Characterization of R-Loop Structures Using Single-Molecule R-Loop Footprinting and Sequencing

Abstract

R-loops are three-stranded structures that form during transcription when the nascent RNA hybridizes with the template DNA resulting in a DNA:RNA hybrid and a looped-out single-stranded DNA (ssDNA) strand. These structures are important for normal cellular processes and aberrant R-loop formation has been implicated in a number of pathological outcomes, including certain cancers and neurodegenerative diseases. Mapping R-loops has primarily been performed using DRIP (DNA:RNA immunoprecipitation) based methods that are dependent on the anti-DNA:RNA hybrid S9.6 antibody and short-read sequencing. While DRIP-based methods are robust and report R-loop formation genome-wide, they only do so at the population average level; interrogating R-loop formation at the single molecule level is not feasible with such approaches. Here we present single molecule R-loop footprinting (SMRF-seq), a method that relies on the chemical reactivity of the displaced ssDNA strand to non-denaturing sodium bisulfite and single molecule long-read sequencing as a readout, to characterize R-loops. SMRF-seq can be used independently of S9.6 to generate high resolution, strand-specific, maps of individual R-loops at ultra-deep coverage on kilobases-length DNA fragments.

1. Introduction

R-loops are three-stranded nucleic acid structures consisting of a DNA:RNA hybrid and a displaced single stranded DNA. R-loops are abundant non-B DNA structures in genomes, covering 5–10% of the genomic space in organisms ranging from yeasts, plants, and mammals [1–6]. While evidence exists to support that R-loops can form in trans [7–9], nuclear R-loops are predominantly understood to form co-transcriptionally from the re-invasion of the nascent transcript into the duplex DNA template. This is evidenced from the overwhelming genic distribution of R-loops, their co-directionality and correlation with transcription, and conclusions from in vitro transcription reactions [10]. Investigations into the possible roles of R-loops have suggested that these structures influence a range of cellular processes [11–13]. Under normal conditions, R-loops participate in the regulation of gene expression [2, 14, 15], class switch recombination in B cells [16,17], and efficient transcription termination [18,19]. Deregulation of R-loop metabolism has also been linked to genome instability [20–22] ultimately contributing to pathological conditions such as certain cancers and neurological disorders [23, 24].

R-loops have been mapped genome-wide using DNA:RNAimmunoprecipitation (DRIP) [1], a method that relies on the S9.6 anti-DNA:RNA hybrid antibody [25] and high-throughput short-read sequencing. Since its initial introduction, DRIP has been widely adopted and several variants of the method have been introduced to improve its resolution and strand-specificity [2, 6, 15]. While robust, these methods nonetheless possess few limitations. A first complication is due to the significant residual affinity of the S9.6 antibody for double-stranded RNA (dsRNA) [26]. This residual affinity can result in inaccurate R-loop maps when RNA strands are directly sequenced unless additional steps are taken to remove contaminating dsRNAs [5,27]. Secondly, DRIP, like ChIP approaches, only provides a population average view of R-loop formation derived from a large cell population. The use of high-throughput short-read sequencing technologies as read-outs further precludes obtaining any information on individual R-loops including their lengths and individual start and stop positions. To overcome these limitations and provide an S9.6-independent method to profile individual R-loops at ultra-deep coverage, we developed Single-Molecule R-loop Footprinting and sequencing (SMRF-seq) [28]. SMRF-seq relies on non-denaturing bisulfite conversion [16] to catalyze the efficient conversion of unpaired cytosines (C) located on the displaced ssDNA strand of R-loops to uracils (U). After long-range locus-specific PCR amplification, molecules amplified from the tagged displaced strand will carry patches of cytosine to thymine (T) conversions flanked by unmodified DNA. These patches represent R-loop “footprints”. The use of Pacific Bioscience’s Single-Molecule Real Time sequencing (SMRT-seq) on long PCR amplicons permits the characterization of collections of individual footprints at high coverage. Overall, SMRF-seq permits R-loop mapping at high resolution, in a strand-specific manner, on kilobases-long single-molecule reads and at high coverage. SMRF-seq, together with its accompanying analysis package Gargamel, will fuel new waves of investigations into the formation and function of R-loop structures.

