Crystal structure of RNA-DNA duplex provides insight into conformational changes induced by RNase H binding

Abstract

RNA-DNA hybrids play essential roles in a variety of biological processes, including DNA replication, transcription, and viral integration. Ribonucleotides incorporated within DNA are hydrolyzed by RNase H enzymes in a removal process that is necessary for maintaining genomic stability. In order to understand the structural determinants involved in recognition of a hybrid substrate by RNase H we have determined the crystal structure of a dodecameric non-polypurine/polypyrimidine tract RNA-DNA duplex. A comparison to the same sequence bound to RNase H, reveals structural changes to the duplex that include widening of the major groove to 12.5 Å from 4.2 Å and decreasing the degree of bending along the axis which may play a crucial role in the ribonucleotide recognition and cleavage mechanism within RNase H. This structure allows a direct comparison to be made about the conformational changes induced in RNA-DNA hybrids upon binding to RNase H and may provide insight into how dysfunction in the endonuclease causes disease.

Abbreviations

AGS

Aicardi Goutiéres syndrome

HBD

hybrid binding domain

NMR

nuclear magnetic resonance

PPT

polypyrimidine/polypurine track

RNase

ribonuclease

Introduction

RNA-DNA hybrids exist in all cells and play crucial roles in genomic and mitochondrial DNA replication as well as in transcription. The RNA primers in Okazaki fragments are required for lagging strand synthesis during genomic DNA replication and R-loops are generated in mitochondrial DNA replication.^1-5 RNA-DNA intermediates are formed during transcription and are part of class I retrotransposon element amplifications. They are also a necessary intermediate in retrovirus replication.^6,7 Recently it has been found that ribonucleotides are inserted into DNA during normal replication at a rate of approximately one ribonucleotide for every 7,600 nucleotides added.^8,9 The presence of these single ribonucleotides within DNA act as a strand discrimination signal for mismatch repair of leading strand replication errors.^10,11 Interestingly, removal of these ribonucleotides from DNA is essential for long-term maintenance of genome integrity and development.^9,12

Ribonucleotides within RNA-DNA hybrids are recognized and hydrolyzed by the RNase H enzymes. Three classes of RNase H enzymes exist (for reviews see references 13,14). RNase H1 is a single domain enzyme that cleaves ribonucleotides from DNA but must have at least 4 ribonucleotides present for activity. RNase H2 recognizes and cleaves a single ribonucleotide within a dsDNA duplex. RNase H3 is structurally similar to RNase H2 but catalytically functions like RNase H1. In humans, mutations in the RNase H2 enzyme leads to the autosomal recessive neurological disorder, Aicardi-Goutières syndrome (AGS), which is a severe autoimmune disorder characterized by the loss of white matter in the brain, increased white cells and cytokine interferon α,^15-19 presumably due to the loss of ability to process ribonucleotides within DNA. All of the ribonucleases have similar catalytic activity and substrate specificity. A nucleophilic attack by water hydrolyzes the phosphate backbone through the assistance of a divalent metal ion (Mg²⁺ or Mn²⁺) and forms a product with a 5′-phosphate and 3'-hydroxyl group.²⁰

Although RNase H can bind both dsDNA and dsRNA, neither are hydrolyzed by the enzyme. Understanding how RNA-DNA hybrids are recognized and processed by RNase H enzymes is an important step in determining how enzyme dysfunction leads to disease. While structures of RNase H enzyme in complex with RNA-DNA and dsDNA duplexes have been determined,^21-26 the recognition requirements for double-stranded RNA-DNA hybrids are still unclear. Relatively few structures of RNA-DNA hybrids alone have been determined, most of which are 10 bp or less or contain polypurine or polypyrimidine tracts (PPT).

