Enzyme sub‐functionalization driven by regulation

Abstract

The emergence of functional novelties during protein evolution has puzzled scientists for many years. Most proposed models focus on repeated duplication‐divergence cycles, but the entanglement of selection pressures acting on the control of transcriptional and enzymatic activity, for example, by metabolites, has not been addressed so far. In this issue of EMBO Reports, Noda‐Garcia et al 1 describe two glutamate dehydrogenase paralogs from Bacillus subtilis with very similar sequences and under two distinct modes of activity control. The functional divergence of these two enzymes during evolution is driven by an interlinked combination of differences between their enzymatic properties and their transcriptional regulation. This article thus illuminates another level of complexity in molecular evolution that may help understand the hitherto unexplained co‐existence of paralogous genes that at first sight appear to be functionally redundant.

The molecular basis of how new functions arise is a crucial but still poorly understood issue in evolutionary biology and has ramifications for almost every biological discipline. Many concepts have been proposed over the last five or so decades. Most of these assume that new functions require new protein coding genes (CDS), and that these new CDSs (paralogs) are derived from existing ones by duplication and subsequently diverge by accumulation of adaptive mutations (Fig 1A). Some of the processes which were proposed to lead to novel functionalities involve: (i) transcriptional differentiation, that is, the two CDS copies are differentially expressed under specific conditions or in different tissues 2; (ii) neo‐functionalization, that is, after the new gene loses some of its old function, it is free to assume a new function since its previous function is backed up by the old duplicate CDS 3; and (iii) sub‐functionalization, that is, both, the new copy and the old one, had multiple functions and specialize on either function after duplication 4.

Figure 1. Evolution of proteins through gene duplication.

(A) Sub‐functionalization model in which an ancestral gene encoding for an enzyme with a main activity A and latent activity B is duplicated. The adaptive path (toward the latent function) effectively results in a shift in substrate preference. (B) Divergent evolution of the sensitivity of hexamer formation to pH and glutamate level for glutamate dehydrogenases GudB and RocG in B. subtilis NCIB 3610 as a result of differential regulation. (C) Promoter swapping between RocG and GudB has a negative effect on fitness when arginine is used as a growth substrate. (D) Reductive deamination of glutamate catalyzed by the hexameric forms of GudB and RocG.

However, as so often in biology, reality is not clear‐cut and no answer applies to all systems. In fact, all three (and perhaps several more) processes may co‐occur, co‐evolve, and interact. For example, recent simulations suggested that, within one system, all three strategies might prevail, with respective probabilities depending on the mutation rate, the rate of gene duplications and, above all, the strength of the fitness pressure calling the adaptive benefit of the novel gene 5. Over the last years, it became clear that, at least in enzymes, a crucial property of a protein's ability to adopt a new functionality is its promiscuity, that is, its multi‐functionality, for example, by acting on several substrates or by generating various products. The functional divergence of related sulfo‐ and phosphohydrolases, for example, is accompanied by an inversion of substrate specificity, that is, the phosphodiesterases have latent sulfatase activity and vice versa 6, 7.

The entanglement of transcriptional regulation and functional specificity is particularly intriguing, since changes in the promoter region are required for the former and changes in the coding region for the latter. So far, such changes have been considered separate processes although it is quite clear that, for example, promoters are replaced with other promoters quite often during evolution. It is also obvious that a rapid change in transcription control would help make a protein's function more specific in a given context and, consequently, less prone to be lost again by drift. In promoters, a single nucleotide exchange can influence the regulation of a whole biological system. Indeed, many paralogous CDSs are expressed differentially which indicates that already slightly divergent duplicates might have co‐evolved in short time with their promoters 2.

In this issue of EMBO Reports, Noda‐Garcia et al 1 at the Weizman Institute (Rehovot, IL) have devoted a thorough and exceptionally well‐designed study to approach this question. As a case study, they use two paralogous enzymes, RocG and GudB, which are highly similar (~75% identity) NAD⁺‐utilizing glutamate dehydrogenases (GDHs) present in strain NCIB 3610 of the well‐studied Gram‐positive bacterium Bacillus subtilis and several related Bacilli. GDH catalyzes the conversion of glutamate into ammonia, an essential nitrogen source, and 2‐oxoglutarate that feeds directly into the central carbon and energy metabolism via the citric acid cycle. As a result, the presence of GDH enables B. subtilis to use glutamate, or any other amino acids the bacterium can convert into glutamate (e.g. proline or arginine), as a source of nitrogen, carbon and energy.

