Download PDF
Synlett 2014; 25(11): 1511-1517
DOI: 10.1055/s-0033-1340183
account
© Georg Thieme Verlag Stuttgart · New York

RNA as an Emergent Entity: An Understanding Gained Through Studying its Nonfunctional Alternatives

Ramanarayanan Krishnamurthy*
Department of Chemistry, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA   Fax: +1(858)7849573   Email: rkrishna@scripps.edu
› Author Affiliations
Further Information

Publication History

Received: 06 January 2014

Accepted after revision: 12 February 2014

Publication Date:
26 March 2014 (online)

 
 PDF Download Permissions and Reprints All articles of this category


Dedicated to Professor Albert Eschenmoser who mentored me in this field of research

Abstract

RNA is arguably the most central member of the class of biomolecules in the extant biochemistry of the earth and, in a number of theories, is postulated to have played a key role in the origin of life. The most prominent among these theories is the ‘RNA world’ theory, which hypothesizes that RNA functioned both as the carrier of information and as a catalyst in many reactions. With new functions of RNA being discovered almost daily, there is much to learn about the operational versatility of this biopolymer. The past–present–future omnipresence of RNA has made it a molecule of intense scrutiny from the perspectives of structure, function, and applications. In this article, I present an understanding of the nuanced interrelationship between the structural components of RNA, and its propensity to undergo self-assembly, an understanding that was not gained by studying RNA, but rather was gained through a comparative study of its nonfunctional alternatives. The results from these juxtapositional studies suggest that RNA might be considered as an emergent structural entity and as a product of chemical evolution.

1 Introduction

2 isoGNA: An Isomer of a Glycerol-Derived Acyclic Nucleic Acid

3 Pentulose Nucleic Acids

4 The Role of the Pent(ul)ofuranosyl Moiety: The ‘A-ha’ Moment

5 Conclusion


#

Key words

chemical evolution - DNA - oligonucleotides - RNA - self-assembly

Biographical Sketch

Zoom Image

Ramanarayanan Krishna­murthy received his B.Sc. in chemistry from Vi­vekananda College (University of Madras), and his M.Sc. in chemistry from the Indian Institute of Technology, Bombay. He obtained his Ph.D. from The Ohio State University, Columbus, Ohio, under the guidance of Professor David Hart; this was followed by postdoctoral work at the Swiss Federal Institute (ETH) Zürich, with Professor Eschenmoser, and as a NASA-NSCORT fellow with Professor Gustaf Arrhenius at the Scripps Institution of Oceanography, UCSD, La Jolla. He then rejoined Professor Eschenmoser at the Skaggs Institute of Chemical Biology at The Scripps Research Institute (TSRI), La Jolla. He is currently an associate professor of chemistry at TSRI.

1

Introduction

Understanding of the structure–function relationship in RNA (and DNA) has been facilitated by many approaches.[1] The classical approach, as exemplified by Watson, Crick, and their colleagues, is to look at the structure as a whole and to relate it to a biological function.[2] The ‘bottom-up’ approach, as exemplified by Westheimer’s method of analyzing Why Nature Chose Phosphates, is to elucidate the function of each part of the polymeric structure.[3] Another strategy is the systematic study of the Chemical Etiology of Nucleic Acid Structure, as delineated by Eschenmoser, which involves asking such questions as, ‘Why pentose and not hexose nucleic acids?’ or ‘Why a pentofuranose and not a pentopyranose?’[4] This methodology entails the synthesis of potentially natural alternatives to RNA that contain, for example, hexo- or pentopyranose sugars in place of ribofuranose. Comparing the relevant chemical properties (e.g., base-pairing capabilities) of such alternative structures with those of RNA or DNA provides insights into the factors that are responsible for the structural uniqueness of the relevant nucleic acid. For example, it was deduced through a study of homo-DNA that the helical structure of DNA is largely due to the furanosyl sugar moiety, rather than the nucleobases or phosphates.[5]

Continuing this line of investigation with questions such as, ‘Why the canonical nucleobases?’ or ‘Why not alternative heterocycles?’[6] has not only led to further insights into the roles played by individual structural elements of RNA and DNA, but has also highlighted the role played by water as a crucial arbiter in the selection of the individual parts of canonical nucleic acids at a functional level.[7]

In this account, I outline recent developments from our laboratory that are based on investigations of two oligonucleotide alternatives: a glycerol-(3′→1′)-linked oligonucleic acid (isoGNA)[8] and pentulose nucleic acid (pentulose-NA)[9]. The first system revealed the unpredictable effects of the backbone on the base-pairing behavior, whereas the second system shed light on the crucial role played by the furanosyl ring in determining the optimal orientation of nucleobases. Such comparative studies, in providing insights into structure–function interrelationships of RNA in an aqueous environment, support the view that the emergence of RNA on the early Earth could have been a consequence of selection by chemical evolution.


