Tertiary motifs revealed in analyses of higher order RNA junctions

Abstract

RNA junctions are secondary structure elements formed when three or more helices come together. They are present in diverse RNA molecules with various fundamental functions in the cell. To better understand the intricate architecture of three-dimensional RNAs, we analyze currently solved 3D RNA junctions in terms of basepair interactions and three-dimensional configurations. First, we study basepair interaction diagrams for solved RNA junctions with five to ten helices and discuss common features. Second, we compare these higher-order junctions to those containing three or four helices and identify global motif patterns such as coaxial-stacking and parallel and perpendicular helical configurations. These analyses show that higher order junctions organize their helical components in parallel and helical configurations similar to lower order junctions. Their sub-junctions also resemble local helical configurations found in three and four-way junctions, and are stabilized by similar long-range interaction preferences such as A-minor interactions. Furthermore, loop regions within junctions are high in adenine but low in cytosine. And, in agreement with previous studies, we suggest that coaxial stacking between helices likely forms when the common single stranded loop is small in size; however, other factors such as stacking interactions involving non-canonical basepairs and proteins can greatly determine or disrupt coaxial stacking. Finally, we introduce the ribo-base interactions: when combined with the along-groove packing motif, these ribo-base interactions form novel motifs involved in perpendicular helix-helix interactions. Overall, these analyses suggest recurrent tertiary motifs that stabilize junction architecture, pack helices, and help form helical configurations that occur as sub-elements of larger junction networks. The frequent occurrence of similar helical motifs suggest Nature’s finite and perhaps limited repertoire of RNA helical conformation preferences. More generally, studies of RNA junctions and tertiary building blocks can ultimately help in the difficult task of RNA 3D structure prediction.

INTRODUCTION

RNA molecules adopt well-defined three-dimensional structures of high complexity to perform specific functions in the cell. These complex architectures piece together basic secondary structural elements such as helices, hairpins, internal loops and junctions, which bind together via tertiary interactions to form compact structures of active RNAs.

An RNA junction can be defined as the point of connection between different helical segments^{1; 2} (Fig. 1A). This secondary structure arrangement is present in a wide range of RNA molecules and is involved in a variety of different functional roles, including the self-cleaving catalytic properties of the hammerhead ribozyme³, promotion of functional folded states of the hairpin ribozyme⁴, recognition of the binding pocket domain by purine riboswitches^{5; 6}, and translation initiation of the HCV virus at the internal ribosome entry site (IRES)⁷. Several junctions also occur within ribosomal RNA subunits^{8; 9; 10} where they play important roles and often bind to ribosomal proteins¹¹. While more is known about other secondary structure elements such as hairpins and internal loops¹², our current understanding of the more complex junction element, especially higher order junctions, is limited. Because junctions serve as major architectural features in RNA, it is essential to better understand structural, energetic, and dynamic aspects of these elements.

Figure 1.

Junction architecture for E. Coli 23S rRNA (2AW4_2073 from Table 1). (A) Secondary structure diagram of the six-way junction element composed of six helices labeled H₁ to H₆ (color-coded) and six loop regions labeled J_1/2 to J_6/1 with nucleotide positions marked in black. Helices and loop regions are labeled uniquely according to the 5′ to 3′ orientation of the entire RNA structure, by labeling H₁ as the first helix encountered while entering the junction in the 5′ to 3′ direction; subsequent helices within the junction are labeled as one moves along the nucleotide chain in the same direction. Lines inside the helices represent the canonical WC basepairs G-C, A-U, and the G-U wobble basepair. (B) Network interaction diagram representing basepairs of the same six-way junction according to the Leontis-Westhof symbology³⁵. (C) 3D representation diagram containing a coaxial stacking between helices H₁ (red) and H₂ (blue).

RNA crystallography, NMR, and other experimental techniques such as fluorescence resonance energy transfer (FRET) and small-angle X-ray scattering (SAXS) have offered unprecedented opportunities to analyze RNA tertiary (3D) structure^{10; 13; 14; 15; 16; 17; 18}. Such views have revealed structural properties of junctions such as coaxial stacking of helices and long-range tertiary interactions^{19; 20; 21; 22} (see Fig. 1B-C). For instance, Lilley et al.^{23; 24; 25} analyzed the conformations of specific examples of three-way and four-way junctions (junctions composed of three and four helical arms respectively) in DNA and RNA using FRET and observed transitional changes and flexibility in their helical configuration under Mg²⁺ and Na⁺ concentration variations. Lescoute and Westhof²⁶ compiled and analyzed the topology of three-way junctions in folded RNAs, categorizing these junctions into three families and specifying rules to predict coaxial stacking, which occurs when two separate helical regions stack to form coaxial helices as a pseudo-continuous helix (see Fig. 1C). The loops connecting the stacked helices constrain the configuration space that these helical axes can explore. Laing and Schlick²⁷ analyzed the topology of solved 3D four-way junctions and grouped them into 9 families according to coaxial stacking interactions and helical conformation signatures. Tyagi and Mathews²⁸ predicted coaxial stacking based on free energy minimization. Finally, Bindewald et al.²⁹ developed RNAJunction, a database which contains information on RNA structural elements including junctions.

