A mapping platform for mitotic crossover by single-cell multi-omics

Abstract

Mitotic crossovers have the potential to cause large-scale genome rearrangements. Here, we describe high-throughput, single-cell, whole-genome sequencing methods for mapping crossovers genome-wide at scale. The methods are generalizable to various eukaryotes and to other end points requiring high-throughput, high-coverage single cell sequencing.

1. Introduction

DNA experiences various types of lesions that can lead to double-strand breaks (DSBs). DSBs can be repaired by error-prone end joining using limited sequence homology, or in an error-free manner using extensive homology by homologous recombination (HR) (Ceccaldi, Rondinelli, & D’Andrea, 2016). HR has two classes: local gene conversion and crossover (CO) (St Charles et al., 2012; Symington, Rothstein, & Lisby, 2014). The latter involves a linkage switch upstream and downstream of the break (Fig. 1A). Crossover can be a major source of large-scale genome rearrangement.

Fig. 1.

Mitotic crossover. A. Mitotic DSB repair. DSB can be repaired by end joining causing small indels, or by HR. While gene conversion involves local homology, crossover can cause extensive genome rearrangements. B—H. Types of mitotic crossover. Blue and black are homologs with SNPs, identical sister chromatids are of the same color, linked by centromere (oval). Homolog-specific copy number plots depict chromosomal alterations analyzed by WGS;for example, a chromosome with an inter-sister crossover still harbors one copy of the blue chromosome and one copy of the black chromosome. Abbreviations are defined in the text. B. SCE. Heterozygosity is maintained across the chromosome, but strand-switch can be detected by Strand-seq of the replication template strand (solid arrow). Without SCE, template strands continuously map to Watson (“W”) or Crick (“C”) strands (blue chromatids);with SCE, template strands have a switch (black chromatids). Reads mapped to the W and C strands are shown in orange and green, respectively. C. cnLOH. Inter-homolog crossover leads to reciprocal cnLOH in daughter cells (red bracket). D. CNV. Interstitial deletion between intra-chromatid repeats (orange boxes) as an example. E. inversion. Intra-chromatid repeats-mediated inversion detected by Strand-seq. F. translocation. Left: unbalanced translocation, recombined chromatids segregate apart, causing terminal CNV. Right: balanced translocation, recombined chromatids co-segregate, causing no change in copy number or heterozygosity, but change in Hi-C signal as now black chromosome is interacting with green chromosome. G. aneuploidy. Unresolved crossover intermediate is one cause of nondisjunction: monosomy and trisomy in reciprocal daughter cells. H. UPD. Reciprocal UPD shown as example. UPD can also be formed by duplication of monosomy or chromosome loss of trisomy.

Mitotic COs can result in seven types of chromosome rearrangements depending on the HR partner (Fig. 1B-H). Classes1–2 occur between allelic sequences. The vast majority occur between sister chromatids generated by replication. Sister chromatid exchange (SCE, class1) is error-free. The minority, 1% in mammalian cell lines and 3% in yeast (Claussin et al., 2017; Johnson, Liu, & Jasin, 1999; Kadyk & Hartwell, 1992; van Wietmarschen et al., 2018), occur between homologs bearing SNPs, causing copy-neutral loss-of-heterozygosity (cnLOH, class2). Classes3–5 use non-allelic repeats. Non-allelic HR (NAHR, (Sudmant et al., 2015)) can cause copy number variations (CNV, class3) and inversions (class4) with intra-chromatid repeats, or translocations (class5) with inter-chromosomal repeats. Lastly, unresolved HR intermediates and other errors could cause aneuploidy (class6) or uniparental disomy (UPD, class7) (Andersen & Petes, 2012). Unlike mutations, these CO outcomes are difficult to map without bias by standard bulk whole-genome sequencing (WGS). For example, the 99% error-free SCE is almost entirely missed in tumor signature studies that detect “scars” in the genome (Alexandrov et al., 2020).

