Smart-seq3xpress

• This protocol requires a some type of liquid handler capable of doing nanoliter dispenses. We have tested and used ( Formulatrix Mantis, Dispendix I.Dot & Dispendix I.Dot Mini ). Other (non contact) liquid dispensers should work as well, as long as they can dispense the required volumes accurately.

• All volumes have been scaled to nanoliter volumes, as such the volume your cell is dispensed in matters. We have tested this protocol with an array of FACS machines and cell printers including BD FACSMelody, BD Fusion, BD Influx, Sony SH800S, Cellenion CellenOne, Cytena F.SIGHT Omics, which all typically dispenses the cell in ~5-10nl or less. If your instrument dispenses in higher volumes (> 50nl) , the protocol may either not work or not be as efficient.

• Consider what buffer you use to dispense / sort your single cells in. Since the relative difference between sorted cell volume and lysis volume has overall decreased, common additives like FBS, BSA, EDTA can potentially interfere and affect downstream molecular reaction if present in high enough amounts. As such we recommend if possible to sort in PBS alone, or as recommended by 10x Genomics a solution of PBS + 0.04% BSA at most. Refrain also from using buffers with Mg2+ and Ca2+ or other metal ions for sorting. If EDTA is an absolute must, try and keep the amounts low. Avoid other additives like DNAseI, and Sodium Azide.

• Not all RNase Inhibitors are compatible and some can have severe negative impact on the downstream reaction and ultimately library quality. It is highly recommened to use the RNase Inhibitor used in the protocol () . Alternative RNase Inhibitors, that have been tested and seemed compatible are,

,

• Usually yes it is possible, but user optimization might be required, and results may vary. It is no longer Smart-seq3xpress, and our support will therefore be limited.

This protocol should be carried out in a clean environment. Use ethanol, RNAseZAP, DNA-OFF, or similar to prepare work bench before start.

Work quickly and preferably on ice.

Try and prepare master-mixes right before use, while the plate(s) are finishing the previous incubation step.

This step is optional and can be disregarded if you choose to use or similar beads for cleaning up your final libraries.

These 22% PEG beads for clean-up is prepared similar to the mcSCRB-seq protocol (mcSCRB-seq)

Weight out the PEG8000 in a 50mL Falcon tube, and add all ingredients except IGEPAL CA-630 together. DO NOT add the total amount of water but wait until the PEG is completely dissolved until filling the Falcon tube up to 50mL.

Incubate at 37-40°C and vortex regularly to help dissolve the PEG8000.

Meanwhile the PEG8000 solution is dissolving prepare the Sera-Mag Speed Beads.

Resuspend the bead stock, and pipette 1000µL of bead stock into a 1.5mL tube.

Place on magnet stand and let beads collect. Remove supernatant

Add 1000µL of a 10mM Tris-HCl pH 8, 1mM EDTA (TE) solution and resuspend beads before retuning the tube on the magnet stand.

Remove supernatant and repeat the wash above once more.

Afterwards remove supernatant and add and re suspend the beads in 900µL 10mM Tris-HCl pH 8, 1mM EDTA (TE).

Add the bead solution to the PEG8000 solution above, after it is dissolved.

Add IGEPAL-CA630, and fill up with the remaining H2O to 50mL. Mix it well.

We have tested multiple types of hydrophobic inert overlays such as Silicone Oils, Hydrocarbons, and the commercial Vapor-Lock (Qiagen). They can have different properties, such as viscosities and solidifying temperature. We commenly use Vapor-Lock, Sillicone oil 25 cSt, Silicone Oil 100 cSt (The higher viscosity is better suited for shipping plates).

CAUTION!!! Do not dispense these silicone oils / overlays with your non contact liquid handler. The solutions can "creep" everywhere. Use either manual multichannel pipettes or semi-manual (e.g. Integra ViaFlow) / automatic dispensing (e.g. Agilent Bravo, Tecan Fluent) with tips, and prepare and store in bulk.

Add 3µL of overlay to each well of a 384 well plate. The amount of overlay can be increased if desired.

