Serendipity Reveals the Function and Physiological Role of a Large Family of Proteins

ABSTRACT

Microbial metabolism involves a complex set of interactions between metabolic pathways that include proteins of both known and uncharacterized function. While investigating the physiological strategy used by actinomycetes with two RpoB paralogs, Damiano et al. uncovered the endonuclease activity of a member of the Rid family (F. Damiano, M. Calcagnile, D. Pasanisi, A. Talà, et al., J Bacteriol, 204:e00462-21, 2022, https://doi.org/10.1128/JB.00462-21). While this finding was peripheral to the original question posed by the authors, it has considerable significance. The study by Damiano et al. highlights how unexpected, but fundamental, information can be gained by following phenotypic leads.

TEXT

Microbial metabolism includes a collection of processes, both regulatory and metabolic, that are integrated to generate the robust and responsive metabolism characteristic of microbes. Our current understanding of physiology and metabolism has been facilitated by historical and continuing efforts to identify components and define regulatory networks. Yet a better understanding of metabolism requires the integration of basic knowledge about each component into a robust, malleable metabolic network that can meet challenges from the environment.

The process of building a system piece by piece can be complicated by the presence of multiple components with the same demonstrated or presumed activity. Historically, these “redundancies” have been identified phenotypically and/or with in vitro biochemical assays. For instance, many organisms have multiple acetohydroxy acid synthases in branched-chain amino acid biosynthesis (1); pyridoxal 5′-phosphate (PLP)-dependent amino acid transaminases are characterized by promiscuous substrate specificity (2), among many others. Such apparent redundancies complicate the assignment of a specific role for each enzyme in the metabolic network. Further biochemical and genetic analyses can identify regulatory and/or kinetic differences between the proteins that can provide a physiological context for each one of them. Clearly, these overlapping functions are critical for the robust and adaptable metabolism microbes need to ensure maximal fitness.

With the number of sequenced genomes increasing every day and genome annotation becoming more automated, different caveats with respect to redundancy arise. It is not uncommon for genomes to have multiple genes with the same annotated function or be assigned to the same gene family. Yet, despite these computational designations, the likelihood that multiple genes are performing the same role under the same conditions in the organism is minimal. Rather, functional annotations may not provide information about the preferred substrate, gene expression differences, etc., and each of these parameters helps discriminate between components inferred to be “redundant.” Advanced understanding of the metabolic/regulatory properties of cellular components will provide clarity about the role that each component plays in the survival and fitness of microbes in their environments.

The manuscript by Damiano et al. in this issue (3) provides an example of metabolic knowledge gained when the specific roles and regulation of apparently redundant proteins are probed more deeply. In pursuing the differential role of two paralogs of the beta subunit (RpoB) of the RNA polymerase (RNAP) of Nonomuraea gerenzanensis, these authors serendipitously came upon the biochemical and physiological role of a member of the large Rid superfamily of proteins. Interestingly, the genome of this organism encodes 10 putative members of the Rid superfamily, and the authors identified the role for a specific member, namely, Rid7C.

Nonomuraea gerenzanensis is an actinomycete of industrial importance (4). This organism is known for its production of a glycopeptide (A40926) that has activity against Neisseria (5) and is the precursor to a novel antimicrobial (dalbavancin) that is active against Gram-positive pathogens, including methicillin-resistant Staphylococcus aureus (MRSA) (6, 7). In N. gerenzanensis, as in several other actinomycetes, the genome encodes two RpoB paralogs; one is encoded by rpoB(S) (the wild-type gene) and the other one by rpoB(R) (a mutant gene). Consequently, this organism encodes two types of RNA polymerases (RNAPs) (8). The biological significance of this apparent redundancy is of interest. A detailed understanding of RNAPs and sequence comparison between the paralogs in N. gerenzanensis suggested RpoB(S) and RpoB(R) could have distinct specificities in gene expression (8, 9). Consistently, RpoB(R) specifically controls production of some secondary metabolites and a number of adaptive behaviors in response to environmental pH (7). Further, expression of rpoB(R) from N. gerenzanensis in the heterologous host S. lividans strain 1326 increased the production of two antibiotics, undecylprodigiosine (Red) and actinorhodin (Act) (9). In total, results from multiple studies support the hypothesis that presence of RpoB(R) differentially regulates various genes relevant to the developmental strategy of actinomycetes and, as such, allows greater fitness under a variety of stresses (9).

