Protocol for flow cytometry-assisted single-nucleus RNA sequencing of human and mouse adipose tissue with sample multiplexing

Summary

Adipocyte size and fragility and commercial kit costs impose significant limitations on single-cell RNA sequencing of adipose tissue. Accordingly, we developed a workflow to isolate and sample-barcode nuclei from individual adipose tissue samples, integrating flow cytometry for quality control, counting, and precise nuclei pooling for direct loading onto the popular 10× Chromium controller. This approach can eliminate batch confounding, and significantly reduces poor-quality nuclei, ambient RNA contamination, and droplet loading-associated reagent waste, resulting in pronounced improvements in information content and cost efficiency.

Graphical abstract

Highlights

• •Nuclei and attached ribosome isolation from fresh/frozen human/mouse adipose tissue
• •Sample indexing with oligo-tagged nuclear antibodies for multiplexing
• •Chromatin staining and flow cytometry provide QC, counting, and balanced pooling
• •Optimized workflow generates pool for direct loading onto 10× Chromium device

Publisher’s note: Undertaking any experimental protocol requires adherence to local institutional guidelines for laboratory safety and ethics.

Adipocyte size and fragility and commercial kit costs impose significant limitations on single-cell RNA sequencing of adipose tissue. Accordingly, we developed a workflow to isolate and sample-barcode nuclei from individual adipose tissue samples, integrating flow cytometry for quality control, counting, and precise nuclei pooling for direct loading onto the popular 10× Chromium controller. This approach can eliminate batch confounding, and significantly reduces poor-quality nuclei, ambient RNA contamination, and droplet loading-associated reagent waste, resulting in pronounced improvements in information content and cost efficiency.

Before you begin

The protocol describes the steps to isolate nuclei from 50–300 mg of flash-frozen human adipose tissue samples or a single murine fat pad. For human samples, it has been optimized and validated for subcutaneous white adipose tissue (WAT) from the periumbilical region and visceral WAT samples from the preperitoneal space and omentum across a broad range of ages (21–73) and Body Mass Indices (22.1–60.3). We have additionally obtained similar yields/quality from human lymphedema and perivascular adipose tissue samples. For mouse samples, we performed optimization and validation for subcutaneous inguinal WAT and visceral perigonadal WAT samples from lean, obese High Fat Diet-fed mice of both sexes on a C57BL/6J background (ages up to 6 months). Notably, in both species, we observe higher nuclear yields, cDNA quality, and transcript complexity from visceral compared to subcutaneous adipose samples. In this protocol, we employ the term "adipose tissue" to encompass any of the above depots.

• 1.Adipose tissue should be flash frozen in liquid nitrogen as soon as possible after harvesting.

Note: In our studies, human adipose samples are frozen in the operating room immediately after harvesting. For murine samples, harvest samples one at a time to minimize time to freeze.

• 2.After initial flash freezing, samples can be stored at −80°C for at least 1 year without noticeable loss of quality.
• 3.The protocol has been optimized for sample of 50–300 mg, so for larger samples, it is best to portion prior to freezing.

Note: If the frozen sample is > 300 mg, divide the sample and run in parallel; excess tissue/volume ratios increase nuclear aggregation and ambient RNA detection and reduce hash antibody labeling and cDNA synthesis efficiency.

• 4.For hashed runs, up to 24 individual samples can be combined on a single 10× Chromium lane using the available BioLegend Total-Seq barcoded antibodies.

Note: We have not tested the compatibility of the 10× Genomics 3′ CellPlex multiplexing system.

• 5.Care should be taken in experimental organization and hashing setup to minimize batch confounding.¹
• 6.This protocol is optimized to work with snap-frozen and fresh samples. We have tested fresh murine samples, which show comparable quality results.

Note: If using fresh tissue, keep it in ice-cold PBS until use.

• 7.With only slight modifications to nuclei isolation steps as noted below, this workflow has been applied to murine liver and hypothalamus samples, showing similar improvements (increased transcript/reduced ambient detection per nucleus) as observed in adipose tissue.