These combined analyses of RNA structures have unraveled recurrent structural motifs across a variety of RNA molecules. Our previous work on annotation and analysis of RNA tertiary motifs²² based on a representative set of high-resolution RNA structures showed that coaxial helices are abundant tertiary motifs that often cooperate with other long-range interactions such as A-minor motifs to stabilize RNA’s structure. Building upon existing work on three-way and four-way junctions^{26; 27}, we extend the analysis here to higher order junctions (five to ten-way junctions) and combine our findings to describe common motifs, including recurrent helical configurations, occurring across all junctions found in solved structures, regardless of their degree of branching. Our analysis reveals novel interaction motifs formed between perpendicular alignments of helices as well as common internal basepairs that help form long-range interactions. We also discuss how junctions arrange their helical arms in similar configurations, regardless of their degree of branching. Statistical data showing basepair and base stacking preferences are also reported.

RESULTS

Network interaction diagrams (see Fig. 1B) indicating basepair interactions have proven useful in understanding RNA tertiary motifs^{30; 31; 32} and in investigating the topology of three and four-way junctions^{26; 27}. Here, we extend such analyses to higher order junctions from degrees 5 to 10. We begin with a description of the higher order junctions using network interaction diagrams. For clarification, we label and color code helices sequentially according to the 5′ to 3′ orientation of the entire RNA as shown in Fig. 1A. A helix here is required to contain at least two consecutive Watson-Crick (WC) basepairs G-C, A-U and G-U. The single stranded region between each pair of consecutive helices H_i and H_i+1 is labeled by J_i/i+1. Each junction element is labeled by its PDB code³³ followed by the first residue number of the first helix H₁ in the junction. The point where strands cross over is called the point of strand exchange. We use the Leontis and Westhof notation^{34; 35} to study basepair interactions occurring within junctions and to describe common motifs. Our list of 207 junctions contains junctions of degree 3 to 10 (see Tables 1, S1 and S2) and has been assembled by taking all high-resolution RNA structures from the PDB database³³ as of April 2009.

Table 1.

List of RNA 3D structures containing twenty three five-way junctions, nine six-way junctions, four seven-way junctions, one nine-way junction, and two ten-way junctions. The name describes the PDB code and the number of the first residue of helix H₁ in the junction. The nomenclature is based on ¹ and the helices are numbered according to the scheme in Leffers et al⁵². Single line between rows separates junctions with the same number of coaxial helices. Double line between rows separates the junction’s degree of branching.

Name	Degree	RNA type	Coaxial stacks	Helical alignments	Nomenclature	Domain	Helix numbers
1U6B_8	5WJ	GI intron Azoarcus	H1H2, H3H4		HS3HS32HS13HS2		P2-3-8-7.1-10
1Y0Q_43	5WJ	GI intron Twort	H2H3, H4H5		HS42HS10HS3HS5		P3-4-6-7.1-7.2
1S72_238	5WJ	23S rRNA H. Marismortui	H1H2, H3H5		HS4HS5HS6HS6HS2	I	H14-15-16-21- 22
2J01_267	5WJ	23S rRNA T. Thermophilus	H1H2, H3H5		HS4HS5HS5HS6HS2	I	H14-15-16-21- 22
2BTE_6	5WJ	Leuyl-tRNA T. Thermophilus	H1H5, H2H3		HS23HS2H		H1-2-3-4-5
2NR0_6	5WJ	Leucyl-tRNA E. Coli	H1H5, H2H3		HS2HS4HS1HS2H		H1-2-3-4-5
2J01_45	5WJ	23S rRNA T. Thermophilus	H2H5, H3H4		HS6HS42HS1HS1	I	H4A-5-8-9-10
2AW4_46	5WJ	23S rRNA E. Coli	H2H5, H3H4		HS6HS42HS1HS1	I	H4A-5-8-9-10
1NKW_45	5WJ	23S rRNA D. Radiodurans	H2H5, H3H4		HS6HS42HS1HS1	I	H4A-5-8-9-10
3BWP_23	5WJ	GII intron O. Iheyensis	H2H3		HS3HS2HS2HS3HS3		IA-IB-IC-ID1- I(ii)
2J00_35	5WJ	16S rRNA T. Thermophilus	H3H4		HS2HS2HS4HS6HS2	5′	H3-4-16-17-18
2AVY_35	5WJ	16S rRNA E. Coli	H3H4		HS2HS2HS5HS7HS2	5′	H3-4-16-17-18
1NKW_592	5WJ	23S rRNA D. Radiodurans	H4H5		HS4HS2HS22HS5	II	H26-27-32-36- 46
1S72_640	5WJ	23S rRNA H. Marismortui	H4H5		HS4HS2HS2HS4HS10	II	H26-27-32-36- 46
2AW4_583	5WJ	23S rRNA E. Coli	H4H5		HS4HS2HS2HS2HS8	II	H26-27-32-36- 46
2J01_583	5WJ	23S rRNA T. Thermophilus	H4H5		HS4HS2HS2HS2HS8	II	H26-27-32-36- 46
1S72_657	5WJ	23S rRNA H. Marismortui	H4H5		HS2HS3HS5HS1HS3	II	H27-28-29-30- 31
2AVY_57	5WJ	16S rRNA E. Coli	H4H5	H2H3	HS2HS6HS1HS1HS3	5′	H5-6-7-13-14
2J00_56	5WJ	16S rRNA T. Thermophilus	H4H5	H2H3	HS3HS6HS1HS1HS4	5′	H5-6-7-13-14
1NKW_2036	5WJ	23S rRNA D. Radiodurans	H2H3	H4H5	HS9HS8HS11HS7HS8	V	H73-74-89-90- 93
1S72_2097	5WJ	23S rRNA H. Marismortui	H2H3	H4H5	HS6HS8HS9HS4HS4	V	H73-74-89-90- 93
2AW4_2056	5WJ	23S rRNA E. Coli	H2H3	H4H5	HS6HS8HS9HS4HS4	V	H73-74-89-90- 93
2J01_2053	5WJ	23S rRNA T. Thermophilus	H2H3	H4H5	HS9HS8HS11HS7HS8	V	H73-34-89-90- H 93
2A64_20	6WJ	RNase P type B B. Stearothermophilus	H1H2, H5H6		HS1HS13HS3HS32HS8		P2-3-5-15-15.1- 15.2
1NKW_2056	6WJ	23S rRNA D. Radiodurans	H1H2	H3H6	HS2HS22HS2HS10HS13	V	H74-75-80-81- 82-88
2J01_2073	6WJ	23S rRNA T. Thermophilus	H1H2	H3H6	HS2HS22HS2HS10HS13	V	H74-75-80-81- 82-88
1S72_2114	6WJ	23S rRNA H. Marismortui	H1H2	H3H6	HS2HS3HS2HS2HS10H S10	V	H74-75-80-81- 82-88
2AW4_2073	6WJ	23S rRNA E. Coli	H1H2	H3H6	HS2HS22HS2HS10HS13	V	H74-75-80-81- 82-88
1S72_38	6WJ	23S rRNA H. Marismortui	H2H3		HS2HS4HS11HS3HS3H S9	I	H4-4A-11-12- 13-14
2J01_43	6WJ	23S rRNA T. Thermophilus	H2H3		2HS4HS11HS3HS3HS9	I	H4-4A-11-12- 13-14
2AW4_44	6WJ	23S rRNA E. Coli	H2H3		2HS4HS11HS3HS3HS9	I	H4-4A-11-12- 13-14
1NKW_43	6WJ	23S rRNA D. Radiodurans	H2H3		2HS4HS11HS3HS3HS9	I	H4-4A-11-12- 13-14
1NKW_829	7WJ	23S rRNA D. Radiodurans	H2H3, H6H7	H1H4	HS42HS2HS4HS3HS3H S7	II	H36-37-38-39- 40-41-45
2AW4_816	7WJ	23S rRNA E. Coli	H2H3, H6H7	H1H4	HS4HS2HS5HS4HS32H S4	II	H36-37-38-39- 40-41-45
2J01_816	7WJ	23S rRNA T. Thermophilus	H2H3, H6H7	H1H4	HS42HS2HS5HS32HS4	II	H36-37-38-39- 40-41-45
1S72_909	7WJ	23S rRNA H. Marismortui	H2H3, H6H7	H1H4	HS4HS2HS5HS4HS32H S4	II	H36-37-38-39- 40-41-45
2J01_6	9WJ	23S rRNA T. Thermophilus	H4H5		HS7HS7HS18HS7HS11H S2HS4HS19HS5	I	H1-2-25-26-47- 72-73-94-99
1NKW_7	10WJ	23S rRNA D. Radiodurans	H4H5, H9H10		HS8HS9HS18HS7HS11H S2HS4HS2HS4HS4	I	H1-2-25-26-47- 72-73-94-98-99
2AW4_7	10WJ	23S rRNA E. Coli	H4H5, H9H10		HS6HS7HS18HS7HS11H S2HS4HS2HS5HS5	I	H1-2-25-26-47- 72-73-94-98-99