# 2

isoGNA: An Isomer of a Glycerol-Derived Acyclic Nucleic Acid

Asking the question, ‘What is the minimal backbone structure that would support a functional informational system?’[10] led us to consider a glycerol-(3′→1′)-linked oligonucleotide (isoGNA), derived by paring back the RNA backbone to its bare-minimum phosphate skeleton (Figure [1]A). On the basis of a qualitative conformational analysis,[11] isoGNA was expected to exhibit base-pairing properties that would differ from those of the known glycerol-based oligonucleotides flexible nucleic acid (FNA),[12] or glycerol nucleic acid (GNA)[13]. For example, isoGNA might adopt other possible conformations (Figures [1]B and 1C) that would permit the system to base-pair not only with RNA, but also with DNA, in contrast to GNA.

Zoom Image
Figure 1 Trimming of RNA gave rise to isoGNA, which is drawn in the following idealized conformations: representation A is derived from RNA and was, therefore, hypothesized to pair with RNA; and representations B and C were thought to be compatible for pairing with DNA.

With such expectations Venkateshwarlu Punna began investigating the base-pairing properties of adenine (A)- and thymine (T)-containing sequences of isoGNA.[8] ­Because he observed no base-pairing of isoGNA at the ­12-mer level with complementary DNA and RNA sequences, further studies were conducted at the 16-mer level. IsoGNA(T)16 and isoGNA(A)16 formed a reasonably strong base-pairing system. However, this duplex was weaker than RNA, slower to form, and exhibited hysteresis. Such behavior can be understood in terms of greater flexibility in the backbone and shortening of interstrand distances (increasing electrostatic repulsion) in isoGNA duplexes in comparison with RNA. The cross-pairing of isoGNA-16mers with complementary RNA or DNA sequences showed dichotomous behavior. The homopyrimidinic isoGNA(T)16 with d(A)18 (or poly-rA) unexpectedly showed no cooperative melting behavior, whereas the inverse combination, homo-purinic isoGNA(A)16 formed duplexes with both complementary DNA (dT18 and poly-dT) and complementary RNA (rU18 and poly-U) sequences. There was no (or minimal) hysteresis, unlike that observed in the all-isoGNA intrasystem pairing. Such asymmetric observations (i.e., disparate base-pairing kinetics and lopsided preferences as to which backbone bears the purines or pyrimidines) for the same set of canonical nucleobases were the first among a series of surprise results, and served as an indication of the control exerted by the backbone on the behavior of nucleobases.

Phaneendrasai Karri, in continuing this project and confirming the previous results, came upon a surprising observation: whereas the homogeneous isoGNA(T)16 and isoGNA(A)16 sequences showed base pairing, the heterogeneous isoGNA-(A,T)-containing sequences, isoGNA-(TA)8 and isoGNA-(T8A8), did not base pair.[8] Moreover, no duplex formation could be discerned with any of the irregular mixed isoGNA-(A,T) 16-mer sequences. The mixed 8-mer as well as 16-mer sequences containing all the four canonical nucleobases (A, T, G, and C) in the isoGNA backbone also showed no base pairing. Such a behavior within a given backbone system, where the homogeneous sequences undergo base pairing but the heterogeneous sequences do not, is unprecedented. Karri found that this unusual behavior extended to the cross-pairing of isoGNA with DNA and RNA: none of the heterogeneous sequences of isoGNA containing any combination of canonical nucleobases (A, T, G and C) cross-paired with corresponding complementary sequences from RNA or DNA.[8] This dichotomous behavior of the heterogeneous isoGNA sequences is ascribed to possible differences in the orientational preferences of the pyrimidine and purine bases with respect to the backbone.