In vitro, RocG and GudB show similar levels of catalytic efficiency (k _cat/K _M). However, they are differently and differentially regulated. Expression of RocG is induced by arginine and to a lesser extent by proline. Use of these amino acids as a growth substrate requires GDH activity. Whenever other, more easily usable carbon and nitrogen sources, such as glucose and ammonium, are present, GDH activity is not required. Under these circumstances, the expression of RocG may even be disadvantageous. Indeed, both glucose and ammonium repress RocG expression. In contrast, GudB is expressed under a strong constitutive promotor and is the major provider of GDH activity in NCIB 3610, even under conditions where RocG is expressed. Strikingly, the energetically costly expression of GudB does not impair organismal fitness in the presence of glucose and ammonium, conditions for which GDH activity is unnecessary. At first sight, either RocG or GudB is redundant. However, the deletion of either of the two GDH paralogs significantly reduces the growth rate when the bacterium uses arginine as a growth substrate.

Noda‐Garcia et al 1 next wanted to understand how transcriptional control and regulation of GDH activity interact and how the two regulatory mechanisms, which are apparently different for GudB and RocG, might have co‐evolved with their respective enzyme copies. Since both enzymes have a very similar level of GDH activity, swapping the promotor and coding regions with one another, that is, express RocG under control of the native GudB promotor and vice versa, would be expected to have little to no effect. However, complementation of a double GDH knockout (∆gudB/∆rocG) with RocG under control of the native GudB promotor showed decreased fitness compared to complementation of ∆gudB/∆rocG with RocG under control of its own promotor, when using arginine as a growth substrate. The same adverse effect of being under control of the non‐native promotor was observed for GudB. Furthermore, constitutive expression of RocG (i.e. under GudB promotor control) in a GudB single knockout strain (∆gudB) has a negative effect on fitness when the bacterium is growing in glucose‐ammonia medium. Complementation of the ∆gudB strain with RocG expression under control of its own promotor did not show any effect on fitness in glucose‐ammonia medium. This promotor‐gene incompatibility suggests that for both RocG and GudB the enzyme has co‐evolved with its respective promoter.

Both enzymes are active as a homohexamer, and for both enzymes, the fraction of assembled homohexamers is affected by pH and the glutamate concentration (Fig 1B and D). An elevated level of glutamate also increases the stability of the hexameric complex as well the thermostability of both GDHs. However, under suboptimal conditions (lower pH, low glutamate), most of RocG, but little if any GudB assembles into hexamers, rendering GudB essentially inactive. At higher pH and high glutamate concentrations, both GudB and RocG assembled into practically 100% hexamers. Furthermore, the positive effect of high glutamate concentrations on stability is stronger for GudB. Apparently, the adverse effects of constitutive expression of GudB are countered by regulation of its activity at the enzyme level, in contrast to the control at the level of enzyme expression seen for RocG.

Intriguingly, the evolution of both paralogs is strongly affected by the regulatory regime they are acting under (Fig 1B), to such an extent that there is no cross‐compatibility between the promotor regions and the enzymes under their control (Fig 1C). This establishes also a striking example of epistasis between the promoter and the gene it regulates, in particular for RocG. Epistasis often results in “ratchets”, that is, a strict order of mutational events is required to avoid evolutionary steps that result in reduced fitness in enzymes 8 or genetic networks 9. Recent developments in phylogenetic reconstruction software and the lowering costs for custom gene synthesis provide the means to reconstruct ancestral sequences and the likely evolutionary pathways which have led, for example, to the sub‐functionalization of a duplicated gene 10. The present paper thus also paves the way for future studies to further disentangle the intricate relationship between the divergence of paralogs and the regulation of their in vivo activity (here the effect of a metabolite on the assembly of the GDH multimer) versus induction/repression of protein expression by particular metabolites. Future studies into this—or other—systems could involve, for example, the reconstruction of ancestral and intermediate sequences of both GDH paralogs, RocG and GudB, and their respective promoters. These studies may provide an answer to the “chicken‐egg” problem of transcriptional control and sub‐functionalization of novel genes—an important step to elucidate the emergence and evolution of complex biological structures and systems.

Noda‐Garcia et al 1 indeed suggest that the process they describe may be more common than generally thought, in particular when paralogs with no obvious functional divergence persist in more than one closely related species. Therefore, their study also has implications for artificial rewiring of networks of proteins in synthetic biology, since a simple cut and paste of functional genes and regulatory elements may impact genetic stability of artificial signaling networks.

See also: L Noda‐Garcia et al (July 2017)