Quick pulse centrifugation to 1000x g,0h 0m 0s to ensure all is collected in the bottom.

Put on seal and store at Room temperature until use.

Prepare lysis buffer reaction mix. The mix is prepared for 500 samples. Adjust this to suit your liquid handlers dead volume and pipetting comfort level.

Reaction concentrations for PEG8000, OligodT30VN and dNTPs, are adjusted to and reflect their final concentration in the reverse transcription reaction (0.4uL).

• Ensure that PEG is fully mixed into solution, by either pipetting up and down until the liquid is clear, or start with vortexing the required master-mix volume of water and PEG together before adding the remaining reagents.

Add 0.3µL of lysis mix to each well of a 384 well plate containing overlay.

Quick pulse centrifugation to 1000x g,0h 0m 0s to ensure lysis mix is collected in the bottom underneath the overlay before storage until use. For short term storage On ice . For longer term storage, store the lysis plates in a freezer at -20°C or at -80°C if the lysis plates contain Spike-in RNAs.

Sort single cells into each well containing 0.3µL lysis reaction mix overlayed by 3µL of preferred overlay.

As mentioned in the Considerations section . Be mindful of the solution cells are sorted in, as additives such as BSA, FBS and EDTA, can have negative impact on the downstream molecular reactions.

Seal plates with a cold storage foil seal, do a quick pulse centrifugation to 1000x g,0h 0m 0s and immediately store at -80°C or on dry ice.

Remove the plate of sorted cells from the -80 freezer and incubate in a thermocycler with heated lid at 72°C for 0h 10m 0s followed by a 4°C hold. Keep the -80C storage seal on for this step, even if it seems loose (read above).

While the plate is incubating in step 7, prepare RT reaction mix. The mix is prepared for 500 samples. Adjust this to suit your liquid handlers dead volume and pipetting comfort level.

Add 0.1µL of RT reaction mix into each well, seal the plate with a PCR seal and pulse centrifuge the plate quickly to 1000x g,0h 0m 0s to ensure RT mix merges properly with the lysed cell mix.

Incubate the plate in a thermocycler as following.

While the plate is incubating in step 8, prepare PCR reaction mix. The mix is prepared for 500 samples. Adjust this to suit your liquid handlers dead volume and pipetting comfort level.

Add 0.6µL of PCR reaction mix to each well seal the plate with PCR seal and pulse centrifuge the plate quickly to 1000x g,0h 0m 0s to ensure RT mix merges properly with the lysed cell mix.

Run the following PCR program in a thermocycler.

*To determine the number of PCR cycles it mainly depends on your cells and RNA content. We use as a standard 12 cycles for HEK cells, and 16 cycles for PBMCs.

However, for most cell types a wide range of PCR cycles can be used, yielding similar quality data. Below is a schematic of selected sequencing output metrics from a PCR cycle evaluation run on PBMCs. Each 384 well plate of sorted PBMCs from the same donor, same batch, underwent a different number of PCR cyclers in preampfification. (Take into consideration heterogeneity of PBMCs as each column is a unique 384 well plate of cells)

The primary concerns when choosing PCR cycles is

1. To amplify cDNA enough to be well above the genomic DNA levels.

2. Using more cycles means also needing to use more TDE1 Tn5 enzyme per cell to be able to get a good ratio of UMI containing reads to internal reads (~50%). This is not such a massive concern since we use minuscule amounts of TDE1 per cell, If you need to adjust the amount of TDE1 Tn5, we recommend adjustments in the order of +/- 0.0005uL per cell increments.

3. Choosing lower PCR cycles might mean you will need to scale up the amounts of PCR cycles for the tagmentation PCR (see below

Above image in higher quality format.

PBMC_PCRcycles_statsv2.pdf

After the preamplification PCR dilute the 1µL cDNA by adding 9µL of

Again, seal the plate and pulse centrifuge to 1000x g,0h 0m 0s to ensure that everything is collected beneath the overlay.