Institutional permissions

All human and murine studies should be performed in accordance with relevant institutional and national guidelines and regulations. Here, informed consent was obtained and human adipose tissue was collected under the BIDMC Committee on Clinical Investigations Institutional Review Board 2011P000079 and University of Pittsburgh Medical Center STUDY 19010309. Murine experiments were approved by the BIDMC IACUC and carried out in accordance with the Declaration of Helsinki. Participants provided written informed consent for their study participation.

General preparation

Timing: 30 min

• 8.
  Prepare workstation:

  • a.
    Prepare an RNase-clean workspace. Spray and wipe down bench, pipettors, and instrument surfaces using RNase AWAY surface decontaminant, followed by wipe down with 70% ethanol to remove remaining RNase AWAY.
  • b.
    Pre-cool swinging-bucket centrifuge to 4 °C at least 10–15 min prior to tissue lysis.
  • c.
    Set up 2 ice buckets/CoolBox systems, one for reagents and one for sample preparation.
  • d.
    Turn on gentleMACS Dissociator and set it to the “mr_adipose_01” program.
  • e.
    Thaw at room temperature the following: single cell 3′ v3.1 Gel Beads, RT Reagent B, Template Switch Oligo (TSO), and Reducing Agent B.

    CRITICAL: If using a new kit, add 80 μL Low TE to the TSO and vortex for 15 seconds.

    Note: “mr_adipose_01” is not standardly provided with gentleMACS instrument. Contact macstec@miltenyibiotec.de for information on installation.

  • f.
    Prechill the following per sample: 1 x 50 mL conical tube, 1 x gentleMACS C-tube, 1 × 40 μm Nylon Strainer, 1 × 20 μm pre-separation filters, and 2 x 5 mL FACS tube.
• 9.Prepare 50 mL 2× ST buffer stock on ice (can be stored at 4°C, for up to 1 month).
• 10.Prepare the following solutions on ice, on the day of the experiment: TST buffer (2 mL per sample), 1× ST buffer (3 mL per sample), and nuclei resuspension buffer (NRB) (6 mL per sample).