In our previous analysis of four-way RNA junctions²⁷, we identified nine broad four-way junction families according to coaxial stacking patterns and helical configurations (Fig. 2). Helices within these junctions stabilize their conformations using common tertiary motifs like coaxial stacking, loop-helix interaction, and loop-loop interactions. Novel interactions involving A-minor motifs and coaxial stacking were revealed repeatedly at the point of strand exchange in many elements within families cH, cL and cK (Fig. 2B-D). In our analysis of higher order junctions, we find more disorder in the organization of their components. Still, similar to three-way and four-way junctions, helices tend to arrange locally in parallel and perpendicular patterns. Similar repeating motifs such as the A-minor interactions and the sarcin/ricin like motifs are also commonly encountered.

Figure 2.

Classification of RNA four-way junctions into nine families according to their coaxial stacking properties, perpendicular helical configurations, and flexible helical arms (see inset box). Diagrams (A-C) consist of junction families with two coaxial helices sorted by their interhelical angles. Diagrams (D-E) Consist of junction families containing one coaxial stacking, while diagrams (F-I) contain no coaxial stacking but can be characterized by helical alignments and perpendicular configurations stabilized by key tertiary motifs.

Higher order junctions

Due to the small number of examples available for higher order junctions (Fig. 3), it is not possible to design a classification scheme similar to the families assigned in junctions with small degrees^{26; 27}. However, a number of recurrent interaction patterns and motifs can be observed, and their helical elements can be organized using coaxial stacking patterns and other helical arrangements as described below.

Figure 3.

Histogram from a total of 207 RNA junctions sorted by branching degree ranging from 3 (3WJ) to 10 (10WJ).

Five-way junctions

Five-way junctions resemble lower-order junctions in terms of their helical arrangements. For instance, Fig. 4A-C shows junction diagrams with two coaxial stacking interactions (seen as aligned colored helices) analogous to families in four-way junctions²⁷. Specifically in Fig. 4A, a junction found in the Azoarcus intron³⁶ contains all its helical axes aligned roughly in a coplanar and parallel arrangement and stabilized by long-range interactions, forming a crossing at the point of strand exchange similar to elements in the four-way junction family cH. A-minor interactions³⁷ (denoted by empty and solid triangles known as Sugar-Sugar interactions) are the most conserved interactions responsible for such crossings. Similarly, the junction 2BTE_6 in Fig. 4B corresponds to the transfer RNA, where four helices form the well known “L” shape while an extra helix bulges out of the “L” shape. Also of interest, both Figs. 4B and 4C contain junction examples with a pair of perpendicular coaxial stacking interactions. While the pattern in Fig. 4B is a coaxial stacking produced between consecutive helices, that in Fig. 4C is a coaxial stacking between pairs of non-consecutive helices (H₂H₅ and H₃H₅ for each case). Thus, coaxial stacking interactions are not exclusively formed between neighboring helices.

