The Biology of Free Guanidine as Revealed by Riboswitches

Abstract

The experimental validation of three distinct riboswitch classes has revealed that many bacterial cells naturally produce guanidine, and that living systems have evolved a variety of genes involved in the metabolism and transport of this toxic metabolite. There are numerous biochemical curiosities and mysteries that spring from these advances, which will make for interesting research topics in the coming years.

Recently,^1-3 we have experimentally validated the existence of three novel riboswitch classes (Figure 1) that naturally sense free guanidine. Perhaps the most surprising implication of this discovery is that biology cares about free guanidine at all. We biochemists are most familiar with the guanidyl moiety as a prominent substructure of fundamental metabolites such as arginine, creatine, and guanine, or as a general denaturant of proteins in its free form. In addition, large amounts of guanidine have been produced for use as feed-stocks for the production of plastics or as slow-reacting explosives for battleship guns and automobile air bags. Curiously, the scientific literature is almost devoid of discussions about the biological manipulation of free guanidine, or its potential roles in the basic metabolism of cells. Three decades ago, some of this attention was directed at assessing whether bacteria could be used as bioremediation agents to clean up sites that became contaminated during explosives manufacturing and testing, but no definitive understanding of natural guanidine biology emerged from this work.

Figure 1.

The family portrait of guanidine riboswitches. Depicted are the conserved sequences and predicted secondary structure models for (A) guanidine-I, (B) guanidine-II, and (C) guanidine-III riboswitch aptamers as determined by comparative sequence analysis.^1-3 A more detailed model for guanidine-I, based on x-ray data, is published elsewhere.⁵ Circles and numbers identify nucleotides in guanidine-I riboswitches that form the ligand binding site, as depicted in Figure 2.

What have these three riboswitch classes taught us about guanidine biology? Also, how can we be so sure any of these conclusions regarding the importance of free guanidine are correct? Certainly, the riboswitch route to revealing insights on guanidine metabolism did not follow an easy path. In 2004, a structured noncoding RNA element called the ‘ykkC motif’ was reported as a candidate riboswitch.⁴ Unfortunately, for the next 10 years, we and others did not succeed in identifying the natural ligand for members of this RNA class, which we now call class I, or ‘guanidine-I’, riboswitches.¹ Frequently, the ligands for newfound riboswitches are easily identified by evaluating the functions of the genes associated with the RNA aptamer. If a gene controlled by a riboswitch candidate either makes or transports compound X, then you likely have a riboswitch that senses compound X. However, despite having an abundance of clearly annotated genes associated with ykkC motif RNAs, these did not quickly lead to the ligand.

There were three major problems that conspired to frustrate any attempts to find the natural ligand for ykkC motif RNAs. First, gene annotations can be wrong or misleading, which was certainly the case for this riboswitch class. Genes most commonly associated with the RNA motif were annotated as coding for urea carboxylase enzymes, although we now know¹ that these proteins, with much greater efficiency, actually carboxylate guanidine (a close analog of urea). Members of another common class of genes associated with ykkC motif RNAs are known to code for efflux pumps targeting certain classes of antibiotics. However, we¹ and Dr. Randy Stockbridge at the University of Michigan (personal communication) have produced evidence to suggest that the drug transporters are an evolutionary offshoot of a much broader collection of transporters whose natural (and likely primal) function is to selectively eject guanidine from cells.

Second, and remarkably, there appears to be at least two riboswitch classes commingled within the ykkC motif collection of RNAs. Three quarters of these RNAs eventually proved to represent guanidine-I riboswitches, but the rest have key nucleotides altered near the binding pocket that alter ligand specificity.¹ This fact confounded efforts to choose a single ligand candidate because the original pool of associated genes correspond to at least two different ligands. These ykkC motif subtypes first had to be teased apart to make sense out of the mixed clues derived from gene associations.

Third, with no rich history of studies on guanidine metabolism, we and others had no reason to consider the possibility that riboswitches would exist that sense free guanidine. It was not until we hypothesized that the ykkC motif might recognize a toxic compound and screened a wide variety of them that we identified guanidine as a potential riboswitch ligand. In fact, we also carefully considered the possibility that the natural ligand for the three riboswitch classes might not be guanidine, but rather some other biological compound (known or unknown to science) that carries a guanidinyl moiety, such as arginine or a related metabolite. However, there is no biochemical or genetic evidence to suggest that a compound other than guanidine is the natural ligand. Furthermore, structural biologists have already created atomic-resolution models of guanidine-I riboswitch aptamers based on x-ray data,⁵ which reveal these RNAs form a precise binding pocket for a single guanidinium ion (Figure 2). The ligand binding site appears to exploit all possible hydrogen bonding contacts, cation-π interactions, and shape complementarity to achieve selective binding and the rejection of even close analogs of guanidine, including urea.