Genome-wide mapping of DNA rearrangements has the advantage of naturally following to the end product and complements other biochemical, cell biology and molecular biology approaches that study snapshots of HR intermediates (Canela et al., 2016; Crosetto et al., 2013; Khil, Smagulova, Brick, Camerini-Otero, & Petukhova, 2012; Mimitou, Yamada, & Keeney, 2017; Sriramachandran et al., 2020; Wu et al., 2021). Such studies are particularly lacking for mitotic HR. In S. cerevisiae, genome instability screens are typically done in haploid with phenotypic markers (Putnam, Pennaneach, & Kolodner, 2005; Putnam et al., 2016); yet the assay is limited in the possible types of events that can be studied. Losing a marker in a diploid can be a result of mutation, cnLOH, deletion, monosomy, or UPD. Only mutations and deletions are possible in haploids subject to bias of survival. In addition, marker-based screens are confined to specific regions of the genome. A few studies focused on genome-wide mapping of mitotic HR in hybrid diploid yeast by analyzing cnLOH (Loeillet et al., 2020; Sampaio et al., 2020; St Charles et al., 2012; Sui et al., 2020; Yin, Dominska, Yim, & Petes, 2017; Yin & Petes, 2013). These global maps revealed the complexity of HR beyond simple explanations by current biochemical models, and uncovered variations not available in studies of induced DSB at defined loci, both regarding distinct types of rearrangements in various mutants, and non-random distributions of the events.

One can imagine extending such genome-wide mapping to various mutants and to human cells. However, such bulk sequencing assays are limiting because: (1) SNP-based analyses can only map inter-homolog HR; yet mitotic HR strongly prefers identical sister chromatids, particularly in mammalian cells; (2) the mapping requires cloning daughter cells experimentally, which is impossible for cells during development in vivo and is hard to scale up. Limited in throughput, roughly only 20 yeast mutants have been analyzed (Andersen, Sloan, Petes, & Jinks-Robertson, 2015; Andersen et al., 2016; Jinks-Robertson & Petes, 2021; Loeillet et al., 2020; Song, Dominska, Greenwell, & Petes, 2014; Yin & Petes, 2014, 2015; Zheng, Zhang, Wu, Mieczkowski, & Petes, 2016).

In Fig. 2A, we detail sequencing technologies that best map the seven types of mitotic CO detailed in Fig. 1. Global mapping of class1 (SCE) requires single-cell Strand-seq (Falconer et al., 2012), which monitors the strand switch of the replication template strand. Classes2–7 alter genetic information and are in theory detectable by bulk WGS; yet most can substantially benefit from single-cell assays, as they are not only copy-neutral but often reciprocal (like cnLOH and UPD-(Andersen & Petes, 2012; St Charles et al., 2012)). Single-cell WGS of daughter cells bypasses cloning and thus is highly scalable (Yin et al., 2019). Strand-seq uniquely detects error-free and mutational events, and provides a unifying tool for mapping all the seven types of CO outcomes without bias (Porubsky et al., 2021, 2020; Sanders et al., 2016).

Fig. 2.

sci-L3 technologies enable high-throughput, global mapping of mitotic crossover. A. Sequencing assays required for detecting 7 types of mitotic crossover. Strand-seq and Hi-C uniquely detect error-free SCE and balanced translocations, respectively. B. Overview of sci-L3: single-cell combinatorial indexing (“sci”), linear amplification (“L”) and 3-level (“3”). Cells are distributed into separate pools and barcoded for the 1st (colored edge), the 2nd (colored fill) and the 3rd round (different shapes). In-between each round, cells are pooled and sorted into new separate pools. A unique combination of three rounds of barcodes defines a single cell. We use IVT-based linear amplification to reduce exponential bias, achieve uniform coverage and prevent multiplicative errors. Sci-L3 enables single-cell WGS, target-seq or DNA and RNA co-assay. C. Expansion of sci-L3 toolset. By adding BrdU labeling or proximity ligation, sci-L3-WGS can be easily adapted to Strand-seq or Hi-C, respectively. By omitting nucleosome depletion, the sci-L3-DNA/RNA co-assay can be converted to an ATAC/RNA co-assay.

Various methods of single-cell genome sequencing methods have been developed and applied to characterize genome instability events genome-wide. Technology development in single-cell WGS largely focuses on two aspects: minimizing amplification bias (Chen et al., 2017) and increasing throughput (Cusanovich et al., 2015; Vitak et al., 2017). Here, we describe single-cell DNA sequencing methods developed by our laboratory that combine improvements in both uniformity in genome amplification and throughput as scalable global assays that unbiasedly detect various types of CO outcomes (Yin et al., 2019).