Figure 4.

Network interaction diagrams of five-way junctions sorted by coaxial stacking and helical configurations. The network symbology follows the Leontis-Westhof notation³⁵ (see inset boxes). Figures A-C show junction diagrams with two coaxial stacks aligned either in parallel (A) or perpendicular to each other (B-C), while figures D-F portraits junction diagrams with one coaxial stack perpendicular to at least one helical arm.

Fig. 4D-F shows junction diagrams with one coaxial stacking perpendicular to at least one helix. Specifically, Fig. 4D illustrates a junction with one coaxial stack and one helical alignment (helices aligned without stacking interactions) arranged in a perpendicular configuration. As observed in three and four-way junctions, such perpendicular arrangements among helices are stabilized by loop-loop interactions (2BTE_6 in Fig. 4B and 2AVY_57 in Fig. 4D), loop-helix interactions (2J01_45 in Fig. 4C) or helix-helix interactions (1S72_657 and 2AVY_35 in Fig. 4F). Loop-loop interactions typically involve Hoogsteen or Sugar edge interactions, but can also involve WC basepairs. Loop-helix interactions primarily involve Sugar-Sugar interactions forming A-minor motifs. Helix-helix interactions involve minor groove interactions and will be discussed below in more detail. Junction diagrams in Fig 4F resemble family cK of four-way junctions, which are composed of one coaxial stacking between two helices, while a third helix aligns perpendicular to the coaxial stack. The remaining two helices are arranged based upon the length of their flanking loop elements.

Six to ten-way junctions

In contrast to the compact globular shapes that many protein structures have, RNA molecules prefer rather compact prolate ellipsoidal shapes^{13; 38}. This property reflects the way junctions form by keeping most of their helical axes roughly coplanar. Compared to junctions with a low degree of branching, higher order junctions are more disordered in the organization of their components; still, the basic helical arrangements such as coaxial stacking (present in every high order junction), parallel, and perpendicular helical axes are retained, as described next.

Fig. 5A shows a six-way junction from the ribonuclease P, forming a coaxial helix H₁H₂ and helices H₃ and H₄ in a plane, while the coaxial helix H₅H₆ leaving this plane. The conformation produced by coaxial helices H₁H₂ and H₅H₆ is similar to the antiparallel conformation found in the four-way junction in the hairpin ribozyme³⁹. The diagram in Fig 5B shows a six-way junction with the helical axes in a plane. The single strand J_5/6 contains nucleotides 2385-2387 base pairing with a hairpin loop, forming a (pseudoknot) helix perpendicular to H₄. The homologous six-way junctions found in the H. Marismortui and T. Thermophilus (1S72_2114 and 2J01_2073 in Table 1) shows helices H₃ and H₆ aligned.

Figure 5.

Network interaction diagrams of six-way junctions sorted by coaxial stacking and helical configurations. The network symbology follows the Leontis-Westhof notation³⁵ (see inset boxes). (A) Junction diagram with two coaxial stacks, and (B-C) portraits junction diagrams with one coaxial stack and one perpendicular helical alignment.

In Fig. 5C, the six-way junction 2J01_43 contains helices H₁-H₃ arranged by forming a coaxial helix H₂H₃ which aligns perpendicular to H₁, in a similar conformation to members of family A in three-way junctions such as the M-box riboswitch (2QBZ_53 in Table S1), and three-way junctions found in the large ribosomal subunit (1S72_51, 1S72_1403 and 1S72_2130 in Table S1).

The seven-way junction in Fig. 6A is formed by 3 coaxial helices aligning their axes more or less in a plane. The coaxial stacking between non-neighboring helices H₁ and H₄ is due to a sarcin/ricin motif⁴⁰ formed between strands J_1/2 and J_7/1. The pair of coaxial helices H₂H₃ and H₁H₄ aligns similar to family cH in four-way junctions²⁷, where a crossing occurs at the point of strand exchange caused by A-minor interactions. At the same time, the pair of coaxial helices H₁H₄ and H₆H₇ aligns similar to family H with its extra helix H₅ in between. Helices H₁ and H₃ arrange perpendicular to each other.

Figure 6.

Network interaction diagrams of seven-ten order junctions. (A) seven-way junction, (B) nine-way junction and (C) ten-way junction. The network symbology follows the Leontis-Westhof notation³⁵ (see inset boxes).

The nine and ten-way junctions shown in Fig. 6B-C correspond to the central junction connecting all domains in the 23S rRNA. The ten-way junction contains an extra helix presumably formed through evolutionary variation. Note that in both cases the strand J_3/4 forms a “helical region” composed of alternating WC canonical and non-WC basepairs. Our definition of a helix requires at least two consecutive WC canonical basepairs to be formed; therefore, this region is considered as a strand. Both junctions are non-planar due to the high degree of branching and form three small globular helical regions. The first region is composed of helices H₁ and H₈-H₉ (and H₁₀ for the ten-way junction) arranging similarly to family cK in four-way junctions²⁷. Helices H₂, H₃, H₆, and H₇ align similar to family X in four-way junctions. The third region is the coaxial helix between H₄ and H₅.