To further understand these results, it became important to define the mechanism used to regulate rpoB(R), which could then define the strategy to control the relevant genes. Two distinct, developmentally regulated, rpoB(R)-containing mRNAs were identified. These transcripts were initially attributed to different transcription start sites and labeled TSS1 and TSS2 (9). The existence of two mRNAs suggested a potential regulatory mechanism. Data consistently showed that the transcript starting at TSS2 was overrepresented in conditions where RpoB(R) was presumed to be activating genes. Damiano et al. (3) elegantly established that the transcript starting at TSS2 was derived from one starting at TSS1. This result indicated the two transcripts were the consequence of an RNA processing event rather than two sites of transcription initiation. In silico secondary structure predictions suggested that the mRNA starting at TSS1 formed hairpin structures that would occlude the Shine-Dalgarno (SD) site and, thus, quite likely suppress translation of RpoB(R). The connection of this work to the Rid superfamily of proteins was derived from the authors’ hypothesis that an endonuclease cleavage event at TSS2 would yield a transcript with an accessible SD site and the concomitant increased translation of RpoB(R). In pursuit of this idea, Damiano et al. implicated a member of the Rid/YER057/UK114 protein superfamily, GenPept accession no. SBP00267.1 (Rid7C), in the posttranscriptional regulation of rpoB(R) (3).

The Rid/YER057c/UK114 (or Rid) superfamily of proteins is broadly distributed and highly conserved (reviewed in reference 10). Within this superfamily, there are nine subfamilies, namely, RidA, Rid1 to Rid7, and the recently added RutC-like (11, 12). Members of the RidA subfamily are found in all domains of life, while the remaining subfamilies are found only in prokaryotes, predominately in bacteria. Identification and characterization of the biochemical and cellular function of a RidA protein were first accomplished in Salmonella enterica (13). Central to this work was the identification of the reactive enamine 2-aminoacrylate (2-AA), an obligate intermediate of serine/threonine dehydratases, as the stressor that can inactivate certain cellular enzymes (reviewed in reference 14). Subsequent work in S. enterica defined a paradigm of endogenous metabolic stress and clarified the role of RidA in modulating this stress, which has been studied in multiple organisms (reviewed in reference 15).

The crux of the exciting contribution by Damiano et al. is the absence in Rid7C of the active-site Arg residue known to be essential for the deaminase activity of RidA (R105 per S. enterica numbering). In addition to RidAs, members of subfamilies Rid1 to Rid3 and RutC also have this arginine, and each of these proteins has enamine deaminase activity. Members of subfamilies Rid5 to Rid7 lack the above-mentioned key arginine, and no deaminase activity has been described for these proteins. Thus, the presence of the arginine residue is a strong predictor of deaminase activity (11, 16).

Annotation of the larger Rid family remains heterogenous in the literature and data sets. Notably, there are no annotations that reflect the significance of arginine in predicting deaminase activity. Prior to the work by Damiano et al. (3), no biochemical activity had been identified for a member of the Rid family lacking the key arginine. Like many bacteria, the genome of N. gerenzanensis encodes Rid proteins from multiple subfamilies. Several of the alluded rid genes are part of operons of metabolic genes, and some of these Rid proteins do have the above-mentioned key arginine. These Rid proteins can be hypothesized to contribute to the relevant pathways with a deaminase activity, as shown for RutC and suggested for others (17). Other Rid family members are untethered by gene synteny and hypotheses about their cellular role remain elusive. The study by Damiano et al. unexpectedly implicated a member of the Rid7 subfamily (i.e., Rid7C) in the posttranscriptional regulation of the RpoB(R) subunit of RNAP. What is most exciting about these results is the fact that Rid7C lacks the active-site arginine.