Zoom Image
Figure 2 The base-pairing abilities of the canonical nucleobases vary widely depending on the nature of the backbone to which they are attached. FNA: flexible nucleic acid; PNA: peptide nucleic acid; GNA: glycerol nucleic acid; isoGNA: isomer of glycerol nucleic acid. Self-pairing refers to duplex formation within the same system. Cross-pairing refers to duplex formation with another system (e.g., RNA or DNA).

The observation that isoGNA(Tn) sequences exhibited intrasystem pairing with complementary adenine sequences in isoGNA, but not in DNA or RNA, suggested that ­pyrimidine nucleobases on an isoGNA-backbone might have a preference for the complementary partner tagged to an acyclic backbone. Karri[8] verified this hypothesis by investigating the cross-pairing of isoGNA with acyclic dipeptide backbones tagged with non-cononical recognition elements that we had previously investigated in our laboratory.[6]

Moderate to strong duplex formation was observed for isoGNA(T)n with the corresponding complementary 2,4-diamino-1,3,5-triazine 12- and 16-mers tagged to a peptide Asp-Glu backbone. Strong pairing was observed even at the 12-mer level, in contrast to the pairing results with RNA and DNA.[8] Pairing of isoGNA(T)16 with 2,4-diaminopyrimidine-6-carboxamide 16-mer tagged to an Asp-amAla backbone was also observed. In the reverse combination, isoGNA(A)16 was also found to form duplexes with a 2,4-dioxopyrimidine-tagged Asp-Glu backbone. In the limited studies carried out with two other acyclic backbone systems, PNA and GNA, no interaction with complementary isoGNA sequences was observed.[8]

Overall, the base-pairing properties of isoGNA were markedly different from our expectations and from those of other known acyclic oligonucleotide systems (Figure [2]). The results for the intra- and intersystem base pairing of isoGNA, in combination with other studies,[14] highlight the central role of the backbone in dictating and fine tuning the base pairing capacity and the kinetics of duplex formation within a set of canonical nucleobases. A relative shift in the position of the canonical nucleobases in the acyclic backbone (from the 1′ to the 2′ position) or a variation in the linker length results in an unpredictable variation in base-pairing capabilities, even though these systems contain the same canonical nucleobases and have closely related backbones (Figure [2]).[12] [13] [15] These observations highlight the need for discretion regarding the structural requirements for fabricating a ‘constitutionally simple’ informational system containing the canonical nucleobases.

What do these results mean in terms of understanding the emergence of RNA? Based on the hypothesis that RNA might be a descendant of simpler genetic systems, primitive informational polymers have been sought by changing the backbone while keeping the nucleobases constant.[16] The results described show that, in a prebiotic and origins-of-life context, the search for alternative, generationally simpler, and structurally minimal (simple) informational systems need not be constrained to the set of extant nucleobases as recognition elements. The canonical nucleobases might represent a functional optimum with respect to the ribose–phosphate backbone. In other words, if we permit the backbone to be changed (as, for example, in FNA[12]), we should also permit the nucleobases to be changed to generate a primitive informational system. Such a process allows for the possibility that there might be other potential primordial recognition element sets that could function in conjunction with prebiotically relevant and simpler backbones: this is a field that is full of potential and promise, and which needs further exploration.[6] [17]


# 3

Pentulose Nucleic Acids

While the isoGNA work was ongoing, we had started studies on another potentially natural oligonucleotide, pentulose nucleic acid (pentuloseNA), that we considered relevant to the chemical etiology of nucleic acid structure.[4] In many prebiotic experimental investigations on sugar formation (e.g., the formose[18] or glyoxylate scenarios[19]), keto sugars dominate aldo sugars. This observation led us to consider the possibility of oligonucleotides in which the ribose in the RNA is replaced by ribulose or xylulose to give pentuloseNAs (Figure [3]). Once again, on the basis of a qualitative conformational analysis,[11] we expected these systems to have base-pairing properties. How the systems would compare with RNA would be determined by synthesizing and studying them.

Zoom Image
Figure 3 Idealized conformational representation of pentulose nucleic acid, derived from the A-conformation of RNA.[11] Note the direction of the Watson–Crick hydrogen-bonding face (indicated by the arrow).