Another common characteristic of higher order junctions is that long single stranded elements occur to reduce steric clashing caused by junctions with many helical arms, while preserving the preferred prolate and ellipsoid shapes of RNA 3D structures. The single strands connecting two helices often traverse or “jump over” a third helix in between as it occurs in the strand J_3/4 shown in Fig. 5C. Moreover, these single strands interact with several junction components while traversing as in the example 2AVY_35 in Fig. 4F. Here the strand J_4/5 connecting helices H₄ (magenta) and H₅ (orange), interacts with J_3/4 and with itself, then interacts with J_2/3 and finally with J_5/1. These longer strands between helices allow frequent formation of pseudoknots (Figs. 5A-B and 6B-C). Other properties of higher order junctions that are shared by junctions with lower degrees are described in the following sections.

Statistical features in RNA junctions

From our dataset of 207 RNA junctions listed in Tables 1, S1 and S2, more than half are three-way junctions, and the number decreases as the degree of branching increases. Fig. 3 shows that the frequency of junctions arranged by degree of branching can be estimated by the exponential function y = 228.4e^−0.78x ( R² = 0.94 ), but it is not clear how this estimate will change with increase RNA structures. Junctions of higher degree of branching are observed in RNAs of larger size such as the ribonuclease, group II intron, and ribosomal RNA. In contrast, junctions with a small degree of branching occur in a wide range of RNAs, from riboswitches to ribosomal RNAs.

The loop (single stranded) regions connecting helical elements in junctions are composed by uneven proportions of nucleotide composition as shown in Fig. 7A. While a low percentage of Cs (14%) can be noted, loop regions are strikingly A-rich (40%) for two reasons: A-minor interactions are important in stabilizing helical arms, and adenines offer flexibility to the loop regions. Conversely, the lower concentration of Cs in loop regions corresponds to the smaller number of non-WC basepairs known involving cytosine; however, a reasonable number of these Cs (14%) participate in pseudoknot formation or WC GC basepairs between loops within the junction. In addition, the concentration of WC basepairs near the end of helices (first and second position) produce a high concentration of GC (73%) basepairs, compared to lower AU (20%) and GU wobble (7%) basepairs (data not shown); this might be explained by the high stability (3 hydrogen bonds) of GC basepairs.

Figure 7.

Junction statistics. A) Proportion of nucleotides at the single stranded (loop) elements within junctions. B) Frequency distribution of loop regions within junctions arranged by size. Values for any loop, for loops between coaxial stacking and for loops between helices forming no coaxial stacking are given in blue, red and yellow respectively.

Fig. 7B describes the distribution of the loop size for all loops within helical junctions (blue), loops between coaxial helices (stacked loop, shown in red), and loops between helices where no coaxial stacking is present (non-stacked loops shown in yellow). In general, a large number of loops range in size from 0 to 6 with a peak at 2, while the less frequent cases are loops of sizes 14 to 22. Fig. 7B also shows (in red) that coaxial stacking occurs preferentially in helices adjacent to loops of smaller size, and no stacking is observed for helices between loops of size greater than 8. Coaxial stacking of helices adjacent to loops of size 6 or 7 occurs often due to many non-canonical basepair interactions, which in turn stack with such helices, or also due to the presence of pseudoknots forming at the loop. While a preference for coaxial stacking formation between loops of small size can be noted, there are several cases in which helices with a small loop size do not stack. Particularly, Fig. 7B shows a peak at 2 corresponding to loops between non-stacked helices (99 out of 143). Many reasons could explain the absence of coaxial stacking in these cases, for example the influence of external forces such as pseudoknot formation, long-range tertiary interactions, and protein binding.

In agreement with work by Elgavish et al.⁴¹, non-canonical basepairs involving AG occur frequently at the end of helices, particularly a trans AG Hoogsteen-Sugar or cis AG Watson-Watson basepairs. These, along with standard WC GC basepairs forming a pseudoknot, are the most frequent interactions observed at the end of helices in junctions. When a non-canonical basepair AG trans Hoogsteen-Sugar is formed, it often stacks to a trans AU Hoogsteen-Watson basepair. These two basepairs are recurrent interactions observed in many junctions and become parts of larger 3D motifs such as the sarcin/ricin^{31; 40} or UA-handle motifs⁴². But they can also form as independent and stable sub-motifs, often binding to RNA or proteins, and assisting in the formation of coaxial stacking between helices.

Other important basepair interactions found in junctions are the Sugar-Sugar basepairs, which can form A-minor motifs³⁷ and often combine with coaxial helices forming higher order motifs²² (A-minor/Coaxial helix). In addition, when long-range interactions occur in junctions, a vast majority of A-minor motifs are formed between loop regions flanking helices (e g., hairpins and internal loops), while the helical receptors are located near the end of helices²². Other basepair interactions also occur and are composed mostly of purine-purine interactions. Long-range interactions such as A-minor are important elements because they stabilize helical arms in junctions and allow the proper function of RNA molecules.

Ribo-base interactions stabilize perpendicular helical conformations

One of the most common elements in the ribosome, highlighted by the structure determining authors, is the interaction of RNA double helices via minor grooves. Examples of such interactions are A-minor³⁷, ribose zipper⁴³, G-ribo⁴⁴ and along-groove packing motif (AGPM)^{45; 46}, also known as p-interaction⁴⁷. The interactions presented here describe yet another strategy used for packing minor grooves of rRNA helices against each other.