1. Transfer 1µL of diluted cDNA to a new 384 well plate, and put On ice for until tagmentation.

2. Prepare 4x Tagmentation buffer as following. Aliquots of 4xTD buffer can be stored at -20°C for later use.

3. Prepare Tagmentation Mix. Mix is for 500 samples.

NB !!! TDE1 Tn5 enzyme comes as a glycerol viscous solution. When pipetting these small uL amounts to your tagmentation mix, be careful to not transfer extra drops that are stuck to the pipette tip. This can cause overtagmentation, which is not bad, but can cause less than expected UMI reads to be captured in sequencing.

*This is a suggestive amount of Tn5, but overall a good starting point that works well with HEK cells (12xPCR preamplification) and PBMCs (16xPCR preamplification). The amount can be changed to meet user specific criteria in terms of ratio of UMI containing reads vs Internal reads, and can be affected by cell-type or amount of preamplification PCR cycles given.

4. Dispense 1µL of Tagmentation Mix to each well of the 384 well plate containing 1µL of diluted and pre-dispensed cDNA. Seal and pulse centrifuge the plate down at 1000x g,0h 0m 0s .

5. Incubate the plate at 55°C for 0h 10m 0s in a thermocycler.

6. To stop the tagmentation reaction and strip off the Tn5 enzyme, add 0.5µL of 0.2% SDS to each well. Seal the plate and quickly centrifuge the plate down before incubation at Room temperature for 0h 5m 0s

7. Proceed to either substep 11.1 or substep 11.2 depending on your index primer concentration and whether you want to PCR your tagmented libraries in higher or lower volumes. For adding Nextera index primers we recommend preparing index plates with already combined Nextera i5 and i7 indexes at a certain working dilution.

Lower volume indexing PCR, final volume 5uL per well. This can save a bit of cost on Phusion Polymerase

8.1 Add 1µL of premixed index primers.

9.1 Prepare Index PCR mix. Again the calculated amount here for ease is for 500 sample. Please scale this to suit your specific dead volume.

*Addition of Tween-20 to the PCR reaction is necessary to avoid having the remaining SDS affect Phusion DNA polymerase, at the given concentration and amount used in the stop reaction. Tested working amounts of Tween-20 used with Phusion polymerase is 0.005-0.05%. Above 0.05% the reaction starts to get negatively affected by it in our hands. If you change these parameters try and match or be in slight excess with the amount of Tween-20 to the concentration of leftover SDS from the stop solution.

10.1 Add 1.5µL of index PCR mix to each well, seal the plates, apply quick centrifugation to settle everything in the bottom, and incubate in a thermocycler as following:

*The amount of PCR cycles needed depends on the starting material, preamplifation cycles given, and amount of tagmentation performed (amount TDE1 Tn5 used per cell). Therefore some user specific optimization may be necessary to suit the needs and wants and final material required.

However as a point of reference:

• HEK cells (12xPCR cycles in preamplification PCR) we use a standard 12xPCR cycles in tagmentation PCR

• PBMCs (10-18xPCR cycles in preamplification PCR) we use 14xPCR cycles in tagmentation PCR.

Standard volume index PCR, final volume 8uL. If you have index plates in the working concentration used for Smartseq3. (0.5micromolar (µM) )

8.2 Add 3.5µL of premixed index primers to each well.

9.2 Prepare index PCR mix. Again the calculated amount here for ease is for 500 sample. Please scale this to suit your specific dead volume.

*Adding a bit of Tween-20 to the PCR reaction can help protect the DNA polymerase a bit against SDS from previous stop reaction. This trick can also help other polymerases to work better like KAPA, Vent etc.

10.2 Add 2µL of index PCR mix to each well, seal the plates, apply quick centrifugation to settle everything in the bottom, and incubate in a thermocycler as following:

*The amount of PCR cycles needed depends on the starting material, preamplification cycles given, and amount of tagmentation performed (TDE1 Tn5 amount added per cell). As such it might require a bit of user optimization, depending on wants, needs, and final material needed.

