Genome-wide mapping of embedded ribonucleotides and other non-canonical nucleotides using emRiboSeq and EndoSeq

Abstract

Ribonucleotides are the most common non-canonical nucleotides incorporated into the genome of replicating cells. They are efficiently removed by ribonucleotide excision repair initiated by Ribonuclease (RNase) H2 cleavage. In the absence of RNase H2, such embedded ribonucleotides can be used to track DNA polymerase activity in vivo. To determine their precise location in Saccharomyces cerevisiae we developed embedded Ribonucleotide Sequencing (emRiboSeq), which uses recombinant RNase H2 to selectively create ligatable 3’-hydroxyl groups, in contrast to alternative methods that utilize alkaline hydrolysis. EmRiboSeq allows reproducible, strand-specific and potentially quantitative detection of embedded ribonucleotides at single-nucleotide resolution. This protocol can be adapted for the genome-wide mapping of other non-canonical bases by replacing RNase H2 with specific nicking endonucleases, a method we term Endonuclease Sequencing (EndoSeq). With the protocol taking <5 days to complete, these methods allow the in vivo study of DNA replication and repair, including the identification of replication origins and termination regions.

INTRODUCTION

Ribonucleotides are frequently incorporated into DNA by DNA polymerases^1–9 a subject of increasing interest discussed in a number of recent reviews^10–15. Under normal conditions the cell can use their transient presence to its advantage: ribonucleotide incorporation by polymerase mu (Pol-µ) is thought to be beneficial for double-strand break repair by nonhomologous end joining^7,16, whereas ribonucleotide incorporation on the leading strand by Pol-ε promotes mismatch repair^17,18. However, failure to remove embedded ribonucleotides, i.e. absence of RNase H2, results in genome instability^6,8,19–25. This may be linked to heritable autoinflammatory disorders such as Aicardi–Goutières syndrome^21,25–27 and Systemic lupus erythematosus²¹, and could have relevance for the neurodegenerative disorder Ataxia with Oculomotor Apraxia 1^28,29. To map the genome-wide distribution of these embedded ribonucleotides we developed a next generation sequencing (NGS) method, emRiboSeq³⁰, which relies on cleavage of the phosphodiester bond directly 5’ of a DNA-embedded ribonucleotide with recombinant RNase H2. In the cell, cleavage by RNase H2 initiates the removal of misincorporated ribonucleotides, a process termed ribonucleotide excision repair (RER)³¹. Therefore, the use of RNase H2 deficient cells is essential to allow mapping of ribonucleotides. We successfully used this approach to establish that ribonucleotide incorporation patterns in the Saccharomyces cerevisiae genome are non-random, and to map the contribution of the replicative DNA polymerases³⁰ (Pol-ε, Pol-δ and Pol-α) using point mutations that alter their propensity to incorporate ribonucleotides^6,18,32. In addition, this accurately identified origins of replication and termination regions at high resolution. In parallel three other groups developed different methods to map embedded ribonucleotides, yielding similar findings^33–36 (see also further comparison below).

Overview of emRiboSeq

Here we describe the details of the emRiboSeq protocol, which could potentially be applied to any organism, although we emphasise its application to S. cerevisiae with its relatively compact genome. In brief, the protocol (Fig. 1) includes isolation and sonication of genomic DNA, followed by end repair, deoxyadenosine-tailing and ligation of fragments to adapters with a deoxythimidine overhang on one end and a non-ligatable 2’,3’-dideoxycytidine on the other. In order to minimise the contribution of nicks or double strand breaks to background, pre-existing 3’-hydroxyl (3’-OH) groups are blocked using terminal transferase and 2’,3’-dideoxyadenosine-triphosphate (ddATP). The library of ligated fragments is then treated with RNase H2 (for emRiboSeq or another appropriate endonuclease for EndoSeq), generating 3’-OH and 5’-phosphate (5’-P) groups. Phosphates are removed to increase subsequent adapter ligation efficiency, followed by denaturation and annealing of single stranded fragments to double stranded adapters with a 3’ random hexamer overhang. This allows ligation of the adapter to only those fragments with free 3’-OH ends. A conjugated biotin molecule in the random hexamer containing strand of the second adapter allows ligated fragments to be captured. Elution of the full length non-biotinylated strand is followed by second strand synthesis to produce a library that requires only size selection prior to sequencing.

Figure 1.

The emRiboSeq protocol. Schematic depicting processing of ribonucleotide (red ‘R’) and non-ribonucleotide containing DNA fragments. The first adapter (blue) attaches to both fragments, but only those that contain ribonucleotides will be captured for sequencing after ligation with the second adapter (pink). For EndoSeq, a different endonuclease is used in place of RNase H2 (Step 47). OH, hydroxyl; P, phosphate; A, deoxyadenosine; T, deoxythymidine; dd, dideoxynucleoside; Bio, biotinylated nucleoside; NNNNNN, random deoxynucleotide hexamer. Adapted from ref. 30.

Applications of the method

Genome-wide mapping of embedded ribonucleotides can offer insights into DNA replication and repair processes, as it allows in vivo tracking of DNA polymerase activities^30,33–35. In addition to the potential for identifying replication origins and termination regions, emRiboSeq will facilitate the study of the causes and consequences of ribonucleotide misincorporation. Furthermore, to examine the distribution of other non-canonical nucleotides only a single step in the emRiboSeq protocol needs to be changed. In EndoSeq, appropriate nicking endonucleases that selectively cleave and generate 3’-OH ends can be used in place of RNase H2. This could be applied to lesions such as apurinic/apyrimidinic (AP) sites (e.g. using AP endonuclease 1, APE1), embedded deoxyinosines (e.g. using E. coli Endo V) or UV-induced damage such as cyclobutane pyrimidine dimers (CPDs, e.g. using UV damage endonuclease, UVDE). In other cases, multiple enzymes may be required to generate the required 3’-OH ends, such as the detection of embedded deoxyuracil by treatment with uracil-DNA glycosylase (UDG) to generate AP sites, which could then be cleaved by AP lyase enzymes (e.g.APE1 or Endo IV). Endonucleases that generate double strand breaks can also be used, although in this case strand-specific information is lost. We established proof of principle of the EndoSeq approach using sequence-specific endonucleases³⁰. Together, emRiboSeq and EndoSeq add important tools to the molecular biology toolbox used to study DNA replication, damage and repair.

