RS and RGG repeats as primitive proteins at the transition between the RNA and RNP worlds

Abstract

For many experimental biologists in the field of nuclear cell biology, low complexity repeats in nuclear proteins constitute a nuisance. They are difficult to express, impossible to crystalize and have low but near ubiquitous unwanted affinities toward many biomolecules. Examples of such nuclear protein repeats are RS (Arg-Ser) repeats in splicing factors, RGG (Arg-Gly-Gly) repeats in hnRNP proteins and FG (Phe-Gly) repeats in nuclear pore components. Here, I would like to present a more positive perspective for at least a subset of repeats and suggest that they are excellent candidates to have constituted the first proteins emerging from an RNA world.

Keywords: :

• chaperones
• early evolution
• RGG domains
• RNA world
• RS domains
• SR proteins

The RNA world hypothesis supposes forms of life that use RNA for both coding and catalytic activity.^1,2 It has been proposed that primordial proteins functioned as RNA chaperones, aiding RNA folding.³ RS and RGG domains are low in complexity, predicted to be unstructured,⁴ subject to simple regulation and directly involved in RNA metabolism. Proteins that contain simple unstructured repeats have been found to act as RNA chaperones^5,6 and the intrinsically unstructured regions of several RNA chaperones have been proposed to possess such chaperone activity.^4,5 RS domains consist of several (5–20) copies of the RS dipeptide, irregularly interrupted by other amino acids. RS domains can directly bind RNA and themselves^7,8 in a rather non-specific manner. Mutagenesis analysis indicated that the exact number and spacing of these repeats are not important for their function.⁹ RS domain self-association is promoted by phosphorylation at serine residues.^7,8 Initially, affinity for single stranded RNA was found to be reduced by SR phosphorylation,^7,8 presumably by ionic repulsion between the negatively charged RNA and phosphoserines. More recent data however indicates that RS domains also possess an intrinsic affinity for double stranded RNA, and that their function in splicing required serine phosphorylation,¹⁰ possibly to properly discriminate between single stranded and double stranded RNA. The RS domain is predicted to be unstructured¹¹ and found mainly in proteins involved in pre-mRNA splicing. However, consistent with an early role in RNA metabolism, RS domains are also present in proteins that function in other sectors of RNA metabolism such as transcription elongation, polyadenylation, translation^12,13 and microRNA processing.¹⁴ Analogous to the current role of RS domains in splicing,^10,15 a function of RS domains in RNA-based forms of life may have been to promote base-pairing between different RNAs that subsequently were transesterified by RNA-based catalysis.

Arguing against an early function is that so far RS domains have been mainly identified in (higher) eukaryotes. However, their possible absence in bacteria and archea may reflect that r-selected prokaryotes have replaced several RNP-catalyzed processes by the more efficient protein-only catalysis,¹⁶ limiting the need for this type of RNA chaperone activity. In that sense budding yeast may represent an intermediate in metabolic advancement with only one clear RS-domain splicing factor Npl3.¹⁷ Advantages of maintaining the less efficient RNP catalysis in higher eukaryotes may include multilevel control or higher combinatorial potential. Also, the inefficiencies may not be rate limiting and even trivial compared with other activities such as transcribing large introns and maintaining a largely non protein coding genome.

RGG domains, irregular repeats of arginine-glycine-glycine, are present in RNA binding proteins such as hnRNP proteins.¹⁸ RGG domains directly interact with RNA and are also predicted to be unstructured.⁴ They are subject tova rather unusual post-translational modification, methylation of arginine. Importantly, similar to phosphorylation of RS repeats, arginine methylation may be reversible¹⁹ and methylated RGG domains are altered in their RNA binding.²⁰

Several predictions arise from this view of simple repeat proteins, one of which is that RS serine ribozyme kinases and phosphatases and RGG arginine ribozyme methyltransferases and demethylases must have existed, remnants of which may still be present in modern day organisms. Interestingly, RNA control of serine phosphorylation has been found in the C-terminal domain (CTD) of RNA polymerase II,²¹ where the non-coding RNA 7SK functions as an inhibitor of the CTD serine 2 kinase P-TEFb. The CTD consists of ~20–50 copies of a seven amino acid repeat, is intrinsically unstructured and lacks catalytic activity. The consensus repeat sequence is Tyr-Ser-Pro-Thr-Ser-Pro-Ser, but non-consensus repeats of the CTD contain two arginine and seven lysine substitutions. Notably, the arginines are modified by site-specific methylation²² which plays a role in snRNA and snoRNA expression. Possibly, the origins of the CTD may also predate protein-based catalysis. Another prediction is that some RS or RGG repeats may function in RNA metabolism in prokaryotes. Lastly, these simple repeat proteins can conceivably be produced without very precise translation machinery and with an incomplete genetic amino acid code. In this light it is interesting that arginine strongly and specifically binds to its own codons in RNA.²³

In summary, simple repeat protein domains should not or not only be viewed by nuclear cell biologists as troublesome regions meant to be masked in database searches, but as interesting protein domains that may shed light on early cellular evolution.

Acknowledgments

I’d like to thank Karla Neugebauer for discussions and collegues for encouragement to submit this manuscript, versions of which have been idly spinning on disks for many years.