In order to understand structural aspects of the double stranded duplex that are important for RNase H recognition and to determine the structural changes to the nucleic acid structure are induced upon binding we have determined the crystal structure of a dodecameric RNA-DNA duplex. The sequence of this RNA-DNA hybrid is identical to one determined in complex with RNase H enzyme²⁰ and contains 50% G/C content. Our results show that, like many other crystallographic hybrid structures, the RNA-DNA duplex takes almost solely the A-form parameters but upon binding to RNase H, widens the major groove to widths much closer to B-DNA and decreases the curvature of the duplex, which may be key steps in the recognition and cleavage of ribonucleotides from DNA.^22-24,26 The RNase H protein contacts the RNA and DNA strands in the minor groove and can differentiate between the A- and B-forms of the duplex as well as the sugar geometry/moiety (ribose or deoxyribose) to ensure specificity for an RNA-DNA hybrid.

Results

This RNA-DNA hybrid will be referred to as the GAA duplex and Figure 1D depicts the numbering system used. The GAA duplex crystallized with 4 RNA-DNA duplexes in the asymmetric unit (Table 1). Superimposing the 4 hybrids indicate that they are structurally similar with a fairly small measured r.m.s.d. between the duplexes ranging from 0.42 Å to 0.75 Å. The differences in structures are primarily due to slight variations in the incline of the bases. The dodecamer RNA-DNA hybrid structure displays a conformation most closely related to A-form RNA and DNA (Fig. 1). Unlike many structures that have currently been determined in the absence of an enzyme, this hybrid has a more uniform pyrimidine/purine sequence of 5′-gacaccugauuc-3′ and a cDNA sequence of 5′-GAATCAGGTGTC-3′ compared to PPT hybrids, which are common in HIV primers and may also contribute to atypical structures of the duplex due to adenine stacking, a-g-a deformations, and “unzipping” caused by unpairing of the bases.^21,22,24,25

Figure 1.

Structure of RNA-DNA hybrid. (A) The stereoimage of the contents in the asymmetric unit shows the packing of 4 RNA-DNA hybrids with each duplex shown in a different color. (B) The RNA-DNA duplex, with RNA strand in green and the DNA strand in red, has a structure most consistent with A-form helices (see Table 2). (C) Hybrid helix viewed along helical axis. (D) The sequence of the RNA-DNA hybrid with the RNA strand in lower case letters and the DNA strand in upper case letters and numbering is shown above the end bases.

RNA-DNA Helical Parameters. The helical parameters of the GAA duplex are almost solely consistent with the A-form helix, which has been also noted in other hybrid duplex structures.^27-29 We compared the parameters of the GAA duplex with typical A-DNA, B-DNA, A-RNA, and RNA-DNA hybrids calculated from structures in the Protein Data Bank (Table 2). The data indicate that typical A-DNA and A-RNA have very similar characteristics and that the GAA hybrid tends to share these A-form traits. The structure of GAA duplex reveals an average minor groove width of 9.8 Å and depth of 1.2 Å, comparable with the A-form family. This leads to a narrower and deeper major groove of 4.2 Å and 9.5 Å, respectively. Figure 2 shows the changes in the minor groove widths and depths between GAA and 1ZBI. The x-displacement measures the distance of the bases from the center of the helix and the GAA duplex (−3.6 Å) closely resembles that of A-RNA (−3.9 Å) which forms a hole down the axis of the helix (Fig. 1C). B-DNA has a very small displacement and thus no radial shift of the bases. The curvature of bound RNA-DNA hybrid is lower than the free GAA hybrid (8.7° vs 17.7°) which is likely a result of the binding of the RNase domain, which is discussed below.

Table 2.

Average helical parameters of nucleic acid duplexes

	GAA Duplex<1	B-DNA2	A-DNA3	A-RNA4	RNase bound to hybrid5	RNA-DNA hybrids6	PPT hybrids7
X-displacement (Å)	−3.6 ± 1.2	0.3 ± 1.0	−4.4 ± 0.7	−3.9 ± 0.8	−3.9 ± 0.8	−4.0 ± 0.6	−3.2 ± 1.8
Bend (°)	17.7	5.7	22.6	13.9	8.7	22.2	17.9
Minor Groove Width (Å)	9.8 ± 0.7	5.4 ± 0.5	10.1 ± 0.8	9.9 ± 0.7	8.6 ± 1.2	9.8 ± 1.2	9.9 ± 0.9
Minor Groove Depth (Å)	1.2 ± 1.3	5.4 ± 0.5	0.7 ± 0.7	0.7 ± 0.8	2.4 ± 1.0	1.0 ± 1.3	1.1 ± 1.4
Major Groove Width (Å)	4.2 ± 2.6	11.3 ± 1.2	4.9 ± 0.5	4.6 ± 2.2	12.5 ± 3.0	3.9 ± 0.6	3.9 ± 0.3
Major Groove Depth (Å)	9.5 ± 1.5	4.7 ± 1.4	11.2 ± 0.4	9.0 ± 0.6	9.4 ± 2.8	10.5 ± 0.8	9.9 ± 0.7

¹Average of the 4 RNA-DNA duplexes in the asymmetric unit.