However as a reference point:

• HEK cells (12xPCR cycles in preamplification PCR) we use as standard 12xPCR cycles in tagmentation PCR. (10xPCR is confirmed working as well)

• PBMCs (10x-18xPCR cycles preamplification PCR) we use 14xPCR cycles in tagmentation PCR.

To save even further on general plastic consumption and tip use, we designed a way of tagmention & indexing in which one prepares predispensed "tagmentation" plates containing premixed dessicated indexes. Hence you can with the same set of tips prepare multiple tagmentation plates containing the same indexes for storage. Of course it is very important to note that you need to change tips when/if you change premixed source indexes so as to not mix/contaminate your premixed stock indexes with different set of indexes.

This method is the most cost-effective implementation of Smart-seq3xpress and especially suited for large-scale projects or preparing large amounts of plates in advance.

Depending on your premixed index plate concentrations, dispense xµL of indexes aiming for a concentration of between 0.2micromolar (µM) -0.5micromolar (µM) for a 5uL final volume reaction. It can also be performed in a 7.5uL final volume reaction, simply adjust the concentration accordingly to fit the higher volume

Examples for a 5uL final reaction volume

For 0.5micromolar (µM) premixed indices add 2.5µL to each well of a 384 well PCR plate (results in0.25micromolar (µM)) .

For 1.0micromolar (µM) premixed indicies add 1.25µL to each well of a 384 well PCR plate. (0.25micromolar (µM) ).

Put the PCR plates onto an open thermocycler and incubate at 95°C without seal for 0h 1m 0s - 0h 5m 0s depending on the amount if index primers added to each well.

In the process or after visually inspect that all the liquid has evaporated from the wells.

Seal and store the plates at -20°C until use.

1. Transfer 1µL of prediluted cDNA from step 10 into a 384 well PCR plate containing dessicated index primers.

2. Prepare tagmentation mix (see Step 11 for the recipe for 4xTagmentation buffer). Shown mix is for 500 samples.

NB !!! TDE1 Tn5 enzyme comes as a glycerol viscous solution. When pipetting these small uL amounts to your tagmentation mix, be careful to not transfer extra drops that are stuck to the pipette tip. This can cause overtagmentation, which is not bad, but can cause less than expected UMI reads to be captured in sequencing.

*This is a suggestive amount of Tn5, but overall a good starting point that works well with HEK cells (12xPCR preamplification) and PBMCs (16xPCR preamplification). The amount can be changed to meet user specific criteria in terms of ratio of UMI containing reads vs Internal reads, and can be affected by cell-type, amount of preamplification PCR cycles given.

3. Dispense 1µL of Tagmentation Mix to each well of the 384 well plate containing 1µL of cDNA. Seal and pulse centrifuge the plate down at 1000x g,0h 0m 0s .

4. Incubate the plate at 55°C for 0h 10m 0s in a thermocycler.

5. To stop the tagmentation reaction and strip off the Tn5 enzyme, add 0.5µL of 0.2% SDS to each well. Seal the plate and quickly centrifuge the plate down before incubation at Room temperature for 0h 5m 0s

6. Prepare Index PCR mix. Again the calculated amount here for ease is for 500 sample. Please scale this to suit your specific dead volume.

7 Add either 2.5µL or 5µL of index PCR mix to each well depending on chosen final volume, seal the plates, apply quick centrifugation to settle everything in the bottom, and incubate in a thermocycler as following:

*The amount of PCR cycles needed depends on the starting material, preamplifation cycles given, and amount of tagmentation performed (amount TDE1 Tn5 used per cell). Therefore some user specific optimization might be necessary to suit the needs and wants and final material required.

However as a point of reference:

• HEK cells (12xPCR cycles in preamplification PCR) we use a standard 12xPCR cycles in tagmentation PCR

• PBMCs (10-18xPCR cycles in preamplification PCR) we use 14xPCR cycles in tagmentation PCR.