Comparison with other methods

Three other methods that allow embedded ribonucleotides to be mapped were developed independently^33–35, highlighting the current interest in studying ribonucleotide incorporation and in vivo polymerase contributions. A useful review and comparison of all four methods, emRiboSeq³⁰, HydEn-seq³³, ribose-seq^35,36 and Pu-seq³⁴, was made by Jinks-Robertson & Klein³⁷. Whereas emRiboSeq uses enzymatic hydrolysis of embedded ribonucleotides (generating 5’-P and 3’-OH ends), the other three methods make use of alkaline hydrolysis of ribonucleotide-containing DNA (generating 5’-OH and 2’,3’-cyclic phosphates). The advantage of chemical hydrolysis is the simplicity of the procedure; a disadvantage is the fact that abasic sites are also prone to hydrolysis³⁸, which may cause higher background. Post-hydrolysis, both emRiboSeq and ribose-seq capture the strand upstream of the nick, whereas HydEn-seq and Pu-seq capture the downstream fragment. The principle of specifically capturing these ends differs for each method. HydEn-seq phosphorylates the 5’-OH ends and uses this to ligate on a first adapter; a second random hexamer-containing adapter is used for second strand synthesis, followed by PCR amplification. Pu-seq uses random priming to perform second strand synthesis in the presence of deoxyuridine (dU) tri-phosphate, generating blunt ends that allow adapter ligation. The dU-containing strand is then degraded and the original strand, which now contains adapters, amplified. Ribose-seq initially ligates a two-sided adapter to the 5’-end of fragments, followed by circular ligation using Arabidopsis thaliana tRNA ligase (AtRNL), an enzyme that specifically allows single stranded ligation to 2’,3’-cyclic phosphates. Subsequent degradation of linear DNA allows specific amplification of circular fragments. Each of the four methods is likely to introduce certain biases, which future in-depth side-by-side comparison will be useful to delineate. Notably, with the exception of emRiboSeq each of these library preparation methods uses PCR pre-amplification, itself a potential source of bias.

Although these four methods achieve broadly the same outcome, the investigator may wish to consider the pros and cons of each before deciding which methodology to employ. The main strengths of emRiboSeq are that it is a robust, reproducible and validated technique, and the absence of pre-sequencing amplification means that no de-duplication is necessary during bioinformatics analysis, resulting in superior read depth. In addition, the lack of pre-amplification also gives it the potential to be used as a quantitative method, although the precision of quantitation will depend on the efficiency of all of the enzymatic steps in the library preparation (each should be performed to completion and under conditions that allow high specificity), as well as the quality of the starting material.

Potential disadvantages of our protocol are its relative length and its dependence on enzymatic cleavage, which could introduce enzyme-specific bias. In our published work we successfully employed recombinant human RNase H2 purified in our own laboratory. More recently, we have also generated emRiboSeq libraries using a commercially available preparation of RNase HII (New England Biolabs) with preliminary analyses showing that these produce similar data, indicating that type 2 RNase H enzymes from other sources can be used. However, careful experimental design and interpretation when using nucleases is important, and potential site specific variation in activity or risk of low level contaminating nuclease activity in both commercial and lab generated enzymes should be kept in mind. Optimal conditions for emRiboseq are those under which all ribonucleotides are cleaved whilst minimising non-specific cleavage. While the use of alkaline hydrolysis in other methods may result in detection of sites additional to ribonucleotides (e.g. abasic residues), this will also circumvent potential problems such as sequence-specific biases in cleavage or contaminating nucleases.

Our method can be adapted to use endonucleases other than RNase H2 to study different non-canonical nucleotides. The EndoSeq method can be compared to Excision-seq, an alternative method that makes use of enzymatic cleavage at modified bases to determine their genome-wide distribution³⁹. However, this method requires the modifications under investigation to be present at a high enough frequency to generate sufficiently small fragments. This is not a limitation for EndoSeq, as fragment size is determined by initial sonication, although the presence of modifications at too high a frequency could be problematic as discussed below.

Experimental Design

Sequencing platform

This protocol was developed using the Ion Torrent™ NGS platform (Life Technologies), but can be adapted to other platforms by changing the adapter sequences, which in our case were designed for compatibility with the capture and sequencing primers used for Ion Torrent™. The optimum length of fragments within a NGS library also differs according to the platform used and consequently the size selection range (Step 76) would also require adjustment. When applying our method to organisms with a substantially larger genome than that of S. cerevisiae read depth may become an issue, in which case the use of a sequencing methodology that generates larger numbers of reads will need to be considered.

EndoSeq: Use of alternative endonucleases

When substituting RNase H2 with alternative endonucleases to determine the genomic distribution of non-canonical nucleotides other than ribonucleotides there are several important considerations. The specificity of the endonuclease is crucial and where possible non-specific activity should be carefully avoided. It is therefore recommended that optimum conditions be established for Step 47. High frequency cleavage (i.e. >1 cut every 400 basepairs, bp) is not desirable as fragments that are cleaved multiple times are likely to be lost during final size selection. In this case, partial digestion may be possible, but this is likely to introduce bias, with sites that are preferentially cleaved overrepresented. Cleavage frequency of a particular nicking endonuclease can be assessed by alkaline gel electrophoresis of enzyme-treated DNA, followed by densitometry^8,30. Existing estimates of ribonucleotide incorporation rates by S. cerevisiae polymerases^5,6,30 mean that multiple cleavage of sonicated fragments should be a rare event in the emRiboSeq protocol, although the occurrence of closely spaced ribonucleotides at specific positions cannot be ruled out. EmRiboSeq would not detect these with only the most 5’ site captured.

