envelopes; Becton Dickinson and Company). Plate media were solidified with 1.3% agar, and the growth at ∼28°C of four R. rubrum strains was compared: the “wild-type” (WT) strain UR2 (10), strain UR616 (ΔcooA::aacC1Ω), and newly created strains bearing the chromosomal alterations rcoM::aacC1Ωlinker (designated UR2216) and cowN::aacC1Ω (UR2546). These strains were constructed by typical methods (10) and verified by amplification of the altered chromosomal region, agarose gel analyses, and sequencing of the PCR products (data not shown).

In the presence of ammonium all strains grew equivalently in aerobic and anaerobic (photoheterotrophic [21]) growth modes, and the addition of CO was without effect (data not shown and Fig. 2 A). Strains also grew comparably in photoheterotrophic conditions in the absence of CO when N₂ served as the sole nitrogen source (Fig. 2B) but unequally if a low level of CO (0.15% initial level) was present (Fig. 2C): growth of the WT (UR2) and cooA (UR616) strains was superior to growth of the strains in which rcoM (UR2216) or cowN (UR2546) had been insertionally inactivated. In contrast, the WT, rcoM, and cowN strains grew similarly when cultivated under non-nitrogen-fixation conditions suitable for anaerobic, CO-dependent catabolism (9), unlike the cooA strain (UR616), which failed to grow (Fig. 2D). These results were consistent with a model wherein the cowN and rcoM gene products are not necessary for CO-dependent growth but are involved in protecting N₂ utilization in the presence of low levels of CO.

FIG. 2.

CowN is required for N₂ fixation in the presence of CO but not for CO-dependent growth. R. rubrum strains UR2 (WT), UR616 (ΔcooA::aacC1Ω), UR2216 (rcoM::aacC1Ωlinker), and UR2546 (cowN::aacC1Ω) were cultivated photosynthetically (PS, plates A to C) on medium supplemented with NH₄Cl in the presence of 0.15% CO (initial level) (A) and on medium requiring N₂ fixation without (B) or with (C) 0.15% CO. Plate D was incubated in the dark under conditions suitable for CO-dependent growth (9).

This protection was confirmed in liquid cultures of R. rubrum growing on N₂. Using the medium formulation specified above (20 ml per 120-ml vial), supplemented with phosphate and malate and supplied with an O₂-free N₂ headspace, photosynthetic cultures exhibited N₂-dependent growth with a doubling time of 8.5 h. Addition of CO to 1% to actively growing cultures of WT strain UR2 induced an ∼5-hour growth lag, followed by resumption of normal growth, while the same treatment of UR2546 (cowN::aacC1Ω) inhibited growth for the duration of the experiment (Fig. 3). The growth inhibition appears to be bacteriostatic: viable counts from samples obtained prior to CO addition and at the 12-hour time-point (Fig. 3) indicated consistent CFU numbers, averaging 3.9 × 10⁸ and 4.0 × 10⁸ CFU/ml/optical density at 680 nm (OD₆₈₀), regardless of strain or growth condition (99% confidence intervals amounted to less than 10% of these numbers). Samples obtained at 24 h showed similar numbers in actively growing cultures (averaging 4.2 × 10⁸ CFU/ml/OD₆₈₀) but a 33% decrease in inhibited cultures, either UR2 maintained under argon or UR2546 cultivated with CO (2.9 × 10⁸ and 2.8 × 10⁸ CFU/ml/OD₆₈₀, respectively). Thus, this decreased viability is not due to a toxic effect of CO. Other results confirm that CO is nontoxic to photoheterotrophic cultures of R. rubrum, regardless of the expression of the coo system, as seen in Fig. 2A and reported previously (11). The bacteriostatic effect of CO and the kinetics of the WT CO response suggest that CowN must be synthesized and then protect nitrogenase from CO. An alternative hypothesis—CowN-dependent CO removal—would not explain the differential growth of strains seen on plates (Fig. 2C).

FIG. 3.

Expression of cowN sustains N₂-dependent growth in the presence of CO. Liquid N₂-dependent cultures of R. rubrum strains UR2 (WT, ○ and •) and UR2546 (cowN::aacC1Ω, □ and ▪), inoculated from N₂-dependent cultures to an OD₆₈₀ of 0.06, were grown to an OD₆₈₀ of approximately 0.2, here designated the zero-hour time point. At approximately 2 h (↑), selected cultures received additions of CO to 1% (• and ▪) or the vial headspace was exchanged with argon (UR2 only, dashed line). Each plot represents the average of results for three cultures. Samples for the determination of CFU were obtained at approximately 1, 12, and 24 h (see text).

Additional R. rubrum strains were created to evaluate the requirement of RcoM and CO for cowN expression in cells supplied with good or poor nitrogen sources (NH₄Cl or glutamate, respectively; NH₄Cl prevents nif expression, while glutamate is an adequate nitrogen source yet allows nif derepression). We inserted a promoterless lacZ/Km^r cassette (12) to create two cowN::lacZ transcriptional fusion strains (Fig. 4): one with the WT rcoM (strain designated UR2504) and a version with a +1-base frameshift in rcoM that created TGA-TAA stops at codons 6 and 7 (UR2630). These constructs were verified by PCR amplification of the altered chromosomal region and sequencing of the products. Liquid cultures were grown photoheterotrophically with mixing at 30°C in the medium specified above but under an argon headspace and with additions of filter-sterilized and pH-adjusted NH₄Cl (to 10 mM) or glutamate (to 25 mM) as nitrogen sources. CO additions to the headspace (4% initial level) were added to cultures at an OD₆₆₀ of ∼0.1, and log-phase cell samples were obtained at an OD₆₆₀ of ∼1.0 for analysis of β-galactosidase according to standard methods (17). The higher CO level used in this experiment relative to the previous experiments reflects the different (non-N₂-dependent) growth condition and was intended to preclude the possibility that CO utilization could perturb the level of expression, as confirmed by similar results obtained with additions of 2% or 8% CO (data not shown).