Note: Buffers containing RNase inhibitor are freshly prepared to ensure its maximal activity.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
TotalSeq-A0451 anti-nuclear pore complex proteins hashtag 1 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682205
TotalSeq-A0452 anti-nuclear pore complex proteins hashtag 2 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682207
TotalSeq-A0453 anti-nuclear pore complex proteins hashtag 3 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682209
TotalSeq-A0454 anti-nuclear pore complex proteins hashtag 4 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682211
TotalSeq-A0455 anti-nuclear pore complex proteins hashtag 5 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682213
TotalSeq-A0456 anti-nuclear pore complex proteins hashtag 6 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682215
TotalSeq-A0457 anti-nuclear pore complex proteins hashtag 7 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682217
TotalSeq-A0458 anti-nuclear pore complex proteins hashtag 8 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682219
TotalSeq-A0459 anti-nuclear pore complex proteins hashtag 9 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682221
TotalSeq-A0460 anti-nuclear pore complex proteins hashtag 10 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682223
TotalSeq-A0461 anti-nuclear pore complex proteins hashtag 11 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682225
TotalSeq-A0462 anti-nuclear pore complex proteins hashtag 12 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682227
TotalSeq-A0463 anti-nuclear pore complex proteins hashtag 13 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682229
TotalSeq-A0464 anti-nuclear pore complex proteins hashtag 14 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682231
TotalSeq-A0465 anti-nuclear pore complex proteins hashtag 15 antibody (0.5 mg/mL stock, 1:250 final)	BioLegend	Cat # 682233
Chemicals, peptides, and recombinant proteins
UltraPure H2O	Thermo Fisher Scientific	Cat # AM9759
1 M magnesium chloride (MgCl2)	Sigma-Aldrich	Cat #M1028
5 M sodium chloride (NaCl)	Thermo Fisher Scientific	Cat # AM9759
1 M Tris hydrochloride (Tris-HCl)	Thermo Fisher Scientific	Cat # 15567027
1 M calcium chloride (CaCl2)	VWR	Cat # 97062-820
2% Bovine serum albumin (BSA)	New England Biolabs	Cat #B9000S
Protector RNase inhibitor	Sigma-Aldrich	Cat # 3335402001
10% Tween 20	Bio-Rad	Cat # 1610781
Phosphate-buffered saline (PBS) (1×, pH 7.4)	Thermo Fisher Scientific	Cat # 10010023
Invitrogen NucBlue Live ReadyProbes reagent (Hoechst 33342)	Thermo Fisher Scientific	Cat #R37605
Glycerine (glycerol), 50% (v/v)	Thermo Fisher Scientific	Cat # 329032
SPRIselect bead-based reagent	Beckman Coulter	Cat #B23319
Buffer EB	QIAGEN	Cat # 19086
Ethyl alcohol, pure, 200 proof	Sigma-Aldrich	Cat #E7023-500ML
Invitrogen TE buffer (low EDTA)	Thermo Fisher Scientific	Cat # 12090015
Kapa HiFi HotStart PCR Kit	Roche	Cat # 07958897001
RNase AWAY surface decontaminant	Thermo Fisher Scientific	Cat # 7003
Critical commercial assays
Chromium Next GEM Chip G Single Cell Kit (48 rxns)	10× Genomics	Cat # 1000120
Chromium Next GEM Single Cell 3′ Kit v3.1 (16 rxns)	10× Genomics	Cat # 1000268
Qubit 1× dsDNA HS Assay Kit	Thermo Fisher Scientific	Cat # Q33231
Agilent High Sensitivity DNA Kit	Agilent	Cat # 5067-4626
NextSeq 500/550 v2.5 Seq 75/150/300 cycle kit	Illumina	Cat # 20024906/7/8
Deposited data
Raw and analyzed data	Emont et al., 20222	GEO: GSE176171 and https://singlecell.broadinstitute.org/single_cell/study/SCP1376/a-single-cell-atlas-of-human-and-mouse-white-adipose-tissue#study-summary
Experimental models: Organisms/strains
Mus musculus C57BL/6J, ages 8–26 weeks, male and female, fed chow or high-fat diet for 8–12 weeks	The Jackson Laboratory	Strain # 000664
Oligonucleotides
Single Index Kit T Set A (for Gene Expression Libraries)	10× Genomics	Cat # 1000213
HTO additive primer v2 5′GTGACTGGAGTTCAGACG TGTGCTCTTCCGAT∗C∗T	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
SI-PCR primer 5′AATGATACGGCGACCACCG AGATCTACACTCTTT CCCTACACGACGC∗T∗C	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D701_LONG primer 5′CAAGCAGAAGACGG CATACGAGATCGAGTAAT GTGACTGGAGTTCAG ACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D702_LONG primer 5′CAAGCAGAAGACG GCATACGAGATTCTCCGGA GTGACTGGAGTT CAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D703_LONG primer 5′CAAGCAGAAGACG GCATACGAGATAATGAGCG GTGACTGGAG TTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D704_LONG primer 5′CAAGCAGAAGACG GCATACGAGATGGAATCTC GTGACTGGA GTTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D705_LONG primer 5′CAAGCAGAAGAC GGCATACGAGATTTCTGAAT GTGACTGGAGT TCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D706_LONG primer 5′CAAGCAGAAGAC GGCATACGAGATACGAATTC GTGACTGGAGT TCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D707_LONG primer 5′CAAGCAGAAGAC GGCATACGAGATAGCTTCAG GTGACTGGAGTT CAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D708_LONG primer 5′CAAGCAGAAG ACGGCATACGAGATGCGCATTA GTGAC TGGAGTTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D709_LONG primer 5′CAAGCAGAAGACG GCATACGAGATCATAGCCG GTGACTGG AGTTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D7010_LONG primer 5′CAAGCAGAAG ACGGCATACGAGATTTCGCGGA GTGACTGGA GTTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D7011_LONG primer 5′CAAGCAGAAG ACGGCATACGAGATGCGCGAGAGTGACTG GAGTTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
TruSeq D7012_LONG primer 5′CAAGCAGAAG ACGGCATACGAGATCTATCGCT GTGACT GGAGTTCAGACGTGTGCTC	BioLegend TotalSeq Protocol	https://www.biolegend.com/en-us/protocols/totalseq-a-dual-index-protocol
Software and algorithms
2100 Expert	Aglient	Cat # G2946CA
FlowJo for Mac	BD Biosciences	v 10.7
bcl2fastq2 Conversion Software	Illumina	v 2.20
Cell Ranger	10× Genomics	v 6.1.2
demuxEM	Gaublomme et al.3	v 0.1.7
CellBender	Broad Institute	v 0.2.0
Seurat	Stuart et al.4	v 3.9.9
Other
CoolBox 2XT all-day cooling and freezing workstation	Corning	Cat # 432025
gentleMACS Dissociator	Miltenyi Biotec	Cat # 130-093-235
Microcentrifuge 5424R	Eppendorf	Cat # 05410205
Swinging bucket centrifuge 5702R	Eppendorf	Cat # 5703000010
Chromium Controller	10× Genomics	Cat # PN120270
Mastercycler pro S thermal cycler	Eppendorf	Cat # 950030020
Qubit 3.0 fluorometer	Thermo Fisher Scientific	Cat #Q33216
2100 BioAnalyzer instrument	Agilent	Cat #G2939BA
NextSeq 500 system	Illumina	Cat # SY-415-1001
Invitrogen EVOS FL digital inverted fluorescence microscope	Thermo Fisher Scientific	Cat # 12-563-460
10× Magnetic separator A	10× Genomics	Cat # PN2000067
50 mL Falcon conical centrifuge tube	Corning	Cat # 352070
gentleMACS C-tube	Miltenyi Biotec	Cat # 130-096-334
40 μm nylon strainer	Corning	Cat # 352340
20 μm pre-separation filter	Miltenyi Biotec	Cat # 130-101-812
Falcon 5 mL round bottom high clarity PP test tube	Corning	Cat #352063
Armadillo PCR plate, 96-well	Thermo Fisher Scientific	Cat # AB2396
Pipette tips RT LTS 1000 μL wide-orifice filter	Mettler-Toledo Rainin	Cat # 30389218
Pipette tips RT LTS 1000 μL, filter	Mettler-Toledo Rainin	Cat # 30389213
Pipette tips RT LTS 200 μL, filter	Mettler-Toledo Rainin	Cat # 30389240
Pipette tips RT LTS 20 μL, filter	Mettler-Toledo Rainin	Cat # 30389226
Fisherbrand sterile polystyrene disposable serological pipets (5, 10, 25, 50 mL)	Thermo Fisher Scientific	Cat # 15-676-10C, 10F, 10M, 10Q
Applied Biosystems MicroAmp clear adhesive film	Thermo Fisher Scientific	Cat # 43-063-11
Qubit assay tubes	Thermo Fisher Scientific	Cat #Q32856
C-chip disposable hemocytometer	VWR	Cat # 82030-468
Corning cryogenic vial with orange cap	Thermo Fisher Scientific	Cat # 50-197-4471
Eppendorf DNA LoBind tube, 1.5 mL	USA Scientific	Cat # 4043-1021