Starting material

Isolation of genomic DNA (gDNA) is performed under conditions that minimise the generation of nicks. Where the endonuclease substrate is rapidly removed by endogenous processes it will be necessary to isolate gDNA from cells in which these processes are inactive, either through genetic (knock-out or knock-down) or chemical means (inhibitors). Although other, less efficient mechanisms exist for the removal of embedded ribonucleotides^40–42, the use of RNase H2 null S. cerevisiae in emRiboSeq was shown to be sufficient to allow mapping of ribonucleotides misincorporated by replicative DNA polymerases³⁰. For similar experiments, the use of strains in which at least one of the genes encoding the RNase H2 subunits, RNH201, RNH202 or RNH203, is deleted is therefore recommended. We have used both mid-log phase (A_{600 nm} = 0.5) and stationary phase cultures (A_{600 nm} = 5 to 6) with similar results.

The emRiboSeq protocol was implemented in S. cerevisiae, which offers a number of advantages including ease of genetic engineering and a small, well defined genome. Additionally, RNase H2 deficient yeast proliferates at a normal rate, unlike RNase H2 null mammalian cells, although absence of RNase H2 activity in budding yeast does cause mutation rate increase and genome instability^{6,19,24,42–45}.

If using gDNA from a source other than S. cerevisiae, Steps 1-15 should be replaced with an appropriate protocol for isolating gDNA in which cleavage at embedded ribonucleotides is avoided prior to Step 47. In our protocol this was ensured by performing RNase treatment under high salt conditions (Step 15). Because our protocol does not use pre-amplification a large quantity of starting material is crucial: we recommend starting with 18 µg of DNA, so that 5 µg can be used as input for end repair (Step 33), although we have been successful with input amounts as low as 1 µg.

Controls

It is recommended that control experiments are performed when first carrying out this protocol, especially when using a new endonuclease. These include the preparation of an ‘endonuclease negative’ library where endonuclease treatment (Step 47) is omitted. This should result in the absence of 3’-OH groups available for ligation to the second adapter (Steps 53-55) and, therefore, a lack of amplification when using PCR to verify the quality of the library prior to sequencing (Steps 81-83). A ‘positive’ control experiment that allows the specificity of generated libraries to be estimated can be achieved through substitution of the endonuclease for one with well-defined sequence specificity. We originally performed such controls³⁰ using BciVI (New England Biolabs), which cleaves both strands, and the nicking endonuclease Nb.BtsI (New England Biolabs). The preparation of control libraries using sequence specific endonucleases provides the additional advantage of enabling the bioinformatic processing of sequence data to be optimised using a library with highly predictable results. It should, however, be noted that due to the sensitivity of the EndoSeq method even low levels of star activity can be observed.

MATERIALS

Reagents

For gDNA isolation from S. cerevisiae

• RNase H2 deficient S. cerevisiae, such as SNM106 (a kind gift from T. Kunkel) or its derivatives^6,30 (5 A_{600 nm} unit pellets can be stored at -80°C).

• Glass beads, Ø 0.4-0.6 mm (Sartorius, cat. no. BBI-8541701)

• 10% (vol/vol) Triton X-100 (Sigma, cat. no. T8787) CAUTION Triton X-100 is corrosive. Wear personal protective equipment.

• 20% (wt/vol) Sodium dodecyl sulfate (SDS, Sigma, cat. no. 05030) CAUTION SDS is corrosive and toxic. Wear personal protective equipment and avoid inhalation.

• Sodium Chloride (NaCl, Sigma, cat. no. S3014)

• Tris(hydroxymethyl)methylamine (Tris base, Sigma, cat. no. T4661)

• Boric acid (Sigma, cat. no. B7901)

• Ethylenediaminetetraacetic acid (EDTA, Sigma, cat. no. E6758)

• Phenol pH 7.9 (Sigma, cat. no. P4557) CAUTION Phenol is corrosive and toxic. Wear personal protective equipment and handle it in a fume hood.

• Phenol:chloroform:isoamylalcohol (25:24:1) (Sigma, cat. no. P3803) CAUTION Phenol:chloroform:isoamylalcohol is corrosive and toxic. Wear personal protective equipment and handle it in a fume hood.

• Chloroform (Sigma, cat. no. C2432) CAUTION Chloroform is toxic. Wear personal protective equipment and handle it in a fume hood.

• 100% Ethanol (EtOH, Sigma, cat. no. E7023) CAUTION Ethanol is highly flammable. Handle away from potential sources of ignition.

• Nuclease free distilled water (H₂O, Life-Technologies, cat. no. 10977)

• RNase, DNAse-free (Roche, cat. no. 11 119 915 001)

• Poly(ethylene glycol) 8000 (PEG_8,000, Sigma, cat. no. 89510)

• Agencourt Ampure XP beads (Beckman Coulter, cat. No. A63880)

For library preparation from gDNA

• Agarose (electrophoresis grade, Sigma, cat. no. A9539)

• SYBR^® Gold nucleic acid stain, 10,000x in DMSO (Life Technologies, cat. No. S-11494)

• 100 bp DNA ladder (e.g. Promega, cat. no. G210A, supplied with Blue/Orange 6X Loading Dye)

• NEBNext^® end repair module (New England Biolabs, cat. no. E6050)

• NEBNext^® dA-tailing module (New England Biolabs, cat. no. E6053)

• NEBNext^® Quick Ligation module (New England Biolabs, cat. no. E6056)

• Custom oligonucleotides; ours were synthesised by Eurogentec and purified using manufacturer recommended methods (Table 1).