²Average of PDBs: 3U2N, 1BNA.

³Average of PDBs: 137D, 240D.

⁴Average of PDBs: 1RNA, 4KYY.

⁵Average of PDBs: 4H8K, 1ZBI.

⁶Average of PDBs: 1D87, 1FIX.

⁷Average of PDBs: 1G4Q, 3SSF.

Figure 2.

Comparison of the minor groove widths and depths. The GAA hybrid (solid lines) has wider minor grooves but they are more shallow than 1ZBI (dashed lines).

The structure also reveals the sugar geometries have C3′-endo puckers in an anti-conformation for almost all of the nucleotides but a few C1′-exo and C4′-exo sugar puckers are present primarily in the DNA strand which may be due to the rigidity of the RNA backbone forcing the DNA backbone to adopt those conformations. B-DNA typically takes an anti C2′-endo sugar pucker and A-RNA and A-DNA have an anti C3′-endo conformation but at the current resolution subtle nuances are difficult to differentiate. Unlike hybrid structures determined by NMR or duplexes bound to RNase H, the GAA duplex is primarily C3′-endo in both the RNA and DNA strands whereas bound hybrids and those determined by NMR have an RNA strand with a C3′-endo and a DNA strand with C2'-endo, or related, sugar conformation.^22-24,30-32

To examine how the sequence of the GAA duplex affected the overall structure of the hybrid as compared to previously determined hybrid structures composed of polypurine and polypyrimidine tracks (PPT), 2 PPT RNA-DNA hybrid structures (PDB: 1G4Q and 3SSF) were used as comparative models with the GAA structure. As shown in Table 3, the helical parameters calculated for the other RNA-DNA hybrids are generally similar to the GAA structure with the notable difference that the width of the major groove for GAA is smaller by 0.3 Å than the RNA-DNA and PPT hybrids.

Table 3.

Helical Parameters of RNA-DNA hybrids

	1G4Q PPT	3SSF PPT	1FIX	1D87	1ZBI Bound	4H8K Bound
X-displacement (Å)	−3.4 ± 1.2	−2.9 ± 1.3	−3.8 ± 0.5	−4.2 ± 0.4	−4.0 ± 0.6	−3.8 ± 0.5
Bend (°)	26.0	9.8	17.0	27.4	7.4	10.0
Minor Groove Width (Å)	9.7 ± 0.7	10.1 ± 0.6	9.7 ± 0.7	9.9 ± 0.9	8.4 ± 0.5	8.8 ± 1.1
Minor Groove Depth (Å)	0.8 ± 1.1	1.4 ± 0.9	1.3 ± 0.8	0.8 ± 1.0	2.6 ± 0.5	2.3 ± 0.9
Major Groove Width (Å)	4.07 ± 0.06	3.6 ± 0.3	4.3	3.4	13.2 ± 1.6	11.8 ± 2.5
Major Groove Depth (Å)	11.13 ± 0.12	8.7 ± 0.7	9.9	11.0	9.6 ± 2.6	9.3 ± 1.2