Materials and equipment

2× salt-Tris (ST) buffer

Reagent	Final concentration	Amount
Ultrapure H2O	N/A	43.88 mL
5 M NaCl	292 mM	2.92 mL
1 M MgCl2	42 mM	2.1 mL
1× Tris-HCl pH 7.5	20 mM	1 mL
1 M CaCl2	2 mM	0.1 mL
Total		50 mL

Store at 4°C, up to 1 month.

Tween with salt and Tris (TST) buffer

Reagent	Final concentration	Amount
Ultrapure H2O	N/A	974 μL
2× ST	N/A	1 mL
Protector RNase Inhibitor	0.2 U/μL	10 μL
2% BSA	0.01%	10 μL
10% Tween-20	0.3%	6 μL
Total		2 mL

Store at 4°C, up to 12 h.

1× ST buffer

Reagent	Final concentration	Amount
Ultrapure H2O	N/A	1.485 mL
2× ST	N/A	1.5 mL
Protector RNase inhibitor	0.2 U/μL	15 μL
Total		3 mL

Store at 4°C, up to 12 h.

Nuclei resuspension buffer (NRB)

Reagent	Final concentration	Amount
1× PBS pH 7.4	N/A	5.91 mL
Protector RNase inhibitor	0.2 U/μL	30 μL
2% BSA	0.02%	60 μL
Total		6 mL

Store at 4°C, up to 12 h.