• Terminal Transferase, supplied with 10x reaction buffer and 2.5 mM CoCl₂ (New England Biolabs, cat. no. M0315)

• 10 mM ddATP, Sequencing Grade Na-salt pH 8.3 (Roche, cat. no. 03 732 738 001)

• RNase HII (New England Biolabs, cat. no. M0288) or recombinant human RNase H2 (purified as previously described⁴⁶) for emRiboSeq; other recombinant (nicking) endonuclease of interest for EndoSeq. CRITICAL: alternative sources of recombinant type 2 RNase H enzymes may be used

• 10% (wt/vol) Bovine Serum Albumin Fraction V (BSA, Roche, cat. no. 10 735 086 001)

• Magnesium chloride (MgCl₂, Sigma, cat. no. M2670)

• Nb.BtsI, Supplied with 10x CutSmart^® buffer (New England Biolabs, cat. no. R0707)

• BciVI, Supplied with 10x CutSmart^® buffer (New England Biolabs, cat. no. R0596)

• Shrimp Alkaline Phosphatase (SAP), supplied with 10x reaction buffer (Affymetrix USB^®, cat. no. 70092Z)

• Dynabeads^® M-280 Streptavidin (Life Technologies, cat. no. 11205D)

• Tri-sodium citrate (Sigma, cat. no. C8532)

• Glycogen (Roche, cat. no. 10 901 393 001)

• Sodium acetate (NaOAc, Sigma, cat. no. S2889)

• Sodium hydroxide (NaOH, Sigma, cat. no. 38215) CAUTION Sodium hydroxide is corrosive. Wear personal protective equipment.

• Phusion Flash High-Fidelity PCR Master Mix (Thermo Scientific, cat. no. F-548)

• E-Gel^® SizeSelect™ Agarose Gels, 2% (Life Technologies, cat. no. G6610-02)

• Agilent High Sensitivity DNA Kit (Agilent Technologies, cat. no. 5067-4626)

• Potassium Chloride (KCl, Sigma, cat. no. P9541)

• Ion 318™ Chip Kit v2 (Life Technologies, cat. no. 4484534) or Ion PI™ Chip Kit v2 (Life Technologies, cat. no. 4482413)

• Ion PGM™ Template OT2 200 Kit (Life Technologies, cat. no. 4480974) or Ion PI™ Template OT2 200 Kit v2 (Life Technologies, cat. no. 4485146)

• Ion PGM™ Sequencing 200 Kit v2 (Life Technologies, cat. no. 4482006) or Ion PI™ sequencing kit (Life Technologies, cat. no. 4485149). CRITICAL: These Ion Torrent sequencing products are under constant development. We therefore suggest using the most appropriate equivalent recommended products.

Table 1. Oligonucleotides used in library preparation and quality control.

All sequences are given 5’ to 3’.

Oligonucleotide	Sequence
trP1-top	CCTCTCTATGGGCAGTCGGTGAT-phosphorothioate-T
trP1-bottom	phosphate-ATCACCGACTGCCCATAGAGAGGC-dideoxy
A-top	phosphate-CTGAGTCGGAGACACGCAGGGATGAGATGG-dideoxy
A-bottom	biotin-CCATCTCATCCCTGCGTGTCTCCGACTCAGNNNNNN-C3-phosphoramidite
primer A	CCATCTCATCCCTGCGTGTCTCCGAC
primer trP1	CCTCTCTATGGGCAGTCGGTGATT

Equipment

• Nuclease-free 1.5 ml tubes (Starlab, cat. no. E1415-1500)

• Vortex MS 3 basic (IKA, cat. no. 0003617000)

• Table-top centrifuge (Eppendorf, cat. no. 5424 R)

• Agarose gel electrophoresis equipment

• DynaMag™-2 Magnet, magnetic stand (Life Technologies, cat. no. 12321D)

• Nanodrop spectrophotometer (Thermo Scientific)

• Bioruptor^® Plus with 1.5 ml tube holder (Diagenode, cat. no. B01020001)

• Rotator (IKA, cat. no. 0004015000)

• Thermocycler (Biorad, cat. no. C1000)

• E-Gel^® iBase™ Power System (Life Technologies, cat. no. G6400)

• 2100 Electrophoresis Bioanalyzer Instrument (Agilent Technologies, cat. no. G2939AA)

• Ion OneTouch™ 2 System (Life Technologies, cat. no. 4474779)

• Ion PGM™ Sequencer (Life Technologies, cat. no. 4462921) or Ion Proton™ Sequencer (Life Technologies, cat. no. 4476610)

Reagent Setup

Solutions

CRITICAL: Prepare all of the following solutions using nuclease-free water.

• Lysis buffer (2% Triton X-100, 1% SDS, 0.5 M NaCl, 10 mM Tris, 1 mM EDTA, pH8.0; store at room temperature (RT) for up to six months)

• TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0; store at RT for up to six months)

• 0.5x TBE (44.5 mM Tris, 44.5 mM Boric acid, 1 mM EDTA, pH 8.0; store at RT for up to six months)

• 0.5 M NaCl (store at RT for up to six months)

• 75% (vol/vol) EtOH (store at RT for up to six months)

• 3 M NaOAc pH 5.2 (store at RT for up to six months)

• 30% (wt/vol) PEG_8,000, 3.75 M NaCl (filter sterilise using a 0.22 µm filter and store at 4°C for up to six months; warm to RT before use)

• 5x Oligonucleotide annealing buffer (300 mM KCl, 250 mM Tris, pH 8.0; store at RT for up to six months)

• 2x RNase H2 reaction buffer for use with recombinant human RNase H2⁴⁶ (120 mM KCl, 100 mM Tris-HCl pH 8.0, 20 mM MgCl₂, 0.02% BSA, 0.02% Triton X-100; store at 4°C for up to six months)

• 0.15 M NaOH (make up fresh)

• Saline sodium citrate (SSC) (0.15 M NaCl, 0.015 M sodium citrate, pH 7.0; store at RT for up to six months)

• 1x Bind and Wash buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA, 2 M NaCl; store at RT for up to six months)

• 2x Bind and Wash buffer (20 mM Tris-HCl pH 7.5, 2 mM EDTA, 4 M NaCl; store at RT for up to six months)

Double stranded adapters

Prepare the first adapter (dstrP1) by combining oligonucleotides trP1-top and trP1-bottom, and the second adapter (dsA) by combining A-top and A-bottom (see Table 1 for oligonucleotide sequences).

In each case, prepare 100 µl 40 µM double stranded adapter as follows: Combine 40 µl of each single stranded oligonucleotide (100 µM) with 20 µl 5x oligonucleotide buffer, incubate at 95°C for 5 min in a thermocycler, before removing and allowing to cool gradually to RT (store at -80°C for up to one year).