The suite is called sci-L3: single-cell combinatorial indexing (sci) with linear amplification (L) and 3-level barcoding (3) (Fig. 2B). In a nutshell, sci-L3 minimizes amplification bias with in vitro transcription (IVT)-based linear amplification, enables exponential gains in throughput with 1 M single cells per experiment at low cost by a 3-level split-and-pool barcoding strategy, and the scheme generalizes to multiple assays beyond WGS such as target-seq and DNA/RNA co-assay. We applied sci-L3-WGS to map 86,786 meiotic COs and chromosome mis-segregations in >10k sperm and discovered whole-genome equational segregation in meiosis I in both fertile and infertile mice (Yin et al., 2019). Sci-L3-WGS bypasses cloning daughter cells and can map at scale rare spontaneous mitotic CO, the mapping is SNP-based and thus misses the abundant error-free SCE. However, the sci-L3 scheme can be easily adapted to Strand-seq with small modification of the protocol (Fig. 2C).

2. Overview of sci-L3-WGS

Fixed nuclei first undergo nucleosome depletion. The nuclei are then evenly distributed into X wells and barcoded for the 1st round by Tn5 insertion. The Tn5 transposomes fragment the genome and simultaneously insert a unique barcode for each well. The process of tagging the DNA with well-specific barcodes on average every 0.5−1.5kb is termed “tagmentation.” We then pool and split the nuclei into Y wells and ligate the 2nd round of barcodes and a T7 promoter, enabling subsequent IVT-based linear amplification. All of the nuclei are once again pooled together with X times Y number of barcode combinations after two rounds of split-and-pool. We then FACS sort or dilute to distribute the nuclei into a final round of wells. The number of nuclei per well is much smaller than X times Y such that the nuclei in the same well do not share the same barcode 1 and barcode 2 combinations. Nuclei are then lysed and subjected to downstream IVT, reverse transcription (RT) and second-strand synthesis (SSS) to amplify the genome in a linear fashion. A third round of barcodes is added during SSS. The combination of three rounds of barcodes define the single-cell origin of each amplified and sequenced molecule.

3. Preparing for a sci-L3 experiment

1. Prepare the following stock solutions:

   1. QIAGEN Protease (60mg/mL; 2.8mL)
      Add 2778μL ddH₂O to 166.7mg of lyophilized protease powder to make 60mg/mL stock.
      Store at 4°C. Note that the amount of water to add may vary by lot.
   2. 1× TE (10mM Tris-HCl pH 8.0, 1mM EDTA; 1mL)
      1μL 1M Tris-HCl pH 8.0
      2μL 0.5M EDTA pH 8.0
      988μL ddH₂O
      Store at room temperature
   3. 1× Lysis buffer (60mM Tris-Ac pH 8.3, 2mM EDTA pH 8.0, 15mM DTT; 1mL)
      150μL 10× TAE pH 8.0
      1μL 0.5M EDTA pH 8.0
      15μL 1M DTT
      10μL 1× TE
      824μL ddH₂O
      Store at −20°C
   4. 10× thms buffer (50mM MgCl₂, 100mM Tris pH 8.0; 1mL)
      100μL 1M Tris-HCl pH 8.0
      50μL 1M MgCl₂
      850μL ddH₂O
      Store at room temperature
   5. Tn5 Storage buffer - STB (50% glycerol, 0.5 × TE; 400μL)
      200μL 1× TE pH 8.0
      200μL glycerol
      Store at room temperature
   6. Glycine (2.5M glycine; 1mL)
      Weigh out 187.7mg of glycine powder and make up to 1mL with ddH₂O
      Store at room temperature
   7. 1× LBT (1× LB, 0.1% TritonX-100; 1mL)
      990μL 1× Lysis buffer
      10μL 10% TritonX-100
      Prepare fresh on the day of experiment
2. Anneal oligonucleotides:

   1. Prepare annealing buffer (10mM Tris-HCl pH 8.0, 50mM NaCl, 1mM EDTA, pH 8.0; 1mL)
      10μL 1M Tris-HCl pH 8.0
      10μL 5M NaCl
      2μL 0.5M EDTA pH 8.0
      978μL ddH₂O
      Store at room temperature
   2. Resuspend:
      lyophilized Tn5 mosaic end (ME) oligo and the Tn5 ME ligation oligo in annealing buffer to a 100μM concentration. Mix the two oligos together 1:1 (50μM).
      lyophilized ligation hairpin oligo in annealing buffer to a 50μM concentration
   3. Anneal with gradual cooling:
      95°C 5min, −0.1 °C/cycle, hold for 9s/cycle, 700 cycles to 25°C. This should take 2–3h.
   4. Dilute:
      Tn5 mosaic end double-stranded (MEDS) oligo to a final concentration of 1.5μM with annealing buffer and store in −20°C until ready to use
      Hairpin ligation double-stranded oligo to a final concentration of 1.5μM with annealing buffer and store in −20°C until ready to use