The project began with Geeta Meher, who established a reliable synthetic route to the expensive and suitably derivatized ribulose sugar on a large scale to permit the preparation of the thymidine and adenine ribulose nucleosides.[9] Matthias Stoop joined the project and applied these protocols to the preparation of the corresponding xylulose-derived nucleosides, and later collaborated with Phaneendrasai Karri.[9] During the synthesis of the building blocks and the oligomers, they encountered many barriers (inefficient processes and purification problems) that almost led us to abandon the project. Despite these difficulties, all three chemists persevered and powered the project through to produce adequate quantities of the A- and T-containing oligonucleotides in the (1′-3′)-linked-ribulo and (1′-3′)- and (4′-3′)-linked xylulo series (Figure [4]).

Zoom Image
Figure 4 The three pentuloseNAs synthesized and investigated with respect to their base-pairing properties (B = nucleobase).

After all these laborious efforts, the base-pairing properties of these ribulo- and xylulo-oligonucleotides turned out to be terribly disappointing; neither self pairing among the pentuloseNAs nor cross-pairing with complementary RNA/DNA was observed. Notwithstanding the disappointment, the unexpected lack of pairing gave rise to a conclusion that gained significance when placed in the context of chemical evolution: even though pentulose-sugars might have been predominant in many prebiotically plausible scenarios, the pentulose oligonucleotides (formed potentially by the same pathways that RNA was produced from ribose) would not be capable of forming an informational system sufficiently competent to interfere with the emergence of RNA. In other words, the pentuloseNAs are a dead end. Such a conclusion is ironic (and bittersweet for us) when considered in the context of modern biochemistry; after all, ribose is formed by conversion of ribulose and xylulose! Ribulose- and xylulose-5-phosphate are central products of the pentose phosphate pathways; they are produced first and converted into ribose-5-phosphate, which then becomes the starting point for the synthesis of RNA building blocks.

Zoom Image
Figure 5 (A) Corrected idealized conformation of pentuloseNA (based on X-ray data for the monomeric nucleosides) in which the nucleobases are in the anti-conformation, leading to (B) the intensification of electrostatic interactions within a duplex of pentuloseNA. For definitions of R1 and R2 see Figure [3].

The biggest windfall from this project came, however, when we started looking into the reasons for the failure of the pentulose nucleic acids to form a base-pairing system. Closer inspection of the models, based on information from x-ray analysis of the monomeric ribulose and xylulose nucleosides, showed clearly that the interstrand phosphate distance was much shorter in a pentulose-NA duplex structure than in RNA. Although this is a consequence of having the phosphate backbone on the same side as the nucleobase on the pentulose sugar framework (Figure [5]A), there was an important aspect that we had overlooked from the very beginning of this project. In our qualitative conformational and molecular model analysis, we had placed the nucleobase in a syn-conformation (over the pentulose-ring; Figure [3]) in constructing the duplex, resulting in interstrand distances that were similar to those in RNA. However, the x-ray structures of the monomers clearly showed that the anti-orientation of the nucleobases is the preferred conformation. When the duplex was constructed with the nucleobases in an all-anti-orientation, the interstrand distances were reduced (Figure [5]). Thus, in addition to having the sugar phosphate and the nucleobases on the same side of sugar scaffold, the anti-orientation of the nucleobases brings the two backbones (in the duplex) even closer, intensifying the phosphate–phosphate electrostatic repulsion (Figure [5]B).

We hypothesized that the presence and the rigidity of the pentulose ring structure prevented any adjustment necessary to increase the interstrand distance while producing a conformationally repetitive structure capable of base pairing. It was then only logical to ask what would happen if the pentulose ring structure were to be removed. Could base pairing be restored in such a ring-less system? This thought process led to the fortuitous discovery of the connection between the results from the isoGNA project and those from the pentuloseNA project, and it shed light on an important role played by the furanosyl ring structure in RNA.


