A theoretical perspective of the physical properties of different RNA modifications with respect to RNA duplexes

Highlights

• •This study represents a direct comparison of the physical properties of different RNA modifications, and provides an insight into their functions of forming and maintaining RNA duplexes.

Abstract

Epitranscriptomic variations include >140 different RNA modifications, many of which can serve as disease biomarkers.

Owing to the challenges on synthesizing modified RNA oligos, majority of earlier studies on the effects of RNA modifications to RNA duplexes focused on selected individual epitranscriptomic variation. There are also limited development on the computational modeling of RNA duplexes containing a specific epitranscriptomic variation.

This study aims to theoretically estimate the physical properties of different modified ribonucleosides and compare their variations with respect to altering the molecular structure of an RNA duplex.

With only four canonical ribonucleotides, short RNA sequences within a transcript or between two different transcripts can be complementary to each other. Thermodynamically, it is favorable for two complementary RNA sequences to form an RNA duplex, which is an essential component in many different RNA structures and functions [1]. In general, the RNA duplex structure is maintained by the hydrogen bonds between the Watson-Crick base pairs, and the hydrophobic interactions (or base stacking) between the adjacent nucleobases along the double stranded region [2]. The RNA duplexes can be further stabilized by adopting a helical structure similar to the double-stranded DNA structure. With the helical structure, the stack of base pairs is twisted right-handed, and each base pair is approximately 30° offset from another base pair. Also, by adopting the helical structure, the major and minor grooves are created, which facilitate the RNA to be recognized and bound with RNA-binding proteins [2]

Since the discovery of the first RNA modification, pseudouridine, in 1951, more than 140 additional RNA modifications have been identified [3], [4], [5]. With the possibility to reverse the RNA modifications and regenerate the unmodified RNA, epitranscriptomic variations have been considered as an extra mechanism of post-transcriptional regulation of RNA-dependent cellular activities [6], [7], [8]. Independently, specific RNA modifications and modified RNA have been associated with diseases [9]. For these reasons, there are growing interests in studying the effects of RNA modifications on the molecular interactions and stability of RNA duplexes that contain specific RNA modifications. However, due to the challenges on synthesizing modified RNA oligos, the majority of the earlier studies on the effects of RNA modifications on RNA duplexes were limited to a single RNA modification per study [10], [11], [12], [13]. Since different experimental designs or approaches were used in the earlier studies, the available information may not allow us to compare the effects of different RNA modifications. Furthermore, the effects of many other RNA modifications have not been experimentally determined yet. There are also limited development from using the computational modeling to simulate the RNA duplex structures that contain specific RNA modifications. In this study, we aim to use a theoretical approach to estimate the physical properties of many different modified ribonucleosides and compare their variations with respect to the RNA duplex structure. The results from this study can be used as a guide for determining which specific RNA modifications in an epitranscriptomic profile should be given higher priority in future investigations, because of their relative large disruption to RNA duplexes.

 Among all the RNA modifications, some of them involve modifications at more than one position within the same ribonucleoside structure. To simplify our study, we focused on the RNA modifications that have modified the nucleobase structures at only one specific position. Despite the limitation of our selection, more than half of all the known RNA modifications in prokaryote​s​ and eukaryotes are included in this study. To the best of our knowledge, this study represents the largest set of RNA modifications that are compared side-by-side with each other.

Depending on whether the RNA modification is located at a position that would block one of the hydrogen bonds in the Watson-Crick base pairing, the selected RNA modifications are divided​ ​into two groups. The first group of RNA modifications​ ​does not directly block any ​hydrogen bonds in the Watson-Crick base pairs. Whereas,​ ​the second group of RNA modifications (boxed in Fig 1) prohibits the formation of a hydrogen bond in the Watson-Crick base pairing. For more details, see the method section in supplementary material. As shown in Fig 1, majority of the selected RNA modifications do not directly block the hydrogen bonding between the Watson-Crick base pairs, which in turn support the initial nucleation step for forming the RNA duplexes.

Fig. 1.

The normalized relative variation of hydrophobicity (RVH) of selected RNA modifications (bar chart) plotted with the percent increase in molecular size (triangle) of the corresponding Watson-Crick base pair upon modification. The name of each RNA modification is abbreviated as in the Modomics database. All RNA modifications listed pertain to a single modification located at a specific position within the nucleobase structure. The boxed modifications represent those prohibit the hydrogen bonding in the Watson-Crick base pairing. A. Modifications that arise from adenine base. B. Modifications that arise from uracil base. C. Modifications that arise from cytosine and guanine bases. D. Formulas used to calculate the normalized RVH and% increase in molecular size. * represents the reported value was divided by 10.