5. Materials and equipment

1. Reagents

   • QIAGEN Protease (QIAGEN, cat. no. 19155)
   • 1M Tris-HCl pH 8.0 (Invitrogen, cat. no. 15568025)
   • 0.5M EDTA pH 8.0 (Invitrogen, cat. no. AM9260G)
   • 10× TAE pH 8.0 (Invitrogen, cat. no. AM9869)
   • 1M DTT (Sigma-Aldrich, cat. no. 646563-10X.5ML)
   • TritonX-100 (Sigma-Aldrich, cat. no. 93443-100ML)
   • 1M MgCl₂ (Invitrogen, cat. no. AM9530G)
   • Glycerol (Sigma-Aldrich, cat. no. G5516-100ML)
   • DMEM complete media (Thermofisher Scientific, cat. no. 11054001)
   • 37% Formaldehyde (Sigma-Aldrich, cat. no. F15587)
   • Glycine (Sigma-Aldrich, cat. no. G7403-250G)
   • IGEPAL^® CA-630 (Sigma-Aldrich, cat. no. I8896)
   • 10× NEBuffer™ 2.1 (New England Biolabs, cat. no. B7202S)
   • SDS, 20% Solution (Invitrogen, cat. no. AM9820)
   • EZ-Tn5™ Transposase (Lucigen, cat. no. TNP92110)
   • 5M NaCl (Invitrogen, cat. no. AM9760G)
   • dNTP set 10mM each (Kapa Biosystems, cat. no. KK1017)
   • PEG-4000 (Sigma-Aldrich, cat. no. 25322-68-3)
   • T4 DNA ligase (Thermofisher Scientific, EL0011)
   • DAPI (Thermofisher Scientific, cat. no. D3571)
   • Ethanol, 200 proof, Molecular Biology Grade (FisherScientific, cat. no. 07-678-003)
   • Bst 2.0 WarmStart^® DNA Polymerase (New England Biolabs, cat. no. M0538)
   • HiScribe T7 Quick High Yield RNA Synthesis Kit (New England Biolabs, cat. no. E2050S)
   • SUPERase•In™ RNase Inhibitor (Thermofisher Scientific, cat. no. AM2696)
   • SuperScript™ IV Reverse Transcriptase (Thermofisher Scientific, cat. no. 18090200)
   • RNaseH (New England Biolabs, cat. no. M0297)
   • RNaseA (Invitrogen, cat. no. AM2270)
   • Q5^® High-Fidelity DNA Polymerase (New England Biolabs, cat. no. M0491)
   • NEBNext^® Ultra™ II DNA Library Prep Kit for Illumina^® (New England Biolabs, cat. no. E7645)
   • Agencourt AMPure XP Magnetic Beads (Beckman Coulter, cat. no. A63881)
   • Qubit ™ dsDNA HS Assay Kit (Thermofisher Scientific, cat. no. Q32851)
2. Equipment

   • Eppendorf LoBind^® twin. tec^® PCR Plates, 96 well (Eppendorf, cat. no. 10049-108)
   • Magnetic racks for 1.5mL tubes (DynaMag-2; Thermofisher Scientific, cat no. 12321D)
   • 1.5mL DNA/RNA low retention microcentrifuge tubes (Eppendorf, cat. no. 022431021)
   • 1.7mL microcentrifuge tubes (Genesee Scientific, cat. no. 22-282)
   • Axygen^® 0.2mL MAXYMum Recovery™ Thin Wall PCR Tubes with Flat Cap (Corning, cat. no. PCR-02-L-C)
   • RNA Clean & Concentrator™5 kit (Zymo Research, cat. no. R1016)
   • DNA Clean & Concentrator™5 kit (Zymo Research, cat. no. D4014)
   • Benchtop microcentrifuge (Eppendorf, cat. no. 5420)
   • Refrigerated centrifuge (Eppendorf, cat. no. 5424R)
   • Eppendorf ThermoMixer^® C (Eppendorf, cat. no. 5382000023)
   • Eppendorf SmartBlock™ 1.5mL (Eppendorf, cat. no. 5360000038)
   • Eppendorf SmartBlock™ PCR 96 (Eppendorf, cat. no. 5306000006)
   • Hemacytometer
   • Qubit Fluorometer 4 (Thermofisher Scientific, cat. no. Q33240)
   • Single channel Pipettes (Rainin, P2, P20, P200, P1000)
   • 8-well and 12-well Multi-channel pipettes (Rainin, P20, P200)
   • Low retention tips (Rainin)
   • Reagent Reservoirs (FisherScientific, cat. no. 21381093)