Helices in junctions often align their axes more or less perpendicular to each other via helix-helix interactions along their minor grooves (Fig. 8A). Because the minor groove in A-RNA has a slightly concave shape, the sugar-phosphate backbone of each helix can pack along the minor groove of the other helix. We previously reported perpendicular interactions in four-way junctions where the AGPM motif is present²⁷ (GU-WC interaction in blue shown in Fig. 8A). A full analysis based on all junctions allows us to recognize two new interactions which often cooperate with AGPM motifs. The combined interactions are composed of four WC basepairs, forming an angle of approximately 60° between their corresponding basepair planes, and occurring when helices are closely packed. Because these new interactions involve ribose-base interactions, we denote them as ribo-base type I (RI) and ribo-base type II (RII) interactions (see Fig. 8).

Figure 8.

(A) Perpendicular alignment between helices H2 and H25 in the T. Thermophilus 23S rRNA structure (PDB 2J01). Residues in blue correspond to the AGPM motif (G539-U554 and G17-C523) and residues in red correspond to the ribo-base interaction type I (C540-G553 and C18-G522). (B) Ribo-base interactions types I and II. (C) Consensus motif for the perpendicular interaction of helices composed of four stacked basepairs at each helix.

The ribo-base type I is characterized by a 2-fold symmetry between two canonical WC basepairs connected by hydrogen bonds interactions between the O2′ of a G residue of the first basepair, and an N2 of a G (or N3 of an A) residue of the second basepair, and between O2′ of a G (or A) residue of the second basepair, and N2 of a G residue of the first basepair (see Fig. 8B). Ribo-base type I occurs between a G of the first basepair and a purine (A or G) of the second basepair. Interestingly, when it appears next to a AGPM motif, a WC CG appears stacked below the WC GU wobble basepair. Indeed, this basepair signature is even more conserved than the WC GC receptor of the GU wobble in the AGPM motif (Table S3).

The ribo-base type II consists of a roto-reflection symmetry (rotation by 180° followed by a reflection around its axis) where two WC basepairs interact by hydrogen bonds between the O2′ of a G residue of the first basepair with an N2 of a G (or N3 of an A) residue of the second basepair, and between O2′ of a C or U residue of the second basepair with N2 of a G residue of the first base (Fig. 8B). When appearing next to the AGPM motif, the CG basepair stacked below the GU basepair can be replaced by a GC basepair, as long as a substitution from CG to GC (or AU) on the receptor basepair of the second stack occurs (see Table S3).

We found 45 instances of ribo-base interactions, mostly located in homologous regions of the ribosomal RNAs considered, and most of them form next to the AGPM motif. While most cases occur between helical elements in junctions, other instances also occur in pseudoknots or near internal loops. Sequence and secondary structure signature consensus elements for these motifs are shown on Fig. 8C, where the ribo-base interactions appear next to AGPM. There are, however, cases where AGPM motifs with no ribo-base interaction appears or ribo-base interactions in non-AGPM patterns. These cases usually occur when WC basepairs are replaced by other basepairs such as cis Watson-Watson AG, or when the GU wobble is replaced by a WC AU basepair (Table S3). Furthermore, crystallographic data from a hammerhead ribozyme (PDB: 1HMH) and tRNA-Gly (PDB: 1VAL) shows type I and II interactions forming between a pair of helices which are possibly tightly packed during to the formation of the crystal.

Analogous to AGPM motifs^{45; 46; 47}, ribo-base interactions bring together helical elements and stabilize RNA molecule for proper function. Another possible role is to act as a mechanism for promoting RNA-protein interactions of neighboring purine nucleotides. Klein and coworkers¹¹ reported that proteins L18e and L15 in the H. Marismortui have a high structural homology in the C-terminal domains and both interact with the five-way junction 1S72_657 (Fig. 3F), forming a near identical nucleotide and amino acid composition. Both proteins L18e and L15 each interact near ribo-base interactions type I (C658-G747 with C685-G661, and C696-G689 with C741-G730 respectively). A close examination of both cases reveals purine bases that expose their hydrophobic surfaces at the protein-RNA interaction site. In other instances, when pairs of helices are closely packed through AGPM and ribo-base interactions, the AGPM/ribose-base motif appears near the end of helices flanking a trans AG Hoogsteen-Sugar basepair interactions. This allows these purine bases to expose their hydrophobic surfaces for possible RNA-protein interactions.

Folding similarity among junctions with different degrees of branching

With the available 3D structures of large RNA molecules such as ribosomal RNAs^{8; 9; 10}, group I introns^{36; 48; 49} and RNase P structures^{50; 51}, it is now evident that there is a high degree of structural conservation in tertiary structures between homologous RNAs. This fact reflects the similarity among junction architectures despite differences in secondary structure. For instance, Krasilnikov and coworkers⁵¹ reported 3D structural similarities in the S domain of RNase P between an internal loop in RNase P type A and a four-way junction in RNase P type B. Also, most transfer RNA structures are composed of a four-way junction (e.g. 1EFW_6 in Table S2), but the example shown in Fig. 4B illustrates a tRNA with a five-way junction conformation. Another interesting example is found in the group I introns (see Fig. 8A), where a three-way junction (1U6B_45 in Table S1) in the Azoarcus intron³⁶ and a five-way junction (see 1Y0Q_43 in Table 1 and Fig. 4B) in the Twort intron⁴⁸ align their corresponding helices P3, P4 and P6 with a high degree of similarity (RMSD 1.09 Å) despite differences in their secondary structure. This structural similarity is in agreement with the observations that group I introns contain conserved core elements formed by junctions, which provide structural stability with the help of conserved peripheral elements by forming long-range contacts⁵².