To quickly pool a plate or multiple plates together we designed a centrifugaton holder heavily inspired from Quartz-seq2, although more accessible. This holder can easily be 3D printed, and fits together with and a standard SBS format 384 well plate of choice. Design and template can be found here https://github.com/sandberg-lab/Smart-seq3xpress .

Put the holder around the reservoir, and add onto the PCR plate containing the final tagmented libraries upside down. Liquid should stay in the wells, by surface tension until centrifugation, but again do this with a slight amount of caution. See pictures if in doubt.

Once this is assembled, put in a centrifuge and pulse to max 100x g,0h 0m 0s .

Collect the pooled library in a fitting tube (depending on how many plates you choose to spin in the same reservoir) and purify the library with 22% PEG Clean-up beads or similar at a ratio of 0.7 : 1 beads to sample .

Mix the beads and sample gently by pipetting up and down and incubate for 0h 8m 0s at Room temperature

Place on magnet and let beads settle for roughly 0h 5m 0s to 0h 10m 0s before discarding the supernatant

Wash twice with freshly made 80% Ethanol

Remove Ethanol and let the bead pellet air dry for at least 0h 5m 0s (Until the pellet is no longer shiny)

Elute the beads in 40µL of UltraPure Water, resuspend beads and incubate for 0h 5m 0s

Use Qubit fluorometer (Qubit dsDNA HS Assay) or similar to quantify the final library pool / pools.

Run the final library on an Agilent Bioanalyzer (High Sensitivity DNA chip) to inspect quality and get median base-pair length information.

Citation

The sequencing ready library should be sequenced on any MGI or Illumina compatible sequencer, either Single-end or Paired-end, depending on the question and need.

If sequencing on a MGI sequencer convert the final library into MGI compatible single stranded circles utililizing the Universal Library Conversion Kit (App-A). Follow the user manual.

Table of Sequencing primers and oligos needed for MGI sequencing of Nextera Style libraries.

For primary data processing we highly recommend using the zUMIs pipeline. There are many specifics and features in the pipeline that are directly geared towards preparing and preprocessing Smart-seq3 data properly.

Software

For data sequenced on Illumina sequencers:

When sequencing has completed convert the binary base-call files (BCL) to fastq files.

Use bcl2fastq (bcl2fastq v2.20).

Software

Fastq files for zUMIs should be processed without demultiplexing in the following manner, instead of demultiplexing per cell.

#Example of preparing data from a 150bp paired-end sequencing run 
bcl2fastq --use-bases-mask Y150N,I8,I8,Y150N --no-lane-splitting --create-fastq-for-index-reads -R /mnt/storage1/NextSeqNAS/191011_NB502120_0154_AHVG7JBGXB

Remove the flag --no-lane-splitting if the cell barcodes or index primers have been reused on the different lanes.

Fastq files are now ready and compatible with the zUMIs pipeline. An example of how config .yaml files should look like for zUMIs can be seen below.

#Smartseq3xpress.yaml (Illumina Paired-end 150bp) 
project: Smartseq3xpress
sequence_files:
  file1:
    name: /Smartseq3xpress/fastq_files/Read1.fastq.gz
    base_definition:
      - cDNA(25-150)
      - UMI(12-21)
    find_pattern: ATTGCGCAATG;2
  file2:
    name: /Smartseq3xpress/fastq_files/Read2.fastq.gz
    base_definition:
      - cDNA(1-150)
  file3:
    name: /Smartseq3xpress/fastq_files/Index1.fastq.gz
      - BC(1-10)
  file4:
    name: /Smartseq3xpress/fastq_files/Index2.fastq.gz
      - BC(1-10)
reference:
  STAR_index: /genomes/Human/STAR7idx_noGTF/
  GTF_file: /genomes/Human/Homo_sapiens.GRCh38.95.chr.gtf
  additional_STAR_params: '--clip3pAdapterSeq CTGTCTCTTATACACATCT'
  additional_files:
out_dir: /Smartseq3xpress/zUMIs
num_threads: 20
mem_limit: 50
filter_cutoffs:
  BC_filter:
    num_bases: 4
    phred: 20
  UMI_filter:
    num_bases: 3
    phred: 20
barcodes:
  barcode_num: ~
  barcode_file: /Smartseq3xpress/expected_barcodes.txt
  automatic: no
  BarcodeBinning: 1
  demultiplex: no
  nReadsperCell: 100
  discardReads: yes
counting_opts:
  introns: yes
  downsampling: '0'
  strand: 1
  Ham_Dist: 1
  velocyto: no
  primaryHit: yes
  twoPass: no
make_stats: yes
which_Stage: Filtering
samtools_exec: samtools
pigz_exec: pigz
STAR_exec: STAR
Rscript_exec: Rscript