# 4

The Role of the Pent(ul)ofuranosyl Moiety: The ‘A-ha’ Moment

Removal of the pentofuranosyl ring structure in RNA has a long history, rooted in origins-of-life research. The resulting flexible nucleic acids FNAs [also known as unlocked nucleic acids (UNAs) or acyclic nucleic acids (ANAs)] were considered by Joyce et al. in arguing for a structurally simpler ancestor of RNA.[12a] They hypothesized that removing the furanosyl (ring) structure would result in an acyclic nucleic acid with a glycerol phosphate backbone that would not display the shortcomings presented by RNA with respect to chemical evolution. However, experimental investigations showed that this flexible nucleic acid has extremely poor or nonexistent base-pairing properties.[12b] [c] Therefore, removal of the furanosyl ring structure from RNA gave rise to an acyclic oligonucleotide in which the base-pairing capacity was almost eliminated (Figures [6]A and 6B). This observation highlights the role played by the pentofuranosyl structure in organizing the backbone of RNA in a manner conducive to base pairing.

Corresponding removal of the pentulofuranosyl ring structure in pentuloseNA (from either the ribulo or xylulo series) leads to an acyclic system that has the same glycerol backbone as that derived from RNA, but with the difference that the nucleobases are directly connected to the glycerol backbone instead of to a flexible linker (Figures [6]C and 6D). This turned out to be another isomer of glycerol nucleic acid (isoGNA), which we already had investigated and shown to be a (limited) base-pairing system.

Zoom Image
Figure 6 A comparative analysis between RNA and pentuloseNA when the ring structure is removed from the parent compound (B = nucleobase; R = H, Me).

Juxtaposing these two results proved fruitful: removal of the furanosyl ring structure in pentuloseNA converted a nonpairing system (pentuloseNA) into a comparatively stronger pairing system (isoGNA). This is exactly the opposite of what is observed in RNA when the furanosyl ring structure is removed, where a base-pairing system (RNA) is converted into a weak or nonpairing acyclic system (FNA). These drastic and contradictory effects of removing the furanosyl ring component in RNA and in pentuloseNA drew attention to the specific role of the pent(ul)ofuranosyl moiety in enabling or disabling base pairing in a given system (Figure [6]).

Closer inspection revealed that the removal of the ring in ribulose- and xylulose-NAs eliminates the restriction imposed on the strict anti-conformational preference of the canonical nucleobases. In the absence of the furanosyl ring structure, the nucleobases can swing the other way (in the isoGNA). Such a disposition of the nucleobase results in an arrangement in which the Watson–Crick hydrogen-bonding face of the nucleobases point in the opposite direction, which is, importantly, away from the sugar phosphate silhouette. This arrangement permits greater interstrand phosphate distances in the isoGNA duplex than those in the duplex of the parent pentulose-NA, enabling base pairing in isoGNA.[20] Removal of the ring structure in RNA results in an acyclic system with a floppy linker that is unable to accommodate the structural requirements and entropic price to adopt a regular repetitive structure that is capable of forming duplexes through base pairing.

Thus, by comparing RNA and FNA with pentuloseNA and isoGNA, we were able to home in on another important role of the furanosyl ring in RNA: it restricts the rotation of the canonical base, constraining the nucleobase to the preferred anti-conformation with respect to the ring structure. This precise positioning results in the Watson–Crick base-pairing face of the nucleobase pointing away from the sugar-phosphate backbone, ensuring an optimal and maximal interstrand distance of the two phosphate backbones in a duplex. Changing this interstrand distance affects this optimum in one of two ways: (a) shortening this distance results in duplexes that are weaker than in RNA, resulting in a system that is compromised and limited in its base-pairing capability, whereas (b) increasing the distance strengthens the duplexes in comparison with in RNA, resulting in systems that are much less sensitive to nucleobase mismatches, leading to inaccurate transfer of information.[7] Therefore, the pivotal positioning of the phosphate and purine/pyrimidine nucleobases is ensured by this furnanosyl ring, and this arrangement enables RNA/DNA to function (by base pairing) in an aqueous environment at near-neutral pH values and ambient temperatures.


# 5

Conclusion

RNA can be regarded as an emergent structure with emergent properties. The relation between the structure of RNA and its function is manifested in every part and at every level of its structural augmentation. The critical dependence of the function of RNA on the nature of its parts and components (ribose in a furanosyl versus a pyranosyl form; nucleobases in the keto versus enol form) and the way they are put together (nucleobases at the anomeric position as the β-anomer; 3′,5′- versus 2′,5′-linkages, and purine–pyrimidine versus pyrimidine–pyrimidine or purine–purine base pairs) indicates that RNA is a chemically refined structure. In other words, the properties of RNA are expressed only when all the requisite components, in their proper forms, are put together in the proper combination. Any change or deviation from this ideal organization and connectivity will undermine the optimal function of RNA.