Figure 2.

Atomic-resolution structure model of a guanidine binding pocket formed by RNA. The view is of the guanidine-I riboswitch aptamer naturally located upstream of the previously annotated urea carboxylase gene of S. acidophilus. Ambiguous hydrogen bonds are depicted for G45. Not depicted are two guanosine residues (G72 and G88) that form the top and bottom of the binding pocket and potentially form cation-π interactions. Additional details regarding the entire structural model are described elsewhere,⁵ and a similar structural model also has been proposed by Ailong Ke and coworkers from Cornell University (personal communication).

Additional observations also suggest that guanidine metabolism serves a critical role in biology, although it has been quite well hidden by most cells. For example, two additional noncoding RNA motifs, called mini-ykkC and ykkC-III, had been identified by bioinformatics and were proposed to be riboswitches or other types of regulatory elements that respond to the same ligand as the original ykkC motif. Similarly, several other important metabolites are sensed by more than one riboswitch structural class, including the coenzyme S-adenosylmenthionine (SAM), the modified nucleobase prequeuosine-1 (preQ₁), and the bacterial signaling compound cyclic diguanosine monophosphate (c-di-GMP). These riboswitches and their metabolite ligands are very common in bacteria, and the existence of redundant riboswitch classes demonstrates that RNA can selectively bind these targets with multiple different pockets formed entirely by RNA. Moreover, the abundance of such riboswitches reflects how important it is for cells to manage the concentrations of these compounds, which are essential for many bacteria.

Indeed, although mini-ykkC and ykkC-III motifs carry distinct sequences and secondary structures compared to guanidine-I riboswitches (Figure 1), members of both motifs selectively bind guanidine^2,3 and regulate genes that are now implicated in guanidine toxicity response processes. Notably the mini-ykkC RNAs, now called class II or ‘guanidine-II’ riboswitches, are almost comical in their structure and function. The most-conserved portions of this class are comprised of two small and essentially identical hairpin structures that appear to cooperatively bind two guanidine molecules.² We speculate that the tandem hairpins form a ‘hand-to-wrist’ architecture wherein each conserved loop sequence docks with its neighbor’s conserved duplex to form two identical binding sites for the ligand. Presumably, RNA engineers could exploit intermolecular complex formation by this tiny motif to create self-aggregating RNAs that are triggered to assemble only upon the addition of guanidine.

Gene products that function as guanidine exporters or that carboxylate guanidine, presumably leading to the splitting of guanidine and the generation of CO₂ and ammonia that could be bubbled off as waste, indicate there are natural conditions when many bacteria produce too much guanidine. These cells might then have no choice but to expend energy and dispose of precious fixed nitrogen to avoid guanidine toxicity. Early results¹ suggest that the production of free guanidine might be the direct result of cells utilizing unfavorable sources of energy (such as their own metabolites) to survive when preferred sources of food such as glucose are scarce. Cells could efficiently derive energy from guanidine-containing molecules by initially avoiding the degradation of this moiety, and then only expending energy degrading or ejecting guanidine when it reaches intolerable levels.

There remain many unanswered questions regarding the biochemistry of guanidine metabolism and the logic of regulating the expression of genes that help overcome its toxicity. Regardless of the details of the natural regulatory and metabolic circuitry, it should be possible to exploit these components to create synthetic organisms that have new utility. For example, reaching the decades-old goal of guanidine bioremediation by bacteria should be facilitated by a deeper knowledge of the guanidine biology exposed by these riboswitches. Furthermore, pathways could be reorganized to turn bacteria into guanidine production factories. If this metabolic reconfiguration were coupled with the capability to fix atmospheric nitrogen gas, one could exploit domesticated bacteria that generate bioavailable nitrogen by fermentation. Moreover, similar opportunities for discovery and engineering are certain to result from additional explorations into the functions of other candidate riboswitches.