FIG. 4.

Expression of cowN is CO and RcoM dependent and is enhanced in the presence of glutamate, a nitrogen source that allows nif derepression. R. rubrum strains bearing a cowN::lacZ reporter were cultivated under photosynthetic growth conditions in medium containing NH₄Cl or glutamate (Glu), plus or minus added CO (4% initial level). Strain UR2504 bears a WT rcoM gene, while in strain UR2630 rcoM expression is inactivated by a single-base frameshift at codon 6 that creates dual translational stops at codons 6 and 7. Multiple mid-log-phase cultures were assayed for β-galactosidase, and activities in Miller units (MU) are presented with error bars indicating the 99% confidence interval.

UR2504 cultures (rcoM⁺) grown in the presence of CO showed increased cowN::lacZ expression relative to cultures grown without CO additions, and the level of expression was greatly enhanced when cells were supplied with glutamate instead of NH₄Cl (Fig. 4). In contrast, only slight cowN::lacZ expression was observed in the frameshifted rcoM strain (UR2630), regardless of the nitrogen source or CO addition, consistent with RcoM-dependent promotion of cowN expression (Fig. 4). A strain (UR2599) bearing an in-frame 47-codon deletion of the rcoM region that encodes the C-terminal DNA-binding domain likewise failed to induce cowN expression, again consistent with RcoM-dependent cowN expression rather than a polar (e.g., Rho-dependent) transcription effect (data not shown).

The apparently “CO-independent” expression of cowN::lacZ in strain UR2504 grown in the presence of glutamate suggests the involvement of an additional regulatory mechanism, as does the effect of the nitrogen source on the magnitude of CO-induced expression (Fig. 4). However, at least the “CO-independent” cowN::lacZ expression could reflect artifactual endogenous synthesis of CO. Indeed, similar growth conditions (malate as a carbon source, glutamate as a nitrogen source, without added CO) were successfully used to cultivate R. rubrum for the isolation of carbon monoxide dehydrogenase, the accumulation of which requires CO (1, 2, 3, 21). The dependence of maximal cowN expression on (i) a nitrogen source that permits nif derepression, (ii) CO (notwithstanding the elevated “minus-CO” baseline for glutamate-grown cultures), and (iii) the presence of RcoM is in accord with CowN protection of nitrogenase function when CO is present (Fig. 2 and 3). The basis of the nitrogen source effect on CO-dependent cowN expression is unclear, as binding sites for known nif gene transcriptional regulators are not apparent 5′ of cowN.

The genomic linkage of cooA with cowN in other organisms (see Table SA1 in the supplemental material) is likewise consistent with a CO-dependent regulation of CowN. For example, aside from R. rubrum, wherein RcoM is necessary for cowN expression and CooA regulates expression of the catabolic CO-oxidation (coo) system (18), only the sequenced genome of a second photosynthetic bacterium (Rhodopseudomonas palustris BisB18) contains both cowN and coo systems. This organism does not harbor an rcoM gene, and the expression of coo and cowN genes therefore appears to be controlled by dual CooA proteins encoded adjacent to the coo regulon (YP_534336) and cowN (YP_533050). Presumably, any differential regulation of the coo and cowN systems relies on different affinities of the regulators for the CO effector or distinctions in target DNA binding. In addition, the biological roles of coo and cowN systems may reflect nickel availability, for Ni additions are necessary for maximal coo-dependent growth (9, 10) while irrelevant for cowN-dependent protection of nitrogenase.

CowN joins a list of several accessory functions known or suspected to promote nitrogenase assembly and function (6, 19, 20), specifically by preserving the activity in the presence of CO, a prevalent environmental contaminant. In vitro studies will be required to fully elucidate the mechanism of this activity, though a stoichiometric interaction with nitrogenase, not unlike that of the “Shethna” protein (5), seems plausible. That cowN is not evident in all organisms possessing a nif system may reflect several scenarios: (i) organisms could possess nitrogenase enzymes with inherently different resistances to CO—indeed, a recent report indicates that the vanadium-containing nitrogenase from Azotobacter vinelandii can reduce CO (14); (ii) organisms could experience different degrees of CO exposure or possess efficient mechanisms for its elimination; or (iii) perhaps a CowN-like function is performed by another uncharacterized accessory protein. Besides demonstrating its novel function, the identification of CowN may promote biological hydrogen production processes inasmuch as these rely on nitrogenase (15) and CO inhibits all nitrogenase-catalyzed substrate reductions except for the conversion of 2H⁺ to H₂. Thus, inactivation of cowN could increase the degree of CO-bound nitrogenase and thereby focus its catalytic specificity toward H₂ synthesis.

Nucleotide sequence accession numbers.

The R. rubrum CowN protein sequence is available in databases (NCBI accession number, YP_428597 [http://www.ncbi.nlm.nih.gov/]; UniProtKB accession number, Q2RNI5 [http://www.uniprot.org/]).