Carbon substrate re-orders relative growth of a bacterium using Mo-, V-, or Fe-nitrogenase for nitrogen fixation
Submitted to Editor Alyson Santoro ([email protected]) on 10/22/2020 and edited on 1/16/2021
Introduction
After publication of this work, it was brought to our attention that it has not been tested whether R. palustris CGA009 and its derivative strains can use dimethyl sulfoxide (DMSO) as an electron sink (Luxem et al., 2020). Here, we report the results of additional bioinformatic, physiological, and chemical experiments to test whether R. palustris CGA009 can use DMSO as an electron acceptor. The results are ambiguous. Bioinformatic analysis did not identify any proteins known to reduce DMSO, nor did the physiological experiments demonstrate growth with DMSO as the exclusive electron sink. Still, R. palustris CGA009 cultures do produce the product of DMSO respiration, dimethyl sulfide (DMS), exclusively in the presence of DMSO. Here, we discuss the impacts on our initial data interpretation as well as the implications for future researchers interested in using DMSO as an external electron sink for R. palustris CGA009.
Results
We began by searching for the presence of proteins with potential DMSO reduction capability in the R. palustris CGA009 genome. DMSO reduction activity has been demonstrated in several clades of the complex-iron–sulfur-molybdenum (CISM) protein superfamily, including the closely related DMSO, trimethylamine n-oxide (TMAO) and biotin sulfoxide (BSO) reductases that together form the DMSO reductase (DMSOR) protein family (Gon et al., 2000; McCrindle et al., 2005; Kappler and Schäfer, 2014; Kappler et al., 2019). These reductases vary in their selectivity for their substrates and in whether they are involved in anaerobic respiratory processes (Iobbi-Nivol et al., 1996; Pollock and Barber, 1997; Gon et al., 2000; McEwan et al., 2002; Ezraty et al., 2005; McCrindle et al., 2005; Kappler et al., 2019). No annotations to DMSO or TMAO reductases were found in the NCBI or KEGG databases for R. palustris CGA009 (accessed April 2020). Because of the broad substrate specificity of many enzymes in the DMSOR family, it is possible that there are enzymes capable of DMSO reduction whose annotations are less precise (Kappler et al., 2019). To identify such enzymes, we used BLAST (blastp) to search for homologs of known DMSO reductases from closely related strains in the R. palustris CGA009 genome. The query sequences included DorA from R. capsulatus 37b4 (Knäblein et al., 1996; Shaw et al., 1996), DorA and TorA from Rhodobacter spheroides 2.4.1 (Mouncey et al., 1997; Laguna, 2010) and Rhodopseudomonas palustris BisB18 (Oda et al., 2008), the other members of the DMSO reductase operons from R. palustris BisB18 and, for completeness, representative members from throughout the CISM superfamily compiled by (Leu et al., 2020) (Table S1). This search resulted in eleven possible hits (Table S1). To assess the likelihood that these hits are putative DMSO reductases, we made a phylogenetic tree for the CISM superfamily (Fig. 1). None of the hits from R. palustris CGA009 grouped with the Type II DMSOR family enzymes, which include the versatile sulfur- and nitrogen-oxide reductase DmsA. Four of the hits clustered as a sister group to known formate dehydrogenases while one formed an outgroup to the clade containing formate dehydrogenases and arsenate, arsenite, selenate and sulfur reductases. The remaining five hits formed a basal outgroup to all clusters within the superfamily, along with xanthine dehydrogenase (Xdha) from Paenibacillus mucilaginosus, which has not, to our knowledge, been observed to reduce DMSO. Only the BisC homologue RPA4653 (WP_011160185) clusters in a clade with proteins known to reduce DMSO, the Type III DMSOR family enzymes. The primary function of BisC, the catalytic subunit of the biotin sulfoxide (BSO) reductase, appears to be the production of biotin and methionine from their respective sulfoxides (Kappler et al., 2019). However, several representatives of the less well characterized BisC/TorZ/MtsZ subgroup of the Type III DMSOR have also been shown to reduce DMSO (Pollock and Barber, 1997; Kappler et al., 2019). It is possible that this could be the enzymatic pathway responsible for DMSO reduction in R. palustris CGA009. We did not measure whether R. palustris CGA009 synthesized RPA4653 in the presence of DMSO, though we note that it was not detected in the previous proteomic analysis during growth on succinate and acetate without DMSO. Given the inconclusive nature of this result, we performed additional physiological and chemical experiments.