6. Step-by-step method details

6.1. Crosslinking and nucleosome depletion

Timing: 2h 30min

1. Crosslink cells in 10mL DMEM Complete medium with 406μL 37% formaldehyde (final concentration 1.5%) at room temperature (21–25 °C) for 10min while gently inverting the tube

2. Add 800μL 2.5M glycine and incubate on ice for 5min. After, pellet cells at 500g for 5min at 4°C and wash cells in 1mL 1× LBT

Critical: Ensure that the 2.5M glycine solution used is no older than a month.

• 3.Resuspend pellet in 1mL 1× LBT with 10μL 10% IGEPAL CA-630 (0.1% final) and incubate on ice for 20min. After, pellet nuclei at 500g for 5min at 4°C and wash once with 1× NEBuffer2.1 (diluted to 1× with ddH₂O).
• 4.Resuspend pellet in 776μL 1× NEBuffer 2.1+24μL 10% SDS and place the tubes in a thermomixer set to 42°C with vigorous shaking (500rpm) for 30min
• 5.Add 180μL 10% Triton-X and place the tubes in a thermomixer set to 42°C with vigorous shaking (500rpm) for 30min

Note: During the incubation, to save time, assemble the transposome as described in step 7.

• 6.Spin at 500g for 5min at 4°C and wash once in 1mL LBT. After the last wash, leave about 20μL behind, gently resuspend the pellet and count nuclei using a hemocytometer. Make up volume with 1× LBT to about 20,000 cell/μL

6.2. First level indexing with tagmentation

Timing: 1h

• 7.
  Assemble transposome (for 24 wells):

  1. Dilute EZ-Tn5 with STB:
     14μL EZ-Tn5 (~1μM)
     7.5μL STB
  2. Combine 0.8μL diluted EZ-Tn5 with 0.6μL annealed Tn5 MEDS (1.5μM) in a 96-well LoBind plate and incubate at room temperature for 30min.

Critical: Here, we prepare volumes for at least 2 additional wells to account for pipetting errors.

Note: Here, we prepare 24 reactions for 24 first round barcodes, with each barcode located within an individual well. Depending on application, it may be desirable to use more first round barcodes.

Note: During assembly, the transposome dimerizes to a final concentration of ~0.25μM. The assembled transposome complex can be stably stored at −20°C for up to a year.

• 8.Preheat the thermomixer (or thermocycler) to 55°C
• 9.Prepare the MgCl₂ solution:

  6.5μL H₂O ×26: 169μL H₂O

  0.7μL 50mM MgCl₂ ×26: 18.2μL MgCl₂

• 10.Into each well of a new 96-well LoBind plate, add 7.2μL of the MgCl₂ solution (final MgCl₂ concentration of 3.22mM, accounting for EDTA in the lysis buffer and STB), along with 1.5μL nuclei in LBT (approximately 30,000) and 1.19μL of the assembled transposome. Incubate at 55°C for 20min in a thermomixer (or can also use a thermocycler).
• 11.After tagmentation, set the thermomixer at 20°C. Leave the lid open to cool (takes about 30min to cool from 55°C to 20°C).

6.3. Second level indexing with ligation

Timing: 1h

• 12.Prepare RBS solution (for 64 wells):

  24μL 10mM dNTP, 48μL 10× thms buffer, 96μL ddH2O, 144μL LBT.

Note: Here, we prepare 64 reactions for 64 s round barcodes, with each barcode located within an individual well. Depending on application, it may be desirable to use more second round barcodes.

• 13.Pool cells from all well into a trough containing 500μL of LBT and transfer to a 1.5mL conical tube. Wash wells and trough with an additional 500μL LBT and transfer into the same conical tube
• 14.Spin at 500g at 4°C for 8min and resuspend the nuclei pellet in 312μL RBS. Add 40μL into each well of a new 8-well strip
• 15.Prepare ligation mix (for 64 wells):

  48μL 10× T4 ligase buffer, 38.4μL PEG-4000, 8μL T4 DNA ligase

Note: For 64 wells, we prepare enough ligation mix for 80 wells to account for pipetting errors and loss between transfers. A single well contains 0.6μL 10× T4 ligase buffer, 0.48μL PEG-4000, 0.1μL T4 DNA ligase.