In the study described in this issue, Damiano et al. defined the mechanism controlling rpoB(R) translation and, as a result, increased our understanding of the RpoB(R)-dependent increase in antibiotic production. The authors made the key observation that the concurrent expression of Rid7C (and, to a lesser extent, Rid7B) significantly increased rpoB(R)-dependent antibiotic production in S. lividans. This fundamental phenotypic observation was pursued to dissect the mechanistic role Rid7C played in regulating the level of translation of RpoB(R). Understanding this regulation led the authors to an explanation for the increased activity of biosynthetic pathways controlled by this RNAP. In a series of well-designed and rigorously performed experiments, the authors acquired data to support their hypothesis that rpoB(R) is posttranscriptionally regulated. As noted above, the longer of the mRNAs (TSS1) was expected to allow less translation of RpoB(R) than TSS2 since the SD site was occluded by secondary structure. A hypothesized endonuclease attack released the shorter mRNA (TSS2), making the SD site accessible, allowing translation of rpoB(R). Increased translation of rpoB(R) then led to expression of the relevant genes and the resulting antibiotic production. In vivo and in vitro data convincingly showed that Rid7C has endonuclease activity and unequivocally showed that the presence of this protein resulted in the accumulation of the mRNA starting at TSS2 (3). In N. gerenzanensis, the production of the A40926 antibiotic is dependent on rpoB(R), and the authors suggested this metabolite might regulate Rid7C activity, which would, in turn, regulate the translation of rpoB(R). Data from multiple experimental approaches supported a role for A40926 in allosterically inhibiting the endonuclease activity of Rid7C. This finding suggests a negative feedback loop and raises multiple additional questions that will most likely be approached in future work.

The significance of the Damiano et al. study extends well beyond the elegant elucidation of the posttranscriptional regulation of rpoB(R) and its impact on the production of antibiotics in actinomycetes that have this paralog. An unanticipated consequence of pursuing the regulatory questions that prompted this study was the first demonstration of a biochemical activity for a Rid7 subfamily member. It is worth emphasizing that this subfamily does not have the arginine that is critical for the activity of other family members, particularly those in the RidA subfamily (11, 13). Thus, the protein in N. gerenzanensis designated Rid7C becomes the first member of the Rid7 subfamily whose activity has been verified in vivo and in vitro.

As expected with a novel biochemical discovery, this work raises numerous fascinating questions to be addressed in future work. For instance, is endonuclease activity conserved across Rid7 members or potentially across the Rid5 to Rid7 subfamilies which lack the arginine required for deaminase activity? Further, Rid7C was initially purified with an associated RNA, specifically the M1 RNA of RNase P. This result suggests there may be more to the modulation of this enzyme activity than first thought. Is this an artifact of purification from Escherichia coli? Or could RNA provide specificity for the site of endonuclease cuts? Such a scenario might indicate a broader role for the Rid proteins and potentially an involvement in numerous regulatory processes. If true, this would have parallels to the enamine deaminase activities that have been characterized for other Rid proteins. These proteins deaminate multiple substrates in vitro, but the only one shown to have in vivo relevance is 2-AA (15, 16). Likewise, consider that while the human RidA homolog UK114 has been annotated as an endonuclease, it has enamine deaminase activity, as expected by the presence of the relevant arginine residue (13). One wonders whether other arginine-containing Rids (like RidA) have endonuclease activity in addition to deaminase activity. If so, do they require RNA for this activity? It will be exciting to see what residues are required for the nuclease activity of Rid7C and to investigate whether other subfamily members have this or a related activity. It will be important to extend the studies of the Rid subfamily members to learn not only about what activity the proteins have in vitro but to reveal their relevance to cell physiology.

Pursuing the role of apparently redundant genes can uncover regulation and specificity that cannot be gleamed from the sequence of a gene alone. In the case of rpoB(R), a role in gene expression was almost guaranteed based on our extensive knowledge of RNAPs. The connection of this regulatory system to the Rid family members was certainly not intuitive, particularly since there were 10 members of this superfamily in the N. gerenzanensis genome. Perhaps Damiano et al. were drawn to this family of proteins by the general endonuclease annotation that permeates the literature, despite the deaminase activity that has been demonstrated for numerous family members.

Defining protein function often requires metabolic detective work, coupled with more than a little serendipity. The uncertainty over annotation and the inability to predict a specific role in the metabolic network can explain why targeted approaches to define function, or assign regulation, are often unsuccessful. The Rid superfamily not only highlights the challenge in assigning function to genes but underscores the difficulty in annotating genomes to make the relevant information available to the community. How can the community best access the information that can facilitate the analysis of genes of unknown function, or regulatory paradigms, when that information is not a central element of a manuscript?

In closing, understanding metabolism is a long and arduous journey that must be approached from many different angles and with different perspectives and starting interests. The study by Damiano et al. (3) provides a great example of how the pursuit of specific metabolic and regulatory questions can generate functional information that could not be anticipated a priori—this is the beauty and excitement of physiological studies.

The views expressed in this article do not necessarily reflect the views of the journal or of ASM.