2. Material

Prepare all reagents and dilutions using molecular biology-grade nanopure water (ddH₂O).

2.1. General

1. Tris-EDTA (TE) buffer: 10 mM Tris–HCL pH 8.0 and 5 mM EDTA.

2. 200 proof Ethanol.

3. Elution Buffer (EB): 10 mM Tris–HCl pH 8.5.

4. 1.5 mL and 2 mL Microcentrifuge tubes.

5. 8-strip PCR tubes.

6. Bench top Microcentrifuge.

7. DynaMag-2 beads (Life Technologies, 12321D).

8. AMPure XP beads (Beckman Coulter). Set at room temperature before use.

9. Mini tube rotator.

2.2. Nucleic Acid Purification

1. 15 mL conical tubes.

2. Proteinase K. Dilute to 20 mg/mL working solution.

3. Sodium dodecyl sulfate (SDS). Dilute to 20% working solution.

4. 5PRIME Phase Lock Gel Light 2 mL tubes.

5. Phenol:Choloroform:Isoamyl alcohol (25:24:1).

6. 3 M Sodium Acetate (NaOAc) pH 5.2.

7. Wide bore 200 and 1000 μL barrier tips.

2.3. Non-denaturing Bisulfite Treatment

1. DNA Methylation-Lightning.

2. Rotisserie oven.

2.4. Site-Specific Amplification

1. Thermocycler (optional: gradient-enabled block).

2. Phusion Hot Start DNA Polymerase.

3. 5× Phusion HF Buffer.

4. 5× Phusion GC Buffer.

5. dNTPs: Equimolar amounts of ATP, CPT, GTP, TTP. Dilute to 10 mM working solution.

6. 5 M Betaine.

2.5. SMRT-Bell Library Preparation

1. SMRTbell Template Prep Kit 1.0 (Pacific Biosciences, 100-259-100).

2. AMPure PB 5 mL (Pacific Biosciences, 100-265-900).

2.6. Quality Check

1. Spectrophotometer.

2. 2100 Bioanalyzer.

3. GelRed (Phenix Research Products, RGB-4102).

4. 50 × TAE buffer: For a 1 L stock solution, dissolve 242 g Tris base in 600 mL ddH₂O. Add 57.1 mL glacial acetic acid and 100 mL 0.5 M EDTA pH 8. Adjust volume to 1 L with ddH₂O. Dilute stock to 1× working concentration.

5. Agarose gel: dissolve molecular biology grade agarose to a final 0.7% (weight vol.) concentration in 1× TAE by heating. Cool to ~65 °C. Add 1 × final concentration of GelRed. Cast gel and insert comb.

3. Methods

Overview.

SMRF-seq exploits the presence of long stretches of ssDNA on the displaced strands of three-stranded R-loop structures. The intrinsic ssDNA character of this strand can be revealed upon treatment with sodium bisulfite under non-denaturing conditions [16]. This will lead to the conversion of susceptible cytosines to uracils within the displaced strand of R-loops. This genetic tag can subsequently be read out after PCR amplification and library construction by long read, single-molecule sequencing (SMRT-seq) (Fig. 1). Downstream data analysis steps are performed using the available Gargamel pipeline [28] to seek and capture single-molecule R-loop footprints (SMRFs). True R-loop footprints are expected to be strand-specific and sensitive to pre-treatment with Ribonuclease H [2,16], an enzyme that specifically degrades RNA in the context of DNA:RNA hybrids [29]. By contrast, spontaneous DNA breathing or strand-separation events will lead to conversion on both DNA strands in a manner insensitive to Ribonuclease H activity. Minimal length thresholds enforced during analysis serve to reinforce the distinction between characteristically long R-loops and short DNA breathing events [30]. Note that SMRF-seq can be carried out with or without preliminary R-loop enrichment with the S9.6 antibody without distortion of footprints [28]. The protocol described here does not include any S9.6 enrichment. Detailed step-by-step instructions for performing DRIP have been recently published [31].