Growth on the carbon substrate butyrate, which is more reduced than biomass, requires the presence of an electron sink like nitrate, DMSO, bicarbonate for carbon fixation or the absence of fixed nitrogen for nitrogen fixation (Muller, 1933; Hillmer and Gest, 1977; Richardson et al., 1988; McKinlay and Harwood, 2010a, 2010b; McKinlay and Harwood, 2011). Here, we tested whether the presence of DMSO enables phototrophic growth of R. palustris CGA009 on butyrate. As expected, we found that strain CGA009 grew phototrophically on butyrate in the presence of bicarbonate (i.e. carbon fixation; Fig. 2A) and absence of fixed nitrogen (i.e. nitrogen fixation; Fig. 2B) but not on butyrate alone. Butyrate with or without DMSO did not support growth (Fig. 2A) unless carbon or nitrogen fixation were also active. These data suggest that R. palustris CGA009 cannot use DMSO as the exclusive electron sink under the phototrophic growth conditions used in our research study. This finding is compatible with previous experiments demonstrating that DMSO does not support anaerobic growth of R. palustris CGA009 on succinate in the dark (Oda et al., 2008).
Though DMSO did not support growth, we did detect the product of DMSO respiration, DMS, in the culture headspace. DMS was produced in the headspace of both the non-growing, non-nitrogen fixing (Fig. 2C) and the growing, nitrogen fixing (Fig. 2E, F) cultures. DMS production appears to be independent of growth, as suggested by the continual increase of DMS concentrations even in stationary phase and the production of DMS in non-growing cultures (Fig. 2C, E). No abiotic DMS production was detected in sterile DMSO controls. Active cultures that lacked DMSO also did not produce DMS, suggesting that observed DMS was produced from DMSO reduction rather than an alternative pathway. We did not test whether sterilized R. palustris organic matter could catalyze DMS production from DMSO. These data show that R. palustris CGA009 cultures catalyze the reduction of DMSO into DMS.
Discussion
In the published dataset, we observed changes in the growth rate of the Mo- and V-nitrogenase strains based on the addition of DMSO (see Fig. 2B in the published paper). Because we assumed that these strains could use DMSO as an electron acceptor, we interpreted the change in growth rate as evidence that use of the V-nitrogenase isoform as an electron sink could promote growth relative to the use of other electron sinks like DMSO. Though it does not alter our overall findings, we are now less certain of this interpretation. Given the data described above, it is unclear whether or to what extent DMSO acts as a physiological electron acceptor for R. palustris CGA009 and its derivative strains. While we have generated evidence that R. palustris CGA009 cultures produce DMS in the presence of DMSO, we have not established whether this process is catalyzed enzymatically or abiotically in the presence of R. palustris derived organic matter, or how the electron flux associated with DMSO reduction is linked to intracellular electron balance and energy conservation (i.e. the electron transport chain). Here, we discuss these questions in greater detail.
To grow, R. palustris must balance multiple constraints, including electron balance, energy conservation and nutrient acquisition. An important consideration for its impact on growth is whether DMSO reduction is only an electron sink, or also serves other functions like nutrient acquisition or energy conservation. The phylogenetic analysis identified one gene, RPA4653, encoding a homologue of BSO reductase, with potential DMSO reduction activity. In the nearest characterized relative, a Rhodobacter strain, BSO reductase uses the cytosolic electron donors NAD(P)H and is disconnected from energy conservation (Pollock and Barber, 1997, 2001). Though the process responsible for DMS production in the R. palustris cultures was not tested, if this process were to rely on the putative BSO reductase RPA4653, it would suggest that DMSO reduction functions as an electron sink influencing cellular redox homeostasis but cannot serve as an electron acceptor for energy conserving anaerobic respiration. Though in our experiments the R. palustris cultures were grown phototrophically, able to obtain energy from light, this would preclude R. palustris CGA009 growth on DMSO in the dark. As we did not include an autoclaved control culture, we also cannot exclude the possibility of abiotic, organic matter-dependent DMS production.