PROCEDURE

gDNA isolation from yeast TIMING 2-3 h

• 1Add 200 μl of lysis buffer and ˜0.2 ml of glass beads to the pellet from a yeast culture (5 A_{600 nm} units), in a 1.5 ml tube. To isolate sufficient gDNA for a particular sample, we usually carry out Steps 1-14 with 8 tubes per sample (i.e. a total of 40 A_{600 nm} units). These can be pooled in Step 14, and subsequent steps carried out for multiple samples at once, making appropriate adjustments for the larger volume. Amounts given in Steps 1-23 are for a single tube containing 5 A_{600 nm} units.
• 2Add 200 μl of phenol and vortex for 2 min (continuous mixing at 3,000 rpm).
• 3Add 200 μl of TE and vortex for 1 min (continuous mixing at 3,000 rpm).
• 4Centrifuge at 5,000 g for 5 min, 4°C.
• 5Transfer the aqueous phase to a new 1.5 ml tube, add 400 μl of phenol:chloroform:isoamylalcohol (25:24:1) and vortex for 30 s.
• 6Centrifuge for 5 min at 5,000 g, 4°C.
• 7Add 400 μl of chloroform and vortex for 30 s.
• 8Centrifuge at 5,000 g for 2 min, 4°C.
• 9Transfer the aqueous phase to a new 1.5 ml tube, add 1 ml of ice-cold 100% EtOH and mix thoroughly by repeated inversion.
• 10Centrifuge at ≥ 13,000 g for 10 min, 4°C and remove the supernatant.
• 11Wash the pellet with 0.5 ml 75% EtOH.
• 12Centrifuge at ≥ 13,000 g for 10 min, 4°C and remove the supernatant.
• 13Air-dry the pellet at RT.
• 14Resuspend the pellet containing total nucleic acids in 50 µl of 0.5 M NaCl. CRITICAL STEP: Replicates can be pooled at this stage (making appropriate adjustments for the larger volume in subsequent steps).
• 15Add 1 µl of DNase-free RNase (0.5 µg/µl) and incubate for 1 h at RT.

DNA purification

• 16Add 50 µl of Ampure XP after allowing it to equilibrate to RT for at least 5 min (1x volume).
• 17Mix thoroughly and incubate for 1 min at RT. If necessary centrifuge briefly to collect droplets.
• 18Place on a magnetic stand for >1 min. The beads will collect near the magnet and the solution will become clear. Remove and discard the supernatant.
• 19Without removing the tube from the stand, wash twice with 500 µl 75% EtOH. Mixing can be achieved by rotating the tube on the stand. Remove and discard the supernatant.
• 20Ensure the beads are dry before resuspending in the required volume and diluent as detailed in the next step. This can take a few min at RT.
• 21Elute DNA in 50 µl H₂O, by vortexing to resuspend and incubating at RT for 1 min.
• 22Using the magnetic stand to retain the beads, transfer the eluate to a new 1.5 ml tube and discard the beads.
• 23Measure the concentration of eluted gDNA using 1.5 µl on a nanodrop spectrophotometer. Usually ˜3 µg of DNA can be expected per 5 A_{600 nm} units culture. CRITICAL STEP: If visualised on an agarose gel an additional band, which migrates at around 4.5 kb, is present. This appears to be RNA resistant to RNase cleavage at 0.5 M NaCl, but sensitive to subsequent degradation by RNase A at low salt concentration.

PAUSE POINT

gDNA can be stored at -80°C for at least 7 d (storage at -80°C is recommended to prevent any unwanted RNase activity on embedded ribonucleotides).

Sonication TIMING 2-3 h

• 24Prepare 300 µl aliquots of 10 ng/µl DNA in H₂O in 1.5 ml tubes. Ideally, 6 aliquots (18 µg) of DNA should be sonicated per library, however using as few as 3 aliquots can give enough material to proceed.
• 25Sonicate each aliquot to achieve an average length of ˜400 bp using a Bioruptor^® Plus, following the manufacturers recommendations.
• 26Confirm the distribution of fragments by gel electrophoresis: Load 10 µl per aliquot alongside 100 bp DNA ladder on a 0.5x TBE, 1% agarose gel. Electrophorese and visualise using a nucleic acid stain (e.g. SYBR^® Gold). CRITICAL STEP The distribution of fragment sizes will have a strong influence on the concentration of the library following size selection. In our experience, an average length of ˜400 bp can be achieved with 18 cycles of 30 s sonication and 60 s cooling on the ‘high’ setting, however this is likely to require optimisation in each lab. Once the desired distribution of fragments is achieved, ethanol precipitate the DNA. Assuming 290 µl per tube, add 1 µl glycogen, 29 µl 3M NaOAc and 800 µl 100% EtOH.
• 27Precipitate for 1 h at -20°C and centrifuge at ≥ 13,000 g for 30 min, 4°C and remove the supernatant.
• 28Wash the pellet with 0.5 ml 75% EtOH.
• 29Centrifuge at ≥ 13,000 g for 2 min, 4°C and remove all supernatant.
• 30Air dry pellet, dissolve and pool pellets from all aliquots of each library in a total volume of 100 µl H₂O.
• 31Purify DNA by repeating Steps 16-23, using 120 µl Ampure XP (1.2x volume) and eluting in 86.5 µl H₂O.

  CRITICAL STEP The large amount of small fragments generated during sonication can lead to a loss of up to 70% of the initial DNA. Whilst we have prepared libraries using as little as 1 μg, we recommend taking through 5 μg of sonicated DNA to the next steps. Subsequent steps, including reaction and purification conditions, as well as elution volumes, assume 5 μg of sonicated DNA is used and should be scaled accordingly if this is not the case.

PAUSE POINT

Sonicated DNA can be stored at -80°C for at least 7 d.

End repair, dA-tailing and adapter ligation TIMING ~3 h + overnight incubation

• 32To 5 µg of sonicated DNA in 85 µl H₂O add 10 µl 10x buffer and 5 µl enzyme mix from the NEBNext^® end repair module.
• 33Incubate for 1 h at 20°C.
• 34Purify DNA by repeating Steps 16-21, using 120 µl Ampure XP (1.2x volume) and eluting in 42 µl H₂O.