Collection buffer

Reagent	Final concentration	Amount
Protector RNase inhibitor	1 U/μL	1.9 μL
RT Reagent B		22.7 μL
Total		24.6 μL

Store at 4°C, up to 2 h.

Step-by-step method details

Tissue homogenization, nuclei isolation, and hashtag labeling

Timing: 1–2 h

The following steps describe the nuclei isolation, filtration, and wash steps required to achieve a clean nuclei preparation. Nuclei from individual samples are tagged with a common hashtag oligo (HTO) and stained with NucBlue (Hoechst 33342) prior to the assessment of nuclear quality by fluorescence-activated nuclear sorting (FANS). A tween-based buffer (TST) shown to preserve the association of rough endoplasmic reticulum with the nuclear membrane facilitates the capture of mature mRNA and recovers nuclei from the broadest range of cell and tissue types.^5,6

• 1.Place 50–300 mg of fresh or frozen adipose tissue into gentleMACS C-Tube with 2 mL of ice-cold TST buffer (Figure 1A).

Note: Frozen adipose tissue samples of this size do not require mincing or thawing

• 2.
  Click tube onto gentleMACS dissociator.

  • a.
    Run the program "mr_adipose_01" (268 rotations per run) three times, for a total of 1 min and 48 s homogenization time.
  • b.
    Tap suspension to the bottom of the tube and incubate on ice for 10 min (Figure 1B).

Note: Although this protocol has been optimized for use with the gentleMACS dissociator, alternative homogenization methods, such as mincing, douncing, and mortar/pestle with liquid nitrogen can be employed with minor adjustments.

• 3.
  Set a 40 μm strainer atop a 50 mL conical tube on ice and pre-wet the filter with 150 μL of 1× ST buffer.

  • a.
    Using a wide-bore, low retention p1000 tip, transfer suspension to the pre-wet 40 μm nylon filter without pipette mixing.

    Note: You will end up leaving some suspension and tissue behind in the C-tube; that is fine.

  • b.
    Using a wide-bore, low retention p1000 tip, rinse the C-tube with ice-cold 1 mL 1× ST buffer and transfer this suspension to the filter.
  • c.
    Additionally, rinse the filter with 2 mL 1× ST buffer.

    Note: Some fat and other detritus will be left behind on the filter.

  • d.
    Discard filter (Figure 1C).
• 4.
  Set a 20 μm pre-separation filter atop a 5 mL FACS tube and pre-wet the filter with 50 μL of 1× ST buffer.

  • a.
    Using a wide-bore, low retention p1000 tip, transfer all suspension to the filter without pipette mixing.

    Note: If suspension is too saturated and doesn’t easily pass through 20 μm filter, gently tap FACS tube on the ice to facilitate filtration process.

  • b.
    Discard Filter and cap FACS tube.
  • c.
    Immediately spin down in a swinging-bucket centrifuge at 500 g for 5 min at 4°C, with the centrifuge brake set to half-maximal.

    CRITICAL: Be careful not to disturb pellet when returning tube to ice bucket. Pellet may or may not be visible.

    Note: If using an ice bucket, be careful not to disturb the pellet when returning tube from centrifuge. Use an empty FACS tube to pre-form wells in the ice for the sample tubes to be placed.

    Note:This is a good time to start centrifuging hashtag antibody tubes in a fixed angle rotor at 14,000 g at 4°C for 10 min prior to adding to nuclear suspensions.

• 5.
  Remove all but ∼200 μL of supernatant (make sure to not disturb the pellet) using a 5 mL serological pipette, making sure to remove the lipid layer at the top of supernatant (Figure 1D).

  • a.
    Using a 5 mL serological pipette, slowly add 5 mL of ice-cold NRB to the side of the FACS tube, being careful not to disturb the pellet.
  • b.
    Immediately spin down in a swinging-bucket centrifuge at 500 g for 5 min at 4°C, with the centrifuge brake set to half-maximal.

Note: Be careful not to disturb pellet when returning tube to ice bucket (Figure 1E).

• 6.
  Remove all supernatant, taking care not to disturb the pellet.

  Note: Some residual supernatant is fine if the pellet is not visible or small; try to remove as much as you can.

  • a.
    Using a wide-bore, low retention p1000 tip, resuspend pellet in 500 μL of NRB for each hashtag antibody the sample will be split into (i.e., resuspend in 1 mL of NRB if splitting a sample between two hashtag antibodies).