• 16.Add 11μL of ligation mix into each well of a new 8-well strip
• 17.Prepare a new 96-well LoBind plate with 0.8μL (1.5μM) annealed double-stranded ligation hairpin oligos
• 18.Using a multichannel pipette, add 4.9μL of nuclei in RBS and 1.2μL of the ligation mix into the first 8 columns (64 wells) of the 96-well LoBind plate containing the double-stranded ligation hairpin oligos. Mix by pipetting up and down
• 19.Incubate in a thermomixer for 30min at 20°C
• 20.During ligation, prepare a 96-well LoBind plates for FACS by adding 3μL of 1× LBT into each well
• 21.After ligation, combine all well in a trough (approximately 630μL). Add 70μL 50μg/mL DAPI, transfer to an appropriate FACS tube and keep on ice until ready to sort

6.4. FACS

Timing: 1–2h

• 22.Sort 100–300 nuclei per well straight into 1× LBT using DAPI, gating for 2C cells and excluding doublets

Note: Each sorting event is a droplet of approximately 3–5nL of FACS buffer, depending on the size of the nozzle. To keep the salt concentration low, we recommend keeping the total volume sorted into each well to be less than 1μL.

• 23.After sorting, spin down plate at 500g for 5min at 4°C.

Pause Point: sorted nuclei can be frozen at −80°C at this point for up to a year.

6.5. Nuclei lysis

Timing: 5–9h

• 24.
  Lyse sorted nuclei

  1. By incubating the plate at 75 °C for 30min, followed by cooling to 4°C
  2. Into each well add 1μL of freshly diluted QIAGEN protease (8mg/mL diluted with ddH₂O, final concentration 2mg/mL)
  3. Incubate for 4–8h at 55°C, followed by 30min at 75°C to heat-inactivate the protease. For overnight incubations, the plate can stay at 4°C

Pause Point: lysed nuclei can be frozen at −80°C at this point for up to a year.

6.6. Gap extension and linear amplification by in vitro transcription

Timing: 11–17h

Note: We recommend processing no more than 32 wells of samples (approximately 9600 cells) at a time as the subsequent amplification step involves RNA and is subject to sample degradation.

• 25.Prepare gap extension mix (for 24 samples):

  50μL H₂O, 17.5μL 10× thms buffer, 8.76μL of dNTP, 8.74μL of Bst WarmStart 2.0 polymerase

Note: For 24 samples, we prepare enough gap extension mix for 25 samples to account for pipetting errors. A single well contains 2μL H₂O, 0.7μL 10× thms buffer, 0.35μL 10mM dNTPs, 0.35μL Bst WarmStart 2.0 polymerase.

• 26.Add 3.4μL into three 8-well strips and incubate for 5min at 68°C
• 27.Prepare the HiScribe T7 in vitro transcription mix (for 24 samples):

  48μL H₂O, 48μL T7 RNA Polymerase mix, 240μL 10mM rNTP mix

Note: A single well contains 2μL H₂O, 2μL T7 RNA Polymerase mix and 10μL 10mM rNTP mix.

• 28.Add 13.9μL of the T7 mix into each tube for a final IVT reaction volume of 20μL
• 29.Incubate at 37 °C for 10–16h

Critical: RT and SSS steps need to be performed right after IVT and take approximately 4h.

Note: Before proceeding to the RT step, set up 24×1.5mL LoBind tubes, 24× Zymo spin IC columns, 24×1.7mL microcentrifuge tubes and 24× low retention PCR tubes.

6.7. Reverse transcription of linearly amplified samples

Timing: 2h

• 30.Add 2.22μL 0.5M EDTA into each tube to terminate IVT
• 31.
  Purify the IVT RNA using the RNA Clean & Concentrator-5 kit. To each sample:

  1. Add 52μL RNA Binding Buffer and mix well by pipetting up and down
  2. Add 95μL of 100% ethanol, mix well by pipetting up and down and transfer to a Zymo-Spin IC column and spin at 13,000 g for 30s. Discard the flow-through
  3. Add 400μL RNA Prep Buffer and spin at 13,000 g for 30s. Discard the flow-through
  4. Add 700μL RNA Wash Buffer and spin at 13,000 g for 30s. Discard the flow-through
  5. Add 500μL RNA Wash Buffer and spin at 13,000 g for 2min to ensure complete removal of ethanol from the column silica matrix
  6. Transfer each column to a new 1.5mL LoBind tube and add 18μL 0.1× TE (pH 8.0) directly onto the column matrix. Wait 2min to allow RNA to re-dissolve, after which spin at 13,000 g for 1min
• 32.
  Prepare the RT mix (for 24 samples):