Moreover, the modular architecture of folded RNAs implies that distances between interacting parts are conserved in functionally homologous molecules³²; thus, similarities in junctions can be made apparent by observing network interaction diagrams and their 3D motifs. For example, in the large subunit of the ribosomal RNA, a five-way junction in H. Marismortui (see 1S72_657 in Table 1 and Fig. 4F) is structurally similar to the four-way junctions found in homologous counterparts in T. Thermophilus, E. Coli and D. Radiodurans (2J01_600, 2AW4_600, and 1NKW_608 in Table S2). In all cases, four helices interact in pairs via perpendicular motifs caused by ribo-base interactions with AGPM. Similarly, the core junctions whose diagrams are shown in Fig. 6B-C present a highly conserved structural similarity between the nine-way junction found in the T. Thermophilus and the ten-way junctions found in both the E. Coli and D. Radiodurans. These observations suggest that the extra helices that are “left out” might have formed later in evolution for particular advantages in species.

Strikingly, a structural similarity of junctions with diverse degree of branching was also observed in non-homologous elements where junctions with a larger degree of branching arrange their helical elements to form “sub-junctions” of smaller degrees. For instance, the six-way junction 2J01_2073 arranges helices H₁, H₂ and H₃ locally similar to three-way junctions of the C family. Elements in family C consist of one coaxial stacking, and a helix aligning parallel to the coaxial helix, by allowing the single strand connecting the coaxial helix to the parallel helix to structure like a hairpin using the standard U-turn. The six-way junction also forms a U-turn hairpin within the loop J_6/1 between helices H₁ and H₆. Fig 9B shows a pairwise structural alignment (RMSD 1.56 Å) between this six-way junction and the three-way junction 1S72_2551 (Table S1) of the family C. Similarly, the U-turn hairpin motif is also found in the four-way junction 2AW4_1832 (Table S2) within the loop J_3/4, forming a sub-three-way junction element between helices H₂, H₃ and H₄ (helices also labeled 65-67 by Leffers et al⁵³). Another example is found in helical elements H₁-H₄ in the seven-way junction, shown in Fig. 6A, which can be decomposed into a four-way junction of the cH family²⁷ while helices H₅-H₇ can be associated to a three-way junction of the C family²⁶ as observed in Fig. 9C. Here, both the four-way junction 2AVY_141 from Table S2 and the three-way junction 2J00_671 from Table S1 superimposed with the seven-way junction 2AW4_816 (RMSDs 1.88Å and 1.65Å respectively).

Figure 9.

Structural similarity between (A) homologous and (B-C) non-homologous junctions. (A) Alignment between the Azoarcus intron (olive green) and the Twort intron (bright green). (B) A six-way junction (olive green) in the 23S rRNA presenting structural similarity to a three-way (bright green) in the 16S rRNA. (C) A seven-way junction (olive green) in the 23S rRNA presenting structural similarity to a three-way (magenta) and four-way junction (bright green) in the 16S rRNA.

SUMMARY AND DISCUSSION

RNA junctions are important structural elements that serve as major architectural components in RNA. While most junctions found in solved crystal structures are formed by a small number of helical branches, higher order junctions with as many as 10 helices exist. Junctions organize their helical elements using various common interactions, such as long-range interactions, coaxial stacking, and many 3D motifs.

Our analysis of higher order junctions using network interaction diagrams is a complementary and compatible approach to the classification of RNA three-way and four-way junctions given by the Westhof²⁶ and Schlick²⁷ groups, which organize elements according to their helical configurations. Our work also complements other studies. For instance, the SCOR⁵⁴ database lists examples of coaxial helices as elements of tertiary motifs. Similarly, RNA junctions contained in the RNAJunction²⁹ database have been grouped by standard nomenclature² based on the size of each loop region. However, similar junctions from homologous RNAs can differ by single insertions of deletions in the loop regions, leading to different classifications under the standard nomenclature.

In the present analysis, we considered higher order junctions from 5 to 10 helices, and compared coaxial stacking and basepair configuration properties to those noted in lower order junctions. We described statistical properties of helices and loop regions for all these RNA junctions and introduced a new motif composed of ribo-base interactions and the AGPM, which is involved in perpendicular helical arrangements. We noted the folding similarity that exists among junctions with different degrees of branching.

In agreement with previous works^{26; 27; 28}, the data from Fig. 7B indicate a preference for coaxial stacking formation for helices whose common single stranded loop is small in size. However, there are several cases where helices with a small loop between them do not stack. The reasons for the absence of coaxial stacking are diverse. Often, elements in the loop regions within junctions form non-canonical basepairs, which in turn can help reduce the spatial distance between helical arms and facilitate coaxial stacking. In many cases of the family C of the three-way junctions²⁶, a small U-turn motif forms at the end of a helix²⁶, possibly preventing a coaxial stacking on the caped helix. In addition, proteins can disrupt coaxial helices when their presence alters helical orientations. The four-way junction 1S72_1743 (Table S2) found in the H. Marismourtui 23S rRNA, contains a pair of helices (labeled 62-63 by Leffers et al⁵³) with no single stranded region between them, but the helices are distorted by the protein L19e, thus preventing the formation of coaxial stacking.