For data sequenced on a MGI sequencer:

Fastq files should be available after completed sequencing run, in the following format, that is directly compatible with zUMIs pipeline; Read_1.fastq, Read_2.fastq. The index barcodes are located as the last bases of read2.

#Smartseq3xpress.yaml (MGI, Paired-end 150bp) 
project: Smartseq3xpress
sequence_files:
  file1:
    name: /Smartseq3xpress/fastq_files/Read1.fastq.gz
    base_definition:
      - cDNA(25-150)
      - UMI(12-21)
    find_pattern: ATTGCGCAATG;2
  file2:
    name: /Smartseq3xpress/fastq_files/Read2.fastq.gz
    base_definition:
      - cDNA(1-150)
      - BC(151-170)
reference:
  STAR_index: /genomes/Human/STAR7idx_noGTF/
  GTF_file: /genomes/Human/Homo_sapiens.GRCh38.95.chr.gtf
  additional_STAR_params: '--clip3pAdapterSeq CTGTCTCTTATACACATCT'
  additional_files:
out_dir: /Smartseq3xpress/zUMIs
num_threads: 20
mem_limit: 50
filter_cutoffs:
  BC_filter:
    num_bases: 4
    phred: 20
  UMI_filter:
    num_bases: 3
    phred: 20
barcodes:
  barcode_num: ~
  barcode_file: /Smartseq3xpress/expected_barcodes.txt
  automatic: no
  BarcodeBinning: 1
  demultiplex: no
  nReadsperCell: 100
  discardReads: yes
counting_opts:
  introns: yes
  downsampling: '0'
  strand: 1
  Ham_Dist: 1
  velocyto: no
  primaryHit: yes
  twoPass: no
make_stats: yes
which_Stage: Filtering
samtools_exec: samtools
pigz_exec: pigz
STAR_exec: STAR
Rscript_exec: Rscript

FOR MORE INFO:

To get a quick look and generate a small stats file with the most common metrics from the zUMIs output data, one can do as following in R.

#Example of how to quickly generate an overview of zUMIs output data in R 
library(data.table)
dge <- readRDS("/Smartseq3xpress/zUMIs/zUMIs_output/expression/Smartseq3xpress.dgecounts.rds")
stats <- fread("/Smartseq3xpress/zUMIs/zUMIs_output/stats/Smartseq3xpress.readspercell.txt")[!RG %in% "bad"]
stats <- dcast(stats, RG~type, value.var = "N")
stats[, nreadpairs := Ambiguity+Exon+Intergenic+Intron+Unmapped]
reads <- fread("/Smartseq3xpress/zUMIs/zUMIs_output/Smartseq3xpresskept_barcodes_binned.txt.BCUMIstats.txt")
reads[, UMIfraction := nUMItag/(nNontagged+nUMItag)]
stats <- merge(stats,reads, by.x="RG", by.y="XC")
exonic_UMIs <- as.matrix(dge$umicount$exon$all)
exonic_Reads <- as.matrix(dge$readcount$exon$all)
complexity <- data.table( RG = colnames(exonic_UMIs),
                          nGenes = colSums(exonic_Reads>0),
                          nGenesUMIs = colSums(exonic_UMIs>0), 
                          nUMIs = colSums(exonic_UMIs))
stats <- merge(stats,complexity, by = "RG")
stats[,pct_coding := ((Exon+Intron)/nreadpairs)*100]