Zoom Image
Figure 7 RNA as an emergent entity with emergent properties (in an aqueous medium).

It is illuminating to dissect the incremental emergent properties of RNA at each level of complexity. For example, at the sugar-phosphate level, the 5′-phosphate ensures that only the ribofuranosyl form is available (preventing the existence of pyranose forms). At the next level, whereas ribose or ribose-5-phosphate is highly labile under prebiotic conditions[21] and the nucleobase is insoluble in water, the nucleoside/nucleotide combination is not only water soluble, but also more stable. Thus the nucleoside/nucleotide combination has inherent advantageous properties in an aqueous medium that are absent in its individual components. Going one level higher, these individual nucleosides/nucleotides are unable to self-assemble by hydrogen-bond recognition in an aqueous environment.[22] However, when these nucleotides are linked together by either 2′,5′- or 3′,5’-connectivity (e.g., at the hexamer to octamer stages), complementary hydrogen-bond-mediated self-assembly (driven by the collaborative and collective hydrophobicity of the optimized purine–pyrimidine combination) is manifested even at micromolar to nanomolar concentrations in water. Furthermore, the importance of phosphates (as the phosphodiester linker) is wholly apparent only at the level of the polymer.[3] The attributes of RNA are fully manifested only in the completely assembled structure, and they are not exhibited by its individual components. Finally, and most importantly, all these properties of RNA are possible only because of water: RNA is optimized for its function in an aqueous environment (Figure [7]).

Such incremental structural development and progressive manifestation of properties (not present at the levels of, or fully expressed by, the components) in aqueous environs argues strongly for RNA to be viewed as an emergent entity with emergent properties.[23] It is possible that this optimal structure was selected from a library of structures (combinations of various backbones, linkers, nucleobases, and connectivities) on the basis of its optimal function. Such a view is also compatible with RNA as an emergent entity.


#
#

Acknowledgments

I am grateful to all my co-workers who persevered despite obtaining ‘negative’ results. The work described here was supported jointly by NSF and the NASA Astrobiology Program under the NSF Center for Chemical Evolution (Grant CHE-1004570) and NASA Astrobiology: Exobiology and Evolutionary Program (Grants NNX07AK18G and NNX07AK96G). I thank Professors Albert Eschenmoser, Gerald Joyce, and Donna Blackmond for their insightful reading of, and constructive feedback on, this manuscript.



   Permissions and Reprints All articles of this category
Zoom Image
Zoom Image
Figure 1 Trimming of RNA gave rise to isoGNA, which is drawn in the following idealized conformations: representation A is derived from RNA and was, therefore, hypothesized to pair with RNA; and representations B and C were thought to be compatible for pairing with DNA.
Zoom Image
Figure 2 The base-pairing abilities of the canonical nucleobases vary widely depending on the nature of the backbone to which they are attached. FNA: flexible nucleic acid; PNA: peptide nucleic acid; GNA: glycerol nucleic acid; isoGNA: isomer of glycerol nucleic acid. Self-pairing refers to duplex formation within the same system. Cross-pairing refers to duplex formation with another system (e.g., RNA or DNA).
Zoom Image
Figure 3 Idealized conformational representation of pentulose nucleic acid, derived from the A-conformation of RNA.[11] Note the direction of the Watson–Crick hydrogen-bonding face (indicated by the arrow).
Zoom Image
Figure 4 The three pentuloseNAs synthesized and investigated with respect to their base-pairing properties (B = nucleobase).
Zoom Image
Figure 5 (A) Corrected idealized conformation of pentuloseNA (based on X-ray data for the monomeric nucleosides) in which the nucleobases are in the anti-conformation, leading to (B) the intensification of electrostatic interactions within a duplex of pentuloseNA. For definitions of R1 and R2 see Figure [3].
Zoom Image
Figure 6 A comparative analysis between RNA and pentuloseNA when the ring structure is removed from the parent compound (B = nucleobase; R = H, Me).
Zoom Image
Figure 7 RNA as an emergent entity with emergent properties (in an aqueous medium).