To estimate the magnitude of the electron flux diverted to DMSO reduction in cultures grown on succinate and acetate, we compared the number of electrons required to produce the observed amount of DMS to that required for the reduction of N2 into biomass (Fig. 1G). Using these back of the envelope calculations, the electron flux from DMSO reduction into DMS is estimated to be about an order of magnitude lower than that of N2 reduction during exponential growth. The lack of growth on butyrate with DMSO suggests that this flux is too low to support growth in the absence of additional electron sinks. We did not test whether an electron flux of this magnitude is sufficient to explain the observed convergence of Mo- and V-nitrogenase based growth rates reported in the presence of DMSO (see Fig. 2B in published paper). These results relate to two interesting questions, the extent to which microbial growth in the environment is based on singular versus multiple electron sinks and the thresholds at which partial electron acceptors become physiologically relevant determinants of growth rate.
In addition to its possible role as a partial electron sink, DMSO has numerous pleiotropic effects on cells that could also have been responsible for the observed changes in growth rate (see Fig. 2B in published paper). These include increased permeability of the cell membrane and facilitation of membrane fusion, alterations in protein secondary structure to favor beta-sheets over alpha-helices, changes in DNA superstructure stabilizing the Z-DNA conformation, changes in quorum sensing, scavenging of ROS, and alteration in photopigments and light harvesting complex composition (Hakobyan et al., 2012; Tunçer et al., 2018; Camp et al., 2020) and references therein). DMSO does not generally appear to alter cell viability (see e.g. (Zhao et al., 2011) but has been reported to impact nitrogenase activity (Hakobyan et al., 2012) and alter microbial growth rates (Ansel et al., 1969; Hakobyan et al., 2012; Gabrielyan et al., 2015) under some conditions. It is also noteworthy that all known DMSOR use the trace metal molybdenum (Mo), and that, besides the genomic analysis in (Peng et al., 2018), the intracellular prioritization of Mo for Mo-nitrogenase versus CISM superfamily proteins has not been explored. Likewise, possible changes in Mo and vanadium (V) oxyanion bioavailability due to chelation by DMSO have not been studied. Clearly, there are several reasonable explanations for the observed growth rate changes that are independent of using DMSO as an electron sink or terminal electron acceptor for respiration.
Together, these new data neither exclude nor confirm the possibility that the changes in growth rate upon addition of DMSO in our initial experiments (see Fig. 2B in published paper) are due to its use as a partial electron sink. The bioinformatic analysis strongly suggests that R. palustris CGA009 cannot use DMSO as a terminal electron acceptor for anaerobic respiration, but the measured production of DMS suggests that DMSO may still act as an electron sink. The homology between R. palustris CGA009 gene RPA4653 and the DMSO reducing BSO reductases hints that it may be the enzyme responsible for the observed DMS production, though organic matter catalyzed abiotic production cannot be excluded. At an electron flux roughly a tenth that of N2 reduction, it is possible that this is simply too small to enable growth on butyrate, though we were unable to test to what extent this electron flux influenced the growth rates on acetate and succinate. Collectively, the ambiguities described here highlight the importance of studies to better connect DMSOR protein sequence to function, with important implications for metabolic modeling and prediction of the specific biogeochemical impacts of this widespread class of enzymes (McEwan et al., 2002; Kappler et al., 2019; Miralles-Robledillo et al., 2019; Leu et al., 2020).
Conclusion
We found that living cultures of R. palustris CGA009 catalyzed the production of DMS from DMSO. DMS production depended on the presence of substrate DMSO and was independent of growth and nitrogen fixation status. As a sole electron acceptor, the presence of DMSO did not support the growth of R. palustris CGA009 cultures on butyrate, suggesting that DMS production may not be enzymatically catalyzed or integrated into the energy conserving membrane electron transport chain, or that its maximal flux is simply too low to support growth on its own. We were unable to resolve whether the growth rate changes observed in our initial experiments were due to use of DMSO as an electron sink or due to its numerous pleiotropic effects. These data do not alter the major findings in our initial publication, but we apologize to any readers whom we may have misled about the possibility of using DMSO as an electron acceptor for R. palustris CGA009. With this corrigendum, we hope to clarify that DMSO may not be a suitable electron acceptor for experiments with R. palustris CGA009 and its derivative strains and bring attention to its potentially confounding, increasingly well documented pleiotropic effects on cell growth.