Fig. 1.

Overall workflow for both experimental procedures and computational analysis

3.1. Cell Harvest

This section is optimized for the human Ntera-2 and HEK293 cell lines grown in culture. The following steps can be easily adapted to other cell lines. Note that SMRF-seq can be easily performed on R-loops generated upon in vitro transcription of plasmids carrying R-loop prone sequences [32, 33]. If interested in such application, carry out in vitro transcription as described [32] and proceed directly to non-denaturing bisulfite treatment (Subheading 3.3) using the products of in vitro transcription reactions as initial material.

1. Passage cells 16 h before harvesting.

2. After 16 h of growth, count cells and measure cell viability. Optimal cells count should be around 5–6 million cells per sample with >90% viable counts. Cells should be no more than 80–90% confluent at harvest.

3. Add 1.5 mL trypsin to dislodge cells and incubate for 2–5 min at 37 °C until cells come off the plate.

4. Add 5 mL culture medium and pipette mix to remove clumps.

5. Transfer to a new 15 mL conical tube and spin for 3–5 min at 1000 rpm (~200 RCF) to pellet cells.

6. Aspirate media carefully without disturbing cell pellet.

7. Add 10 mL 1 × DPBS and pipette mix to resuspend cells.

8. Re-spin at 200 RCF for 3–5 min to pellet cells.

9. Aspirate supernatant off without disturbing pellet.

3.2. Nucleic Acid Purification

• 1. Add 875 μL TE to the pelleted cells. Pipette mix ~10× to resuspend.

• 2. Transfer 200 μL of resuspended cells into a new 1.5 microcentrifuge tube. Up to four DNA extractions can be performed from the cell material collected.

• 3. Cell lysis: add the following volumes to each tube:

12.5 μL	20% SDS
10 μL	Proteinase K

• 4. Invert 4–5× gently until lysed. The solution should appear clear and viscous.

• 5. Incubate in a 37 °C water bath for 2 h to permit proteinase K digestion. Invert mix and do a brief spin down every hour to remove condensation that forms on the top of the tube.

• 6. Spin down 2 mL Phase Lock tube (PLT) for 1 min at 12,000 RCF.

• 7. Add 177.5 μL TE to each Proteinase K treated sample for a total volume of 400 μL.

• 8. Transfer (pouring or pipetting) Proteinase K treated cell lysate into PLT directly.

• 9. Add 400 μL Phenol:Choloroform:Isoamyl alcohol to lysate. Invert gently a few times (see Note 1).

• 10. Spin down at 12,000 RCF for 5 min at room temperature or at 4 °C if possible (see Note 2).

• 11. Pour top aqueous layer containing nucleic acids (~400 μL) from PLT into new 2 mL tube.

• 12. Add the following to the 400 μL sample to precipitate nucleic acid:

40 μL	3 M sodium acetate (1:10 by volume)
1.0 mL	100% ethanol (2.5× volume)

• 13. Invert gently until DNA precipitate (5–10 min) (see Note 2).

• 14. Spool DNA using wide bore 1000 μL barrier tips and transfer to a new 1.5 mL tube while taking care not to carry over residual supernatant. Otherwise, remove as much supernatant as possible.

• 15. Add 1.5 mL cold 70% ethanol. Incubate 5–10 min on ice.

• 16. Remove supernatant and repeat 70% ethanol wash. Do not spin between washes.

• 17. Remove excess ethanol and air dry while tubes are inverted. This may take ~1 h depending on the size of the pellet.

• 18. Add 50 μL TE and incubate on ice for 30 min to resuspend DNA. Do not pipette to mix during that time as R-loops are sensitive to mechanical stress.

• 19. Using a wide bore 200 μL tip, pipette mix 3×.

• 20. Measure nucleic acid concentration using a spectrophonometer (i.e., Nanodrop). Proceed to bisulfite treatment within 24 h, keeping DNA at −20 °C (see Note 3).