From the perspective of forming the RNA duplex and its stability, the hydrophobic​ ​interactions between the adjacent nucleobases along each of the strands within the duplex have reported to play a bigger role than the hydrogen bond​s​​​ ​between the Watson-Crick base pairs [14]. For this reason, we calculated the ​relative variation of ​hydrophobicity (RVH) of a​ll​ selected RNA modifications by relating the Log P value of each modified ribonucleoside to the Log P value of its corresponding unmodified ribonucleoside. Since the pKa data of many modified ribonucleosides are unavailable, our estimation on the Log P values had excluded any possible protonated or deprotonated ​ribonucleoside structures. In order to compare the RVH of different modified ribonucelosides that have very different molecular sizes, the ​RVH is normalized to the increase in molecular​ ​mass from its corresponding unmodified ribonucleoside (Fig 1D). If the hydrophobicity of a specific modified ribonucleoside is higher than its corresponding unmodified ribonucleoside, the normalized RVH would have a positive value and vice versa. The normalized RVH values range from −35 to 15. All the normalized RVH values are presented in Fig 1A-C in the alphabetical order.

Among all the normalized RVH values, inosine (I) has the biggest decrease in hydrophobicity (Fig 1A). With respect to the stability of RNA duplexes, this indicates the inosine modification has the most destabilizing effect among all the selected RNA modifications. Our result does comply with the report from Znosko, in which an I-U base pair was found to be less stable than an A-U base pair by ∼2.3 kcal/mol [10]. For the most frequent modification of adenosine, namely N6-methyladenosine (m6A), the positive normalized RVH value indicates that the m6A modification would enhance the base stacking and stabilize the RNA duplex. However, the m6A modification increases the size of an m6A-U base pair (Fig 1A), which can potentially disrupt the helical structure of an RNA duplex. According to a recent report from Kool and his associates, the m6A modification would destabilize RNA duplexes and lower their melting temperature [11]. Therefore, if the normalized RVH value and the% increase in size of a specific RNA modification in Fig 1A-C have indicated the opposite effect on the stability of an RNA duplex, more cautious are needed to evaluate the results. In view of this situation, the results of another common example of RNA modification, namely 5-methyluridine (m5U), were evaluated. In terms of the% increase in size as defined in Fig 1D, the m5U modification is exactly the same as m6A modification. However, the normalized RVH value of m5U is higher than that of m6A. Therefore, based on our results, the m5U modification would stabilize the RNA duplex. According to an earlier report from Kool and his associates, the m5U modification did increase the stability of RNA duplexes and their melting temperatures [12]. To further evaluate our results in Fig 1, the results of pseudouridine (Y) modification were also evaluated. In the case of Y modification, there is no change in its size and hydrophobicity when comparing with uridine. Therefore, based on our results, the Y modification would not have any significant effect on the stability of RNA duplexes, which matches the findings reported by Kierzek and his associates[13]

Theoretically, the larger the chemical group is introduced by the RNA modification, the bigger the change in the resulting molecular structure. For each selected RNA modification, the% increase in size to the complementary base pair was calculated as shown in Fig 1D. It is important to note that the additional chemical group that corresponds to a specific RNA modification can be positioned either inside or outside the helical structure of an RNA duplex. For instance, in the case of m1A modification, the methyl group can be positioned between the m1A-U base pair, and pushes the two RNA backbones away from each other. As a result, the helical structure can be distorted with a bulge. Whereas, in the case of m6t6A modification, the larger threonylcarbamyol group ​can be positioned outside the helix along the major groove. Overall, the% increase in size ranges from 0% to ∼30% (Fig 1A-C).

In summary, we determined majority of the known RNA modifications that introduce a single modification do not directly interfere with the hydrogen bonding between the Watson-Crick base pairs, thus allowing the formation of RNA duplexes. Also, we demonstrated the effects of specific RNA modification on the stability of RNA duplexes could be estimated by considering the relative variation in hydrophobicity and the increase in molecular size that result from the RNA modification. Based on the existing knowledge in nucleic acid biochemistry, more accurate estimation on the effects of RNA modification on the stability of RNA duplexes can be achieved if the position of the modified ribonucleoside within an RNA duplex as well as the adjacent RNA sequences are taken into consideration [15]. Nevertheless, the results in Fig 1 represent a direct comparison of the physical properties of different RNA modifications and provide an insight into the variability of their functions in a biological system.

Declaration of Competing Interest

All authors wish to declare no conflict of interest.