Discussion

The structural differences between the bound and unbound hybrid reveal that RNase H interactions unwind the helix to create a local distortion in nucleic acid structure and facilitate protein binding. The free and bound RNase H structures have a difference of 2.0–2.2 Å r.m.s.d. along the Cα backbone, indicating that structural changes will be conferred onto the bound duplex upon binding.^20,28,29 Until now, it has been difficult to determine if RNase H changes the conformation of the RNA-DNA hybrid upon binding, because of the lack of the same nucleic acid structure in both the free and bound state. The free GAA duplex and the hybrid bound to the Bacillus halodurans RNase H²⁰ (PDB 1ZBI), have the same sequence and allow us to make a direct comparison to address this issue. A superimposition of the structure of the GAA hybrid onto the RNase H bound duplex structure reveals dramatic differences between the 2 (Fig. 3). The most distinct difference between the RNase H bound duplex and the GAA hybrid is that the protein bound duplex has a smaller minor groove width (8.4 Å vs 9.8 Å), and a significantly larger major groove width (13.2 Å vs 4.2 Å). The GAA structure also shows a significantly larger bend (17.7° vs 7.4°) (Table 3). By having a greater bend, the duplex can interact more favorably with RNase H due to an increased exposure of the surface area of the minor groove and preventing water molecules from diminishing hydrogen bonding interactions with the binding pockets.²⁶ These structural changes are induced by interactions with the protein, most notably the phosphate binding pocket of RNase H²⁰, created by residues T104, N106, S147, and T148 (Fig. 3). RNase H contributes a considerable helical distortion by moving the bound phosphate about 5 Å away from the unbound helical position, resulting in a pinching of the minor groove and widening of the major groove. This leads to a kink in the sugar-phosphate backbone at residue A6 but the same kink is not seen in the free GAA duplex (Fig. 3B).

Figure 3.

RNase (H)binding induces conformational changes in RNA-DNA hybrid. (A) The structure of the GAA hybrid (yellow) superimposed onto the hybrid of the same sequence bound to RNase H (green) (PDB: 1ZBI), reveals RNase H distorts the DNA strand of the hybrid by rotating the phosphodiester backbone around nucleotide dA6 about 5 Å into the nucleotide binding pocket. (B) The superimposition of the 2 RNA-DNA hybrids shows the RNA strand has a relatively similar conformation, but the DNA strand has been distorted to significantly widen the major groove of the helix (see Tables 2 and 3). This flexibility of the DNA strand likely plays a role in substrate recognition and discrimination.

The structure of the hybrid binding domain (HBD) from RNase H in complex with an RNA-DNA hybrid also has the same sequence as the GAA hybrid.²⁰ The HBD is a small protein domain that preferentially binds RNA-DNA hybrids and functions in protein dimerization and processivity for RNase H.^29,33 The structure shows HBD interacts with the RNA-DNA hybrid solely along the minor groove backbone of the duplex (3BSU), in a manner that is very distinct from RNase H catalytic domains (1ZBI). Interestingly, the duplex bound to the HBD has structural parameters that are similar to the unbound GAA duplex (data not shown), supporting the idea that binding in the active site of RNase H enzymes induces conformational changes in the RNA-DNA duplex.

It has been proposed that the RNase H enzymes recognize ribonucleotides within dsDNA by the local change in conformation to the A-form having a specific minor groove width of approximately 10 Å, which is slightly less than A-DNA and A-RNA, a pitch and rise similar to A-DNA, curvature different from B-DNA, as well as the presence of a 2'-hydroxyl group on the ribose sugar.^20,26,27 The structure of the GAA duplex indicates that this conformation is likely induced by protein binding. RNases H are not able to cleave dsRNA, suggesting that the added flexibility of a DNA strand is required for conformational changes to the duplex upon protein binding.³⁴ This is supported by data showing the flexibility of the DNA strand correlates with the rate of cleavage by RNase H enzymes.^13,35 But the width of the minor groove alone is not enough for ribonucleotide cleavage. It has been demonstrated that the binding of B-DNA to RNase H forces the DNA to have a narrow minor groove but cleavage still does not occur.²⁶ Another requirement for catalysis to occur could be based on the degree of curvature of the helix. B-DNA is fairly linear with a bend less than 6° but RNA-DNA hybrids have bends greater than 16° but are straightened to about 8° upon binding to RNase H.