3.3. Non-denaturing Bisulfite Treatment (See Note 4)

1. Pipette 1.5 μg of DNA from previous isolation bring volume to 20 μL with PCR grade water.

2. Add 130 μL Lightning conversion reagent to 20 μL template. Invert mix ~3×.

3. Incubate at 37 °C in a rotisserie oven for 2 h, protected from light.

4. Add 600 μL M-binding buffer to spin column.

5. Load 150 μL bisulfite-treated sample to column with buffer. Gently invert mix a few times.

6. Spin samples at 10,000 RCF for 30 s. Discard flow-through.

7. Add 100 μL M-wash buffer to column. Spin for 30 s.

8. Add 200 μL L-Desulphonation buffer to column.

9. Incubate for 20 min at room temperature. Then spin for 30 s. Discard flow-through.

10. Add 200 μL M-wash buffer to column. Spin for 30 s.

11. Repeat wash. Air dry for 1 min.

12. Place columns in new 1.5 mL tubes.

13. Add 12.5 μL EB to elute sample. Wait 2 min. Spin for 30 s to collect samples.

14. Quantify DNA using spectrophotometer. Recovery should be around 80%.

3.4. Site-Specific Amplification (See Note 5)

• 1. Assemble PCR reaction in 8-strip PCR tubes as follows (see Note 6):

Reagent	Vol. per reaction (μL)	Final concentration
ddH2O	14.75
5× Phusion HF/GC buffer	5.0	1×
10 mM dNTPs	0.5	200 μM
Template (10 ng/μL)	2.5	25 ng
Forward + reverse primers (10 μM each)	2.0	0.8 μM
Phusion U DNA polymerase (2 U/ μL)	0.25	0.02 U/μL
Total	25.0

For difficult templates, use reaction conditions below:

Reagent	Vol. per reaction (μL)	Final concentration
ddH2O	9.75
5× Phusion HF buffer	5.0	1×
10 mM dNTPs	0.5	200 μM
Template (10 ng/μL)	2.5	25 ng
Forward + reverse primers (10 μM each)	2.0	0.8 μM
5 M Betaine	5.0	1×
Phusion U DNA polymerase (2 U/μL)	0.25	0.02 U/μL
Total	25.0

• 2. Perform gradient PCR using program below: (see Note 7).

Stage	Temp (°C)	Time	Cycles
1. Denaturation	98	30 s	1
2. Amplification Denature	98	10 s	25–35
Annealing	50–72	30 s
Extension	72	2.5 min
3. Final extension	72	5 min	1
4. Hold	4	∞

• 3. Check PCR products via agarose gel electrophoresis. The goal is to obtain a single PCR band of the correct size (Fig. 2). If a single band cannot be achieved, we advise the redesign and optimization of primers. In the event that contaminating bands cannot be completely reduced, then the dominant correct band can be excised from gels post PCR and the DNA purified subsequently.

• 4. Amplification of converted template. Use optimized PCR conditions as determined above. Double total reaction volume to 50 to increase recovery.

• 5. Purify products using AMPure Magbeads (see Note 8).

• 6. Transfer the 50 μL PCR product from the previous step into a new 1.5 mL tube.

• 7. Add 50 μL AMPure XP beads (1×) to the template.

• 8. Gently tap to mix. Briefly spin down.

• 9. Place samples in mini tube rotator and incubate for 10 min.

• 10. Briefly spin down then place tubes on DynaMag-2 magnetic rack. Wait ~1 min to allow beads migrate to the magnet and the supernatant to clear.

• 11. Remove supernatant carefully without pulling any beads.

• 12. Add 1.5 mL 70% ethanol. Cap tubes and invert a few times while on the magnetic rack. Remove supernatant.

• 13. Repeat ethanol wash.

• 14. Air dry for ~1 min. Do not let the bead dry excessively, which is evident when the beads begin to crack.

• 15. Resuspend in 30 μL EB, remove the tubes from the magnetic rack and gently mix by tapping or flicking. Incubate with rotation for 5 min.