  .CRITICAL STEP: Do not separate from the beads as they can be reused and do not interfere with the dA-tailing reaction.

• 35Add 5 µl 10x buffer and 3 µl Klenow Fragment (3´→ 5´ exo^–) from the NEBNext^® dA-tailing module.
• 36Incubate for 1 h at 37°C.
• 37Add 25 µl 30% PEG_8,000, 3.75 M NaCl (0.5x volume).
• 38Purify DNA by repeating Steps 17-21, eluting in 20 µl H₂O.

  CRITICAL STEP: Do not separate from the beads as they can be reused and do not interfere with adapter ligation.

• 39Add 10 µl 5x buffer, 15 µl 40 µM dstrP1 adapter and 5 µl T4 Quick Ligase from the NEBNext^® quick ligation module.
• 40Incubate overnight or for at least 12 h at 16°C. Ligation for shorter times may be possible, but we have not tested this.
• 41Add 25 µl 30% PEG_8,000, 3.75 M NaCl (0.5x volume). CRITICAL STEP: 1x NEBNext Ligation buffer contains 6% PEG_6,000 and may therefore alter the size selection effect, however this nucleation procedure has proved sufficient to significantly deplete the adaptors.
• 42Purify DNA by repeating Steps 17-22, eluting in 195 µl H₂O.

3’ blocking and endonucleolytic treatment TIMING ~4.5 h

• 43Add 25 µl 10x Terminal Transferase reaction buffer, 25 µl 2.5 mM CoCl₂, 2.5 µl 10 mM ddATP, 2.5 µl 20 U/µl Terminal Transferase.
• 44Incubate at 37°C for 2 h.
• 45Purify DNA by repeating Steps 16-23, using 300 µl Ampure XP (1.2x volume) and eluting the 3’ blocked DNA in 51.5 µl H₂O.

  Use the obtained concentration to scale subsequent reactions. Normally 2-3 μg of DNA remains at this point. CRITICAL STEP Subsequent steps, including reaction and purification conditions, as well as elution volumes, assume 2 μg of DNA is used and should be scaled accordingly if this deviates significantly.

• 46Endonuclease treatment: This should be optimised for each new endonuclease. The following reaction conditions were successfully employed for Escherichia coli RNase HII (option A), human RNase H2 (option B; note that alternative sources of type 2 RNase H enzymes may be used) and Nb.BtsI (option C).
  • ARNase HII treatment
    • iTo 2 μg of 3’ blocked DNA in 50 µl add 10 µl 10x reaction buffer (provided), 50 U RNase HII and H₂O up to 100 µl.
    • iiIncubate for 2 h at 37°C and then inactivate by incubating for 20 min at 80°C.
  • BRNase H2 treatment
    • iTo 2 μg of 3’ blocked DNA in 50 µl add 50 µl 2x RNase H2 reaction buffer and 20 pmol RNase H2. CRITICAL STEP: This large excess of enzyme was used to ensure complete digestion, but lower amounts are likely sufficient.
    • iiIncubate for 2 h at 37°C and then inactivate by incubating for 20 min at 80°C.
  • CNb.BtsI treatment
    • iTo 2 μg of 3’ blocked DNA in 50 µl add 10 µl 10x CutSmart^® buffer, 1 µl 100x BSA, 0.5 µl 10 U/µl Nb.BtsI and 28.5 µl H₂O.
    • iiIncubate for 1 h at 37°C and then inactivate by incubating for 20 min at 80°C. CRITICAL STEP: This enzyme exhibits star activity, particularly when high concentrations of enzyme or long incubation times are used.

PAUSE POINT

Endonuclease treated samples can be stored at -20°C for at least 7 d.

Dephosphorylation and second adapter ligation TIMING 2 h + overnight incubation

• 47Purify DNA by repeating Steps 16-21, using 120 µl Ampure XP (1.2x volume) and eluting in 85 µl H₂O.

  CRITICAL STEP: Do not separate from the beads as they can be reused and do not interfere with the de-phosphorylation reaction.

• 48Add 10 µl 10x SAP reaction buffer and 5 µl 1 U/µl SAP.
• 49Incubate at 37°C for 1 h and then inactivate by incubating for 15 min at 65°C.
• 50Add 50 µl 30% PEG_8,000, 3.75 M NaCl (0.5x volume).
• 51Purify DNA by repeating Steps 17-22, eluting in 29 µl H₂O.
• 52Incubate the library at 95°C for 5 min in a thermocycler and then snap cool in an icy water bath. CRITICAL STEP Libraries must be denatured immediately prior to ligation.
• 53Once cooled, immediately add 10 µl 5x buffer, 6 µl 40 µM dsA adapter and 5 µl T4 Quick Ligase from the NEBNext^® quick ligation module.
• 54Incubate overnight or for at least 12 h at 16°C. CRITICAL STEP: Ligation for shorter times may be possible, but we have not tested this.
• 55Purify DNA by repeating Steps 16-22, using 90 µl Ampure XP (1.8x volume) and eluting in 40 µl H₂O.

PAUSE POINT

Eluted DNA fragments can be stored at -20°C for at least 7 d.