Experimental Procedures
Microbial culturing
Cultures were grown and monitored as in the initial manuscript, with the minor difference that the yeast extract was not included in the growth medium. As before, the concentration of added DMSO (Sigma Life Sciences, molecular biology grade) was 40 mM, the concentration of added sodium bicarbonate was 10 mM, and the concentration of added ammonium sulfate was 7.5 mM. We note that, though we continue to refer to the cultures as R. palustris, a recent phylogenetic analysis has suggested that strain CGA009 and its mutant derivatives may in fact be more closely related to R. rutila (Imhoff et al., 2020). This phylogenic reclassification does not impact our studies.
Bioinformatics
We searched for any annotations to ‘DMSO,’ ‘dimethyl sulfoxide,’ ‘dimethyl,’ ‘TMAO’ and ‘trimethylamine’ in the NCBI databases for R. palustris CGA009 (last annotated on 4/16/2020) and BisB18 (last annotated on 4/4/2020). In the KEGG Genome database, we searched for these annotations as well as the genes and enzymes related to the KEGG Compounds DMSO (C11143) and TMAO (C01104) in R. palustris CGA009 and BisB8 genome (KEGG Genome Entry: T00153 and T00336, respectively). This search revealed the previously identified TMAO reductases in R. palustris BisB18 but did not identify any hits for R. palustris CGA009. To confirm the absence of unannotated enzymes capable of DMSO reduction, we also used the NCBI blastp web interface to search for potential DMSOR homologs in the R. palustris genomes, using default parameters (Altschul et al., 1990). To characterize these sequences, we constructed a phylogenetic tree for the CISM superfamily based on a curation from (Leu et al., 2020). Accession numbers and amino acid sequences are included in Table S1. Sequences were aligned using the MAFFT plugin (version 7.450; default parameters; (Katoh et al., 2002; Katoh and Standley, 2013) within Geneious (version 2020.2.4; Biomatters Ltd, New Zealand) and edited to remove sites with >50% gaps. The final alignment length was 776 residues. The tree was generated using the Geneious plugin of FastTree (version 2.1.11; default parameters; (Price et al., 2010) and plotted in FigTree.
DMS concentration measurements
DMS concentrations in the culture headspace were measured on a GC-8A with a flame ionization detector (Shimadzu Instruments; He carrier gas; column, Supelco HayeSep N; column temperature, 110°C; detector temperature, 150°C). The calibration curve was made by diluting a known volume of refrigerated DMS (Sigma Aldrich, ≥99%) directly into a balch tube with 10 ml sterile culture medium, mixing manually through vigorous shaking, subsampling and, like the samples, loading 1 ml onto the instrument using an injection loop. To avoid carryover, the sample loop was flushed with at least 20 ml of room air between injections.
Calculation of relative electron fluxes used for N2 and DMS reduction
We estimate the total number of electrons used for N2 reduction into cellular nitrogen using the approximate value of 10−15 mol N cell−1 and the empirically observed relationship cells mL−1 = 2.29 x 109 x OD660. This estimate ignores the electron flux to H2 during nitrogen fixation, a third of the flux at minimum, as well as the documented variability in the cellular nitrogen quota. Given these assumptions, we note that these calculations should be viewed as order of magnitude estimates, intended to provide an intuitive comparison of electron fluxes rather than precise estimates. Electron fluxes are calculated using the stoichiometry 2 e−: DMS and 6 e−: N2.
Acknowledgements
We thank James (Jake) McKinlay for bringing the DMSO issue to our attention, Linta Reji for help with the bioinformatic analysis, and the entire Zhang Lab for feedback on these experiments and the results.