• 16. Place tubes back on the DynaMag-2 rack and wait ~1 min. Recover DNA in supernatant and place in new 1.5 mL tube.

• 17. Check sample concentration using a spectrophotometer.

• 18. Calculate number of molecules per amplicon:

Fig. 2.

Example gradient PCR with 2990 bp product. Agarose gel stained with 1× GelRed. Run for 20 min at 150 V

$$\text{ mole }\mathrm{dsDNA}(\text{mol})=\left(\frac{\text{ mass of dsDNA}(\text{g})}{\text{ length of dsDNA}(\text{bp})\times 650\left(\frac{\text{g}}{\text{mol}.\text{bp}}\right)}\right),$$

where the average weight of a DNA base-pair is 650 g/mol.

• 19. If pooling non-overlapping amplicons, pool equimolar amounts of each amplicon together with a total final mass no less than 500 ng for library preparation. Suggested starting for library preparation input is 500–700 ng.

3.5. SMRT-Bell Library Preparation (See Note 9)

• 1. Concentrate pooled amplicons (see Note 10).

• 2. Add 0.8× AMPure PB magnetic beads to pooled amplicons (i.e., if starting with 100 μL pooled sample, use 80 μL AMPure PB beads).

• 3. Perform ethanol purification as described in previous section for AMPure bead clean-up, steps 8–16.

• 4. Elute in 22 μL EB. Proceed to end repair step or store in −20 °C until ready to proceed to end-repair.

• 5. Assemble end-repair reaction on ice as follows:

Reagent	Volume (μL)	Final Conc.
Concentrated pooled amplicons	22.0
10× template prep buffer	3.0	1×
10 mM ATP high	3.0	l mM
10 mM dNTP	1.2	0.4 mM
20× end repair mix	l.5	1×
Total volume	30.0

• 6. Mix samples by gentle tapping or flicking the tube and briefly spin down.

• 7. Incubate samples at 25°C for 15 min in thermocycler, and then hold reaction at 4 °C.

• 8. Purify end-repaired template using 0.8× AMPure PB beads (24 μL) and elute in 32 μL EB. This is a good stopping point. End-repaired samples can be stored at −20 °C for a few days. If continuing, proceed directly to adaptor ligation through the end of library preparation.

• 9. For blunt end SMRTbell adaptor ligation assemble reaction on ice as follows to avoid adapter-adapter ligation:

Reagent	Volume (μL)	Final Conc.
Repaired template	32.0
20 μM blunt adapter	1.0	0.5 μM
Gently mix before next step
10× template prep buffer	4.0	1×
1 mM ATP low	2.0	0.05 mM
Gently mix before next step
Ligase (30 U/μL)	1.0	0.75 U/μL
Total volume	40.0

• 10. Gently flick or tap to mix and spin down briefly.

• 11. Incubate for 1 h at 25°C in a thermocycler. Hold at 4 °C for at least 1 min, up to overnight (see Note 11).

• 12. Inactivate ligase by incubating at 65 °C for 10 min in a thermocycler. Hold at 4 °C for at least 1 min. Proceed directly to exonuclease treatment.

• 13. Exonuclease treatment. This will remove unligated and damaged template.

Set reaction on ice as follows:

Reagent	Volume (μL)
Ligated template	40.0
Exo III (100 U/μL)	0.5
Exo VII (100 U/μL)	0.5
Total volume	41.0

• 14. Gently flick or tap to mix reaction then briefly spin down.

• 15. Incubate for 1 h at 37 °C then hold at 4 °C in a thermocycler for at least 1 min. Proceed directly to clean-up step.

• 16. Library Clean-up, first purification. Perform purification using 0.8× AMPure PB magnetic beads (32.8 μL).

• 17. Elute in 50 μL EB then proceed directly to second purification.

• 18. Second purification, perform the last purification using 0.8× AMPure PB magnetic beads (40 μL).

• 19. Elute in 12 μL EB. This will be the library to be sent for sequencing.

• 20. Check concentration on a spectrophotometer. Check with your PacBio sequencing provider the requirements for library submission. If the concentration is too low, repeat library preparation from pooled amplicons.