Single stranded library preparation, second strand synthesis and size selection TIMING 4-5 h

• 56Resuspend Dynabeads^® M-280 Streptavidin and transfer 20 µl to a new 1.5 ml tube.
• 57Using a magnetic stand to retain the beads, discard the supernatant and replace with 1 ml 1x Bind and Wash buffer.
• 58Using a magnetic stand to retain the beads, discard the supernatant and replace with 40 µl 2x Bind and Wash buffer.
• 59Add the whole 40 µl of adapter-ligated library from Step 56 and incubate at RT for 15 min on a rotator.
• 60Using a magnetic stand to retain the beads, discard the supernatant and replace with 50 µl SSC.
• 61Incubate at RT for 5 min on a rotator.
• 62Repeat Steps 61-62 to ensure no unbound fragments are carried through.
• 63Using a magnetic stand to retain the beads, discard the supernatant and replace with 40 µl 0.15 M NaOH.
• 64Incubate at RT for 10 min on a rotator.
• 65Using a magnetic stand to retain the beads, transfer the eluate to a new 1.5 ml tube.
• 66Repeat the elution: add 40 µl 0.15 M NaOH to the beads from Step 66, incubate at RT for 5 min on a rotator, use a magnetic stand to retain the beads and pool the eluate from this step with that collected in Step 66.
• 67Add 120 µl H₂O and ethanol precipitate by adding 1 µl glycogen, 20 µl 3 M NaOAc and 550 µl 100% EtOH.
• 68Precipitate and wash by repeating Steps 28-30.
• 69Dissolve the pellet containing the single stranded library in 9.5 µl H₂O.
• 70Add 0.5 µl 10 µM primer A and 10 µl 2x Phusion Flash Master mix.
• 71In a thermocycler, incubate at 98°C for 1 min, followed by 58°C for 30 seconds, and 72°C for 1 min.
• 72Purify DNA by repeating Steps 16-22, using 36 µl of Ampure XP (1.8x volume) and eluting in 20 µl H₂O.
• 73Set up a 2% E-Gel^® SizeSelect™ agarose gel on the iBase™ Power System according to manufacturer’s instructions.
• 74Load the library directly into the desired well; loading dye is unnecessary. Use 100 bp DNA ladder as a marker.
• 75Once the 200 bp fragment has entered the collection well start collecting the library. Pause the electrophoresis every five seconds, mix and remove the liquid in the collection well, transferring it to a 1.5 ml collection tube, then replenish the well (20 µl H₂O). Stop once the 300 bp fragment has completely entered the collection well. A total of ˜500 µl is usually collected per sample. Alternatively, the sample can be separated by conventional agarose gel electrophoresis and fragments between 200 and 300 bp purified by gel extraction, e.g. we have successfully applied the QIAquick Gel Extraction Kit (QIAGEN, cat.no. 28704) following the manufacturer’s instructions (if this alternative is used, proceed to Step 79).
• 76Assuming 500 µl has been collected, divide size-selected DNA into 2 x 250 µl in 1.5 ml tubes and ethanol precipitate by adding 25 µl of 3 M NaOAc and 700 µl 100% EtOH.
• 77Precipitate and wash by repeating Steps 28-30.
• 78Dissolve each pellet in 20 µl H₂O and purify by repeating Steps 16-22, using 36 µl Ampure XP (1.8x volume) and eluting in 15 µl H₂O.

  CRITICAL STEP Contaminants, such as carried over Ampure beads, can disrupt library quantification using the Bioanalyzer.

PAUSE POINT

Prepared libraries can be stored at -20 °C overnight, or if necessary, for up to 7 d.

Library quality control and sequencing

• 79Transfer 0.5 μl to a new tube for further library verification by PCR. Dilute this aliquot 1 in 20, by adding 9.5 μl H₂O.
• 80Set up three PCR reactions per library, each using 0.5 μl of the 1 in 20 library dilution, 0.5 μl 10 μM primer trP1, 0.5 μl 10 μM primer A, 8.5 H₂O and 10 μl 2x Phusion Flash master mix.
• 81Using a thermocycler incubate the reactions at 98°C for 10 sec, followed by 15 cycles for one reaction, 16 for the second and 17 for the third, with each cycle 1 s at 98°C, 5 s at 58°C and 5 s at 72°C; finally, incubate at 72°C for 1 min. ?Troubleshooting
• 82Load 10 µl of each reaction alongside 100 bp DNA ladder on a 0.5x TBE, 1% agarose gel. Electrophorese and visualise using a nucleic acid stain, allowing a qualitative as well as semi-quantitative assessment: Successfully generated libraries yield a smear between ˜200 and 300 bp that is usually visible after 15 cycles of amplification.
• 83Following the manufacturer’s protocol, confirm the size distribution of the library and estimate its concentration using a Bioanalyzer and High Sensitivity DNA Kit (Agilent). ?Troubleshooting
• 84Assuming the Bioanalyzer trace shows a quantifiable distribution of material between ˜200 and 300 bp, which is confirmed by PCR, the library can then be prepared for Ion PGM™ or Ion Proton™ sequencing following the manufacturer’s protocol. CRITICAL STEP: The Bioanalyzer trace and PCR amplification from a typical library are shown in Fig. 2b and c. The required input for Ion OneTouch™ emulsion PCR is 100 μl of 12 pM for the Ion Proton™ and 26 μl of 25 pM for the Ion PGM™. We recommend using the Ion PI™ Chip to achieve the maximum number of reads. Alternatively the Ion 318™ Chip can be used.

Figure 2.

Library quality control and anticipated results. (a) Sonicated DNA separated by agarose gel electrophoresis (Step 25) shows an average fragment size of approximately 400 bp. (b) Bioanalyzer result (Step 84) for an emRiboSeq library shows a typical trace (left) and gel-like image (right) with a peak for fragments between ˜180 and ˜300 bp in size (black bar). Standards (green and purple bars) of defined size and amount allow quantification. FU, arbitrary fluorescence units. (c) Agarose gel electrophoresis of PCR products after 15, 16 and 17 cycles of amplification (Steps 81-83) of the same library shows product between 200 and 300 bp in size. (d) Sequencing results for libraries generated using Nb.BtsI are highly reproducibility between different strains (POL, wildtype polymerase; pol1-L868M, increased Pol-α ribonucleotide incorporation) after normalizing read counts to sequence tags per million (TPM). The majority of bona fide Nb.BtsI sites were present at maximal frequency, although some sites were present at lower frequencies. This is the result of partial loss during size selection because of their close proximity to other cleavage sites, a highly reproducible finding between independent libraries (Spearman's rho=0.82, p <2.2-e16). Due to star activity Nb.BtsI-like sites were also detected. Other libraries prepared using BciVI restriction enzyme digestion did not show such star activity (data not shown), and allowed calculation of the site specificity for the method (>99.9%). (e) Summed signal at Nb.BtsI sites (dark blue, correct strand; light blue, opposite strand) highlights the strand specificity (>99.9%) and single nucleotide resolution (>99%) of this method. Panels d and e adapted from ref. 30.