  1. RT mix A: 48μL 10mM dNTP, 14.4μL RT RNA primers (50μM), 12μL Superase-In
     Note: A single tube will contain 2μL 10mM dNTP, 0.6μL RT RNA primers (50μM), 0.5μL Superase-In.
  2. RT mix B: 144μL 5× SuperScript IV buffer, 36μL 0.1M DTT, 24μL SUPERase-In, 24μL SuperScript IV
     Note: A single tube will contain 6μL 5× SSIV buffer, 1.5μL 0.1M DTT, 1μL SUPERase-In, 1μL SuperScript IV.
• 33.Distribute 3.1μL of RT mix A into low retention PCR tubes
• 34.Add approximately 18μL of purified IVT RNA and mix by pipetting up and down
• 35.Incubate for 1min at 70°C, followed by 20s at 90 °C in a thermocycler to denature and remove any RNA secondary structure. After denaturation, place samples into a 4°C water bath for fast cooling. If processing more than 8 samples, start cooling after 10 s at 90°C. After approximately 30s in the 4°C water bath, transfer samples to a 96-well ice block
• 36.Add 9.4μL of RT mix B into each tube, mix well by pipetting up and down followed by a quick pulse spin
• 37.Incubate in a thermocycler with the following RT program:

  25°C 1 min

  37°C 1 min

  55°C 15min

  60°C 10min

  65°C 12min

  70°C 8min

  75°C 5min

  80°C 10min

  25°C hold

6.8. RNase treatment

Timing: 40min

• 38.Prepare a mix of 12.5μL RNaseH (0.5μL per sample) and 7.5μL RNaseA (0.3μL per sample).
• 39.Add 0.8μL of the RNase mix to each sample
• 40.Incubate at 37 °C for 30min to eliminate the RNA
• 41.Thaw 24 SSS primers during incubation

6.9. Third level indexing with second-strand synthesis (SSS)

Timing: 1h

• 42.Add 1μL of a unique SSS barcoding primer into each tube
• 43.Prepare the Q5 master mix for SSS (24 samples):

  661.5μL H₂O, 490μL 5× Q5 buffer, 490μL Q5 GC enhancer, 24.5μL Q5 polymerase.

Note: The Q5 master mix for a single sample contains 27μL H₂O, 20μL 5× Q5 buffer, 20μL Q5 GC enhancer, 1μL Q5 polymerase.

• 44.Add 67μL of the Q5 master mix to each RT sample (100μL reaction volume).
• 45.Incubate in a thermocycler with the following SSS program:

  98°C 40s

  58°C 30s

  60°C 30s

  65°C 30s

  70°C 30s

  72°C 6min

  4°C hold

• 46.
  Purify the resulting double stranded DNA using the DNA Clean & Concentrator-5 kit:

  1. To each SSS reaction, add 500μL DNA binding buffer and mix well by pipetting up and down. Load onto a Zymo-Spin IC Column and spin at 13,000 g for 30s. Discard the flow-through
  2. Wash twice with 200μL DNA wash buffer and spin at 13,000 g for 30s. Discard the flow-through
  3. Spin at 13,000 g for 2min
  4. Transfer each column to a new 1.5mL LoBind tube and add 10μL 0.1× TE. Wait 2min to allow DNA to re-dissolve, after which spin at 13,000 g for 1min
• 47.Qubit 0.8μL of the sample using the dsDNA high sensitivity kit

6.10. Library preparation

Timing: 2h 30min

• 48.Bring up dsDNA (20–100ng) after SSS to 50μL with 0.1× TE
• 49.Add 7μL End-repair buffer and 3μL End-repair enzyme (from the NEBNext Ultra II kit, please see manufacturer’s protocol for additional details) and mix well by pipetting up and down
• 50.Incubate for 30min at 20 °C and then 30min at 65°C
• 51.
  Dilute sequencing adaptors with 10mM Tris-HCl, 10mM NaCl depending on sample concentration:

  1. >200ng, no dilution (15μM working adaptor concentration)
  2. 50–200ng, 1:10 dilution (1.5μM working adaptor concentration)
  3. <50ng, 1:20 dilution (0.6μM working adaptor concentration)
• 52.Add 2.5μL diluted adaptor, with 30μL ligation master mix and 1μL ligation enhancer and mix well by pipetting up and down
• 53.Incubate for 15min at 20 °C
• 54.Add 3μL of USER enzyme and mix well by pipetting up and down
• 55.Incubate for 15min at 37 °C
• 56.
  Perform an AMPure XP magnetic bead cleanup:

  1. Prepare 80% ethanol by diluting with ddH₂O

Note: Always prepare fresh 80% ethanol just before performing cleanup.