• 21. Proceed to quality check. Use 1 μL of library to run on a BioAnalyzer High-Sensitivity chip (see Fig. 3 for example).

Fig. 3.

Example BioAnalyzer trace of a SMRTbell library. Pooled amplicons range from 2938 to 4379 bp

3.6. SMRF-Seq Analysis Using Gargamel Pipeline (See Note 12)

• 1. Download SMRT Link package from https://www.pacb.com/support/software-downloads/ (see Note 13) and instructions (https://www.pacb.com/wp-content/uploads/SMRT_Link_Installation_v600.pdf). CCS generation will be performed on the command line, thus GUI interface is not necessary. Perform all system checks to be able to call CCS tool.

• 2. Load smrtlink after opening a terminal for Circular Consensus Sequence (CCS) generation.

module load smrtlink/6.0

• 3. Run CCS tool. Depending on file size, this may take several hours to several days depending on compute capacity and file size. Input file will be under Primary Analysis folder annotated as “subreads.bam”. Default parameters are --minPasses = 3 and --minPredictedAccuracy = 0.9.

ccs <input.subreads.bam> <output.fq>

• 4. Removing PCR duplicates using dedupe2.sh from BBMap package v37.90 (https://sourceforge.net/projects/bbmap/). Thresholds had been optimized for this application specifically.

dedupe2.sh in=ccs.fq out=deduplicated.fq e=30 mid=98 k=31 nam=4

• 5. Analyze data using Gargamel pipeline. This is an open source software designed to call R-loop peaks and visualize patterns of C to T conversions. General usage for this pipeline is available at: https://github.com/srhartono/footLoop/blob/master/README.md

• 6. Download Gargamel pipeline:

  git clonehttps://github.com/srhartono/footLoop

  Required packages for this pipeline are listed in the README.md document.

• 7. Map and assign reads using footLoop (Bismark) (see Note 13).

  This will require an index file and the reference genome to be mapped to (see Fig. 4 for example).

Fig. 4.

Example index file. Columns are as follows: chr, start, end, gene, 0, strand. Tab delimited and no header format

footLoop.pl −r <ccs.fq> −n <output_dir> −l <label> −i
<index.bed> −g <ref.genome.fa> −x −10 −y 10 −L 95p

• 8. Call C to T conversion tracks (footPeak) (see Note 14).

footPeak.pl −n <footLoop_output_dir> −o <output_dir> −w 20 −t 0.55–l 100

• 9. Generate peak clusters (footClust).

footClust.pl −n <footPeak_output_dir>

• 10. Visualize tracks by generating png files (footPeak_graph) (see Note 15).

footPeak_graph.pl −n <footPeak_output_dir> −r 1

(See Fig. 5 for an example)

Fig. 5.

Example footprinted region with PNG output after footPeak_graph.pl run. (a) SNRPN70 footprinted region (hg19) and DRIPc-seq data with (+) and (−) strands shown in red and blue, respectively. (b) Example output showing random 200 peaks reads on the non-template strand. (c) Example output showing random 100 reads for template strand. For panels (b) and (c), each horizontal line corresponds to one read. Red tick lines indicate converted Cs within a called peak, yellow indicates non-converted Cs, green indicates converted Cs not called a peak; gray is missing/ambiguous sequence

• 11. Generate Genome browser tracks (footPeakGTF).

footPeak_GTF.pl −n <footPeak_output_dir>

This script will generate GTF files that can be directly uploaded to USCS Genome Browser. Example output is shown in Fig. 6.

Fig. 6.

Genome Browser snapshot of same SNRPN70 terminal region in previous figure. Converted cytosines within called peaks are annotated as red tick marks on the positive strand (blue tick marks for footprints called in the negative strand). Each horizontal line is a single molecule read mapping to the region

Fig. 7.

Example target region in SNRPN70. Red bar indicates possible target region to amplify. Blue arrow indicates forward and reverse primer sites flanking peaks of DRIPc-seq signal (below)