Bioinformatics

• 85Align sequence reads to the appropriate genome (we use sacCer3) with bowtie2 (we used version 2.0.0). Command: bowtie2 -x sacCer3 -U runName.fastq runName.sam
• 86
  Perform alignment quality filtering and format conversion using Samtools (we used version 0.1.18) and BEDTools (we used version 2.16.2). A Bam file (denoted runName.master.bam below) containing all mapped and unmapped reads is a convenient way to store the complete raw sequence and quality information, as well as the alignment to the reference genome. Samtools view with the –q 30 option only retains reads with a mapping quality score >30 (misalignment probability < 0.001) for analysis. The ribonucleotide incorporation site is one nucleotide upstream and on the opposite strand to the mapped read 5' end, a transformation that can be achieved with the short Perl code illustrated below. All command syntaxes shown assume a Bash shell and standard Unix tools in addition to the specific software mentioned above. Commands:

  samtools view -b -S runName.sam > runName.master.bam
  samtools view -b -q 30 -S runName.sam > runName.bam
  bamToBed -i runName.bam > runName.bed
  perl -alne \\
     ‘if($F[5]=˜s/\+/-/){$F[2]=$F[1];$F[1]--;} \\
     else{$F[5]=˜s/-/\+/;$F[1]=$F[2];$F[2]++} \\
     print join “\t”, @F’ \\
     < runName.bed | sort –k1,1 –k2,2n –k 6 > runName.ribo.bed

• 87
  To facilitate comparison between libraries of differing read depth, normalise read counts to sequence tags per million (TPM =(fpe/total_fpe)*1×10⁶, where fpe is 5’ end reads at site and total_fpe is total number genome mapped 5’ ends). Conversion to ribonucleotide counts per site (and strand) and the TPM normalisation can be achieved with the following commands:

  bedtools groupby -i runName.ribo.bed -g 1,3,6 -c 1 -full \\
          -o count > runName.counts.bed
  export totalfpe=`wc -l runName.ribo.bed | awk ‘{print $1}’`
  cat runName.counts.bed | perl -alne \\
          ‘$F[6]=($F[6]/$ENV{totalfpe})*1e6; print join “\t”, @F;’ \\
          > runName.tmp.bed

• 88Subsequent analysis can be performed in the R statistical environment (or Excel) by importing the tab delimited Bed file runName.tmp.bed. Such files can also be loaded into common genome browsers for visualisation and provided as “processed data files” to the Gene Expression Omnibus (GEO) and similar repositories along with the raw data (fastq file). CRITICAL STEP: Automation of emRiboSeq sequence processing (Steps 86-88) can be achieved using the script found at https://github.com/taylorLab/LaggingStrand/blob/master/emRiboSeqProcessor.sh with further details available in ref. 30 and examples of code performing analysis at https://github.com/taylorLab/LaggingStrand/

Timing

Day 1

• Steps 1-23, gDNA isolation from yeast: 2-3 h.

Day 2

• Steps 24-32, sonication: 2-3 h.

• Steps 33-41, end repair, dA-tailing and adapter ligation: ˜3 h + overnight incubation.

Day 3

• Steps 42-47, 3’ blocking and endonucleolytic treatment: ˜4.5 h.

• Steps 48-55, dephosphorylation and second adapter ligation: ˜2 h + overnight incubation.

Day 4

• Steps 56-79, single stranded library preparation, second strand synthesis and size selection: 4-5 h.

• Steps 80-84, library quality control (˜2h).

Day 5

• Steps 85-89, sequencing followed by bioinformatics processing of data (several days).

Several potential pause points have been highlighted within the protocol. If necessary, it is also possible to pause at many other points. However, it should be noted that Ampure XP beads should be removed prior to freezing and reintroduced before the next purification step.

TROUBLESHOOTING

Step	Problem	Possible reason	Possible solution
82	Amplification of ‘endonuclease negative’ control library	Incomplete removal of 3’-OH groups	Ensure that the 3’ blocking reaction conditions and timing are optimal for the terminal transferase used (Step 44).
84	Poor bioanalyzer trace	Carry over of Ampure beads or other contaminant	Additional purification (a further round of Ampure Bead purification or spin column based purification) may help, but is not guaranteed to improve the trace. N.B. we have successfully generated high quality sequencing data from samples that failed to produce good Bioanalyzer traces. In these cases we estimated the concentration based on the semi-quantitative PCR performed in Step 82, and comparing it with results from a previous library.

ANTICIPATED RESULTS

EmRiboSeq libraries generated using this protocol typically had a concentration of 400-2,000 pM. The total number of reads for the Ion 318™ Chip was typically 4-6 million, and an order of magnitude higher for the Ion PI™ Chip at 50-90 million reads. Of these reads 70-80% mapped unambiguously (phred scaled mapping quality >30) to the yeast genome, resulting in an average read coverage of >35-fold for the Ion 318™ Chip and >400-fold for the PI™ Chip, with 150 base reads. However, for embedded ribonucleotide mapping it is the read 5’ end coverage over the genome rather than sequenced read coverage that matters. For this measure, 0.2x to 0.4x genome average coverage for the Ion 318™ Chip and 3x to 6x coverage with the PI™ Chip was typical. Results for libraries prepared using other endonucleases are likely to be similar, as was the case for control libraries prepared using sequence specific endonucleases. Independent libraries yielded highly reproducible data, and demonstrated >99.9% site and strand specificity and >99% single nucleotide resolution (Fig. 2c and d).

It was previously unknown if embedded ribonucleotides were randomly distributed throughout the genome. The advent of a number of high-throughput sequencing methods, including emRiboSeq, has started to shed light on this. It is now clear that this distribution is non-random, with certain patterns emerging, including those related to polymerase-dependent nucleotide preferences, as well as sequence context^30,33–35. The way is now open to investigate genome-wide ribonucleotide incorporation, uncover which rules underlie the observed distribution and determine the (patho)physiological consequences.