• b)Add 140μL (1.4×) AMPure XP beads and mix by pipetting up and down at least 10 times
• c)Incubate at room temperature for 5min, after which place on a magnetic stand for another 3min

Note: During incubations, prepare the PCR reaction and pre-heat the thermocycler.

• d)Remove supernatant and wash with 200μL 80% ethanol twice, briefly incubating 30–60s each time. Do not remove tubes from the magnet during the washes
• e)After a second wash, remove any residual liquid and air dry the beads until they turn a matte brown color

Critical: Do not over dry beads.

• f)Elute with 17μL 0.1× TE, take off the magnet and mix well by pipetting up and down at least 10 times. Incubate for 2min
• g)Place back onto the magnetic rack and take 16μL of the supernatant into the PCR reaction

• 57.
  For the PCR reaction:

  1. Combine in an 8-well strip, 2μL Universal primer, 2μL Indexed primer, 20μL 2× NEBNext Ultra II Q5 Master Mix, 16μL purified DNA (40μL reaction).
  2. PCR cycling conditions:
     Initial denaturation 98°C 30s.
     4 cycles of:
     • Denaturation 98°C 10s.
     • Annealing/Extension 65°C 75s.
     Final extension 65°C 5min
     Hold 4°C
• 58.After PCR add 34.6μL (0.86×) AMPure XP magnetic beads and repeat step 56, eluting with 10–30μL 0.1× TE
• 59.Qubit 0.8μL of the sample using the dsDNA high sensitivity kit

7. Expected outcomes

With single human or mouse cells, we expect the dsDNA concentration to be at about 50 ng/μL with Qubit quantification at step 47 with 100–300 sorted per well. The concentration and size distribution of the final library depend on how much was used as input for library preparation. We recommend quantifying the molar concentration of the sequencing library using only fragments between 200 and 400bp with BioAnalyzer or TapeStation.

8. Quantification and statistical analysis

To process raw fastq files into single cell BAM files for bespoke downstream analysis, we have created a Snakemake workflow available at https://github.com/recombinationlab/sciL3Pipe. The major step of the workflow include:

1. Because of the bi-directional T7 transcription and the orientation independent ligation of sequencing adaptors, the final library contains reads with barcodes on either read 1 (R1) or reads 2 (R2). The first step involves identifying the read that contains the RT primer sequence, swapping orientation if sequence present in R2. Subsequently, the SSS barcode (third level index) is extracted and used to generate a separate fastq file for each barcode. The splitting by barcode enables parallel execution of all the downstream step

2. The next step involves identifying the Tn5 and ligation barcodes (first and second level indexes) and trimming off all the barcodes using the Tn5 MEDS sequence. Add barcodes are retained within the read names

3. Trimmed reads are aligned using BWA-MEM

4. The output BAM file is finally split into individual cell BAM files using the barcodes saved within the read names

The automated workflow also generates several QC metrics, including genome coverage and collisions if a mixture of cells from two species was used within the experiment.

To run the workflow:

1. Clone the sciL3Pipe repository from github:

   git clone https://github.com/recombinationlab/sciL3Pipe

2. Make sure the required dependencies are installed:

   1. Conda
   2. Python 3
   3. Snakemake

3. The workflow requires a JSON file containing locations and names of the input fastq files. This can be generated using fastq2json.py using the following command:

   python fastq2json.py -fastq_dir /path/to/fastq/directory

4. Users are expected to generate or download their own BWA-MEM genome index. The path to the local index needs to be updated in the configs/config.yaml file

5. Finally, the workflow can be executed using the provided shell run scripts:

   sh run_pipeline_sge.sh (to run on an SGE cluster) or sh run_pipeline_local.

   sh (to run locally).

4.

Key resources table

Reagent or resource	Source	Identifier
Experimental Models:	Cell Lines
HEK293T	ATCC	CRL-3216
3T3	ATCC	CRL-1658
Oligonucleotides
Barcoding oligos for sci-L3	(Yin et al., 2019)	Table S5
Software and Algorithms
sciL3Pipe	This Paper	https://github.com/recombinationlab/sciL3Pipe