    Note: When performing a hash split, aim to use > 8 unique hashtags to increase barcode complexity and maximize the benefits of super-loading. This may require splitting each sample amongst two or more unique hashtags.

  • b.
    Using a wide-bore, low retention p1000 tip, transfer nuclear suspension to new 5 mL FACS tubes, dividing sample into multiple 500 μL aliquots if splitting a sample between multiple hashtags.
• 7.Add 1 drop of NucBlue stain to each 500 μL suspension.

Optional: A small 10 μL aliquot of the nuclei suspension can be loaded on hemocytometer and visualized to check the nuclei preparation. An example is given in Figure 2.

• 8.
  Avoiding the bottom of the antibody tube, add 2 μL of hashtag antibody (use the 0.5 mg/mL stock concentration) to each suspension for a final concentration of 2 μg/mL (1:250 stock dilution).

  • a.
    Mix gently (can use low-speed vortex).
  • b.
    Incubate on ice for 45 min.
  • c.
    Spin down in a swinging-bucket centrifuge at 500 g for 5 min at 4°C, with the centrifuge brake set to half-maximal.
• 9.Remove 450 μL of supernatant, being careful not to disturb the pellet.
• 10.Using a wide-bore, low retention p1000 tip, resuspend each nuclear pellet in 450 μL of NRB (Figure 1F).

Note: We find that a single experienced experimenter can process up to 8 samples concurrently within the 1–1.5 h time frame, minimizing RNA degradation before FANS. To minimize variation in thaw-to-FANS time when hashing > 8 samples, we recommend staggering sample preparation. We begin processing the second batch of samples while the first batch is incubating on ice for 45 min during step 8b. Generally, the end of the first sample set incubation coincides with the start of the second sample set, and the second set can be completed while the first set is sorted. If unable to process 8 samples within this window, we highly recommend engaging an additional experimenter.

Figure 1.

Adipose tissue homogenization and nuclei isolation

(A) Adipose tissue prior to homogenization (step 1).

(B) Adipose tissue finely homogenized (step 2).

(C) Fat and other detritus left behind on 40 μm strainer (step 3c).

(D) Floating lipid layer following the first centrifugation (step 5).

(E) Precipitated nuclear pellet following the second centrifugation (step 6).

(F) Resuspended nuclear pellet (step 9).

Figure 2.

Visualization of nuclei isolated from adipose tissue

(A) Bright-field channel view of isolated adipose tissue nuclei.

(B) DAPI channel view of isolated adipose tissue nuclei.

(C) Bright-field and DAPI channels overlay view of isolated adipose tissue nuclei. Arrowheads indicate cell debris and arrows indicate nuclear aggregates. Nuclei are imaged at 10× magnification.

Fluorescence-activated nuclei sorting and sample pooling

Timing: 30–50 min (will vary based on the number of samples and nuclear concentration)

The following steps outline the process of configuring a flow cytometer to simultaneously: 1) remove debris and ambient RNA, 2) perform a nuclear quality control (QC) check, enriching for high-quality nuclei, and 3) count and pool nuclei from individual samples into a concentrated pool that is directly loadable onto the 10× Chromium device, ensuring targeted representation from individual samples.

In integrating sample multiplexing and flow cytometry into our workflow, this protocol aims to account for two key considerations: 1) Sample hashing detection of droplets containing multiple nuclei allows “super-loading” to minimize reagent waste. 2) As centrifugation can result in significant nuclear loss, centrifugation between FANS counting/pooling and instrument loading should be avoided.

• 11.
  Calculate nuclear loading concentration and collection buffer volume:

  • a.
    The maximum recommended number of cells that can be profiled using the standard 10× Single Cell Gene Expression is 10,000; this protocol aims to double that to 20,000 nuclei.
  • b.
    As a general principle, the number of nuclei loaded onto the 10× Chromium Controller should be ∼ twice the target recovery number, hence load ∼40,000 nuclei.
  • c.
    Sort nuclei directly into the 10× RT Reagent B with an addition of RNase inhibitor (1 U/μL) to reduce transcriptome degradation of collected nuclei during FANS.
  • d.
    Estimate droplet size using the following equations and constants derived from manufacturer specs.