In conclusion, distortions created by RNase H binding induce changes in the minor groove width, curvature of the duplex, and can put strain on specific backbone atoms, likely functioning to fine tune binding affinity and substrate discrimination. Similar mechanisms of helical distortion for nucleic acid binding have been seen with a number of other proteins, most notably prokaryotic transcription factors. DNA binding by the bacteriophage λ Cro protein and repressor, as well as the AmrZ protein from Pseudomonas aeruginosa facilitates helical distortions in DNA in order to discriminate binding sites.^36-39 RNase H binds nucleic acids independent of sequence but needs to distinguish RNA and DNA polynucleotides inferring that substrate flexibility is likely an important factor. The GAA structure finally allows a direct comparison to be made between a bound and unbound RNA-DNA hybrid to determine the structural aspects of the hybrid that provide specificity for RNase H binding. More hybrids with the same sequence that are bound and free are needed to ensure that the minor groove width, degree of the duplex bending, and the presence of the rigid 2′-hydroxyl group are the main requirements for substrate recognition and cleavage.

Material and Methods

Crystallization and X-ray Data Collection. The RNA and DNA dodecanucleotides (5′-gacaccugauuc-3′ and 5′-GAATCAGGTGTC-3′) were ordered from Integrated DNA Technologies with desalting purification. Oligonucleotides were annealed in RNase free water with 10 mM MES, pH 6.5, 5 mM MgCl₂, 20 mM NaCl. Crystals were grown by sitting drop vapor diffusion method at room temperature in conditions containing 20% PEG 3350, 0.2 M magnesium formate, 10% 2-Methyl-2,4-pentanediol. Crystals were frozen in liquid nitrogen prior to data collection. X-ray diffraction data were collected using Cu-Kα radiation on a MicroMax 007 generator and a Saturn 92 CCD detector (Rigaku). Intensity data were processed with the program D*TREK and scaled to 2.8 Å resolution (Table 1).⁴⁰

Table 1.

X-ray data collection and refinement statistics

Data Collection
Wavelength (λ) (Å)	1.54
Space group	P21
Unit cell (a,b,c) (Å)	46.6, 44.5, 83.3
Unit cell (αβγ) (°)	90 105.3 90
Resolution range (Å)	26.3–2.8 (2.9–2.8)
Total Reflections	15,767
Unique Reflections	8,314
Completeness (%)	99.9 (99.9)
Rmerge	0.086 (0.492)
Mean I/σ(I)	13.4 (2.7)
Average Redundancy	7.3 (7.3)
Refinement
Resolution Range (Å)	26.3–2.8 (3.0–2.8)
Working set reflections	15,032 (2,989)
Test set reflections	735 (149)
Number of nucleic acid atoms	1,980
Rcryst	0.221
Rfree	0.249
r.m.s.d. bond lengths (Å)	0.008
r.m.s.d. bond angles (Å)	1.4
Average B factors (Å2)	47.6

Values in parenthesis are for the outermost resolution shell.

Structure Solution and Refinement. The crystals of the RNA-DNA hybrid grew in space group P2₁ with 4 RNA-DNA hybrids in the asymmetric unit. Phases for the data were obtained by molecular replacement using PHASER⁴¹ and the structure of the RNA-DNA component of the RNase H complex (PDB: 1ZBI) as the search model.²⁰ The structure was refined with multiple rounds of simulated annealing and composite omit maps (no crystallographic symmetry averaging) using the program Phenix⁴² followed by rounds of model building using Coot.⁴⁴ PDB_REDO⁴³ was used for refinement validation. The model refinement converged at an R_work = 22 .1% and R_free = 24 .9%. The structure has been deposited in the Protein Data Bank with the PDB accession code 4WKJ.

Helical Analysis. Analysis of the nucleic acid structure was performed using the program Curves+^45,46 as well as on structures from the following PDB files: 3U2N, 1BNA, 137D, 240D, 1RNA, 4KYY, 4H8K, 1ZBI, 1G4Q, 1FIX, 1D87, and 3SSF. The parameters used for Curves+ were wback = 2 .9 Å and wbase = 3 .5 Å.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by funding from the NCI CCSG P30CA012197; the American Heart Association Grant 10GRNT3650033 (TH); and National Institutes of Health [Grant RO1 GM110734 (FWP) and RO1 GM108827 (TH)]. The funding sources had no role in the study design; in data collection; analysis and interpretation of the data; in writing the manuscript; or in the decision to submit for publication.