Furthermore, in some cases, even if the size of a loop J_i/i+1 is small, the size of neighboring loops J_i-1/i and J_i+1/i+2 can be equal or smaller, as observed in elements of four-way junctions families H and cH ²⁷ (Table S2). This can lead to an interconversion of stacking conformers or to a competition for coaxial stacking conformers, which can ultimately be decided by long-range interactions. Indeed, experiments for the hammerhead ribozyme⁵⁵ and hairpin ribozyme⁵⁶ have shown that loop-loop interactions act as important elements in the function of these ribozymes, by stabilizing the correct conformation of these junctions. In particular, A-minor motifs occurring within the junction (e.g., Fig. 2A and 1S72_238 from Fig. 2C), help stabilize the structure, and avoid interconversion of different configurations.

Although in general, due to the conformational flexibility and dynamic character of junctions, a continuum of junction conformations might be possible, our compilation of RNA junction domains based on available structures illustrates Nature’s strong preferences for the arrangement of RNA helical elements in parallel and perpendicular patterns, while keeping the helical axes coplanar. As recently discussed in an essay⁵⁷, most RNA structure and folding data comes from in vitro experiments, where high ionic concentrations can compensate for the lack of in vivo folding factors such as ligands and RNA chaperones. Differences between in vitro and in vivo folds of RNA are still being investigated.

Long-range interactions that stabilize helical elements are very diverse, but often involve Sugar-Sugar interactions in the form of A-minor motifs. Other interactions such as base-ribose and long-range stacking interactions are also observed. One advantage of studying junctions with different number of helices is that it allows recognition of important repeating motifs such as the sugar-edge interactions (A-minor), sarcin/ricin, and trans AG Hoogsteen-Sugar interactions. These sets of non-canonical basepairs play important roles in RNA’s structure and therefore function.

Ribo-base interactions are novel helix-helix interactions found in perpendicular helical conformations. They belong to the same family of helix packing interactions such as the G-ribo⁴⁴, A-minor³⁷, AGPM⁴⁶, and ribose zipper⁴³. Because ribo-base interactions often appear next to the along-groove pacing motif (AGPM), both motifs form parts of a larger motif (AGPM/ribo-base), whose main function is to pack together helical elements and stabilize RNA molecule for proper function. Such motifs can also act as RNA-RNA or RNA-protein binding promoters by helping their flanking trans AG Hoogsteen-Sugar basepair interactions to expose their hydrophobic surfaces for binding.

As more interactions involving RNA base and ribose are discovered, one can foresee the need to extend the current RNA basepair classification given by Leontis and Westhof³⁵ to include ribose-base interactions.

We encountered many examples of higher order junctions that arrange their helical elements similar to lower order junctions. The junction examples belong to both homologous and non-homologous RNAs. One can then ask: how are higher order junctions formed? We propose that some junctions with a high degree of branching are formed from insertions and unions of smaller order junctions under evolutionary pressure; the optimal junction sites for insertions and unions likely correspond to regions that would not dramatically change its internal tertiary structure conformation. Our analysis also suggests that higher order junctions can be decomposed into smaller “sub-junctions”. Ultimately, a better understanding of junction decompositions can help predict RNA three-dimensional structures and functions.

MATERIALS AND METHODS

The data set of our 3D RNA junctions as collected from the RCSB Protein Data Bank³³. Based on available structures as of April 2009, 554 high-resolution structures were selected, with repetitions omitted by choosing the more recent structures. Junction elements were searched within these and analyzed for basepair interactions.

To perform our comprehensive search of k-junctions (3≤ k ≤10) in the set of RNA structures above, we first considered the secondary structure associated with every 3D structure defined in terms of its canonical WC basepairs and the single stranded regions. The search for canonical WC basepairs was performed using the program FR3D⁵⁸. Second, we searched for sets of k distinct strands connecting in a cyclical way by at least two consecutive canonical WC basepairs (Fig. 1A). For simplicity, pseudoknots were automatically removed during the search, but later re-inserted for statistical analysis. Visual inspection was also used to verify the correctness of our procedure. In addition, we compared our search outcome to data available from the RNAJunction database²⁹, to ensure the verity of all junctions.

Crystal structures containing at least one junction each were identified, 43 in total. The structures include the two high resolution crystal structures of the 16S (PDB 2AVY, 2J00) and four 23S rRNA (PDB 1NKW, 1S72, 2AW4, 2J01). Although the 3D shape of equivalent rRNA molecules is highly conserved among species, differences are informative because they help to understand evolutionary changes that Nature allows while keeping their molecular function intact. In total, our dataset thus contains 207 RNA junctions as listed in Tables 1, S1 and S2. Additional detailed junction information such as PDB source, sequence, and residue numbers are available in Tables S4-S10 from the Supplementary Material.

Non-canonical base pairing with alternate hydrogen bonding patterns occur often in RNA. A consensus between FR3D and RNAVIEW⁵⁹ was considered to classify basepairs. Where discrepancies occur, we employed visual programs such as Pymol (DeLano Scientific LLC) and Swiss PDB viewer⁶⁰. Additionally, the junction data were analyzed from different perspectives: sequence signatures, length of loop regions, 3D motifs, and the 3D organization of their helices. Orientation aspects such as in coaxial stacking, helices that form perpendicular inter-helical angles, and helices aligning their axes in parallel without the use of stacking forces were analyzed on by inspection. Pairwise structure alignment between junction domains was done using the ARTS web server⁶¹.

Network interaction diagrams describing basepair interactions are represented symbolically according to the Leontis and Westhof base pairing classification^{34; 35}. The diagrams were created using VMD⁶² and S2S⁶³, a visual aid program based on RNAVIEW.