    Note: For a Beckman Coulter MoFlo Astrios EQ with a 70 μm nozzle configuration:

    $$dropdiam=Instrumentconstant\times streamdiam=1.89\times .065mm=0.1229mm$$

    $$dropvolume=\frac{4}{3}\times \pi \times {\left(\frac{dropdiam}{2}\right)}^3=\frac{4}{3}\times \pi \times {\left(\frac{0.1229mm}{2}\right)}^3=0.00097\mu L=\sim 1nL$$

    CRITICAL: Depending on the make and model of the flow cytometer being used, it may be necessary to alter flow rates/nozzle sizes to provide a sufficiently concentrated pool. While individual cytometer manufacturers may offer ballpark estimates, empirical measurement of the droplet volume on the machine you intend to use provides the most accurate estimate.

  • e.
    In order to load 40,000 nuclei onto a single lane of the 10× GEM Chip G, aim to collect a larger total number of nuclei.

    Note:This approach accounts for potential variability in the volume of collected nuclei, which can be attributed to losses occurring between the detector and the collection tube, as well as liquid evaporation resulting from extended sorting times.

  • f.
    To account for this loss as well as to provide excess volume for ease of pipetting, collect the volume from 52,000 events.

    Note: This amounts to a maximum collected volume of ∼50.4 μL + 22.7 μL RT Reagent B (equal final concentration as standard 10× protocol) + 1.9 μL of RNase inhibitor (1 U/μL final concentration) = 75 μL.

  • g.
    If unable to collect the full 52,000 events, add Ultrapure H₂O needed to maintain the above concentrations.

    Note: Using this approach, we observe an average nuclear capture efficiency of ∼47% for adipose nuclei suspensions prepared by this protocol and loaded onto the 10× Chromium controller device.

    Note: When targeting > 20,000 nuclei total from a set of samples, instead of decreasing the number of multiplexed samples, generate a second pool by FANS using the remaining volume from samples and load it into a separate lane of 10× Chromium Controller.

• 12.Prepare collection plate by adding 24.6 μL collection buffer, add it into a single well of a 96-well plate, seal the plate, and spin it down in a plate centrifuge.

Note: Use a low-profile 96-well plate to allow plate-based platform to be elevated closer to the output nozzle, minimizing the distance between droplets and collection buffer.

• 13.Check registration of output flow immediately before initiating collection by placing a seal over the collection plate and ensuring collection stream hits the seal in the center of the target well.

CRITICAL: This is important to minimize nuclear loss onto the sides of the receiving receptacle.

Note: Ensure both input nuclear suspensions and pool output remain chilled. If an actively chilled collection platform is not available, pre-chill a collection block in the freezer overnight. Use adhesive film to cover the 96-well plate before and after sorting to minimize potential evaporation and contamination.

• 14.
  Gating strategy:

  • a.
    Set a FANS gate based on the Side SCatter-Height (SSC-H) and NucBlue intensity (405-488/59-Height) to select for chromatin-containing nuclei (R1) (Figure 3A).
  • b.
    Set a FANS gate based on the SSC-H and Forward SCatter-Height (FSC-H) to identify single nuclei (R2) (Figure 3B).
  • c.
    Set a FANS gate based on the FSC-H and Forward SCatter-Area (FSC-A) to enrich for high-quality single nuclei (R3) (Figure 3C).

Note: Side-scattered light is proportional to the internal complexity and smoothness of nuclear membrane, therefore particles with high SSC could be indicative of apoptotic nuclei and should be excluded. Forward-scattered light is proportional to the size of the particle, so nuclei display low FSC.

• 15.Sort each sample using the gates in 14a-c into the prepared collection plate (12) until the target events are reached before proceeding to the next one (i.e., if collecting nuclei from 8 hashtag-labeled samples, sort 6,500 nuclei from each one).

Note: On a Beckman Coulter MoFlo Astrios EQ with a 70 μm nozzle, maintaining a 61 PSI flow and efficiency > 85%, we observe a rate of ∼400–1000 events per second.

Note: When splitting samples among multiple hashtag antibodies, set the sort order such that one barcoded suspension from each sample is first run before sorting additional barcoded suspensions of the same sample. This approach minimizes the sample-to-sample variation associated with pre-sort degradation.

Note: Additional in-line assessment of fluorescence can be paired to further enrich/de-enrich specific target cell types (e.g., in mice harboring a recombinase-dependent nuclear marker or through antibody labeling of cell type-specific nuclear envelope proteins (e.g., NeuN antibody, which recognizes neuronal nuclei specifically).

CRITICAL: If not input-limited, sort on the “purify” abort setting to reduce the number of collected non-target particles.

Figure 3.

FANS gating strategy for adipose tissue nuclei sorting

(A) R1 is selected to get all intact nuclei separated from debris.

(B and C) R2 and R3 population is selected for high-quality single nuclei, rejecting doublets, or multiplets. The percent of events at each gate represents the average expected yield.

10× GEM generation

Timing: 30 min

• 16.After FANS collection, seal the 96-well plate with clear adhesive film for transfer on ice to the location of the 10× Chromium controller.
• 17.Using a low retention p200 tip, gently pipette mix FANS pool and transfer 62 μL to a strip tube.

CRITICAL: Adjust volume up to 62 μL with Ultrapure H₂O if experiencing volume loss or lower than expected nuclear yield from samples.

• 18.
  Add the following reagents to FANS pool:

  • a.
    2.4 μL Template Switch Oligo.
  • b.
    2 μL Reducing Agent B.
  • c.
    8.7 μL RT Enzyme C.
• 19.
  Follow 10× Genomics protocol (https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-single-cell-3-reagent-kits-user-guide-v-3-1-chemistry):

  • a.
    Begin at step 1.2a to load and run the chip on the 10× Chromium Controller.

    CRITICAL: Avoid creating bubbles when loading reagents onto Chromium Next GEM Chip G as they can result in a clog during partitioning of barcoded Gel Beads, nuclei, and enzymes into single-nuclei emulsion droplets.

  • b.
    Use a wide-bore, low retention p200 tip to slowly aspirate and transfer Gel Beads-in-emulsion (GEM) at steps 1.4e-g.

RT-PCR and cDNA generation

Timing: 2 h

• 20.
  Follow 10× Genomics protocol until cDNA amplification (https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-single-cell-3-reagent-kits-user-guide-v-3-1-chemistry; step 2.2a). To the cDNA PCR mix, add:

  • a.
    50 μL Amp Mix.
  • b.
    15 μL cDNA primer.
  • c.
    1 μL HTO additive primer v2 (0.2 μM).

Note: Add HTO primer to increase yield of HTO products during cDNA amplification.

• 21.Amplify cDNA with 13 PCR cycles.

Steps	Temperature	Time	Cycles
Initial Denaturation	98°C	3 min
Denaturation	98°C	15 s	13 cycles
Annealing	63°C	20 s
Extension	72°C	1 min
Final extension	72°C	1 min
Hold	4°C	Forever

Note: Perform the standard 13 cycles of cDNA amplification, as recommended for nuclei by 10× Genomics.

• 22.
  Perform 0.6× SPRI clean-up after amplification.

  • a.
    Instead of discarding supernatant at step 2.3d, save supernatant and transfer it to a new strip tube.

    Note: This supernatant contains HTO-derived cDNA fraction and can be stored on ice until the cleanup of the endogenous cDNA is done.

  • b.
    Endogenous transcript cDNA fraction is bound to the SPRI beads– continue cDNA cleanup with 10× Genomics protocol (https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-single-cell-3-reagent-kits-user-guide-v-3-1-chemistry; step 2.3e).
• 23.
  Purify HTOs using SPRI clean-up per manufacturer protocol:

  • a.
    Add 70 μL SPRIselect reagent to supernatant.
  • b.
    Incubate 5 min at room temperature.
  • c.
    Place strip tube on magnet and wait 5 min until solution is clear.
  • d.
    Carefully remove and discard supernatant.
  • e.
    Add 300 μL freshly made 80% ethanol to the tube without disturbing the beads and wait 30 s.
  • f.
    Carefully remove and discard the ethanol wash.
  • g.
    Add 200 μL freshly made 80% ethanol to the tube without disturbing the beads and wait 30 s.
  • h.
    Carefully remove and discard the ethanol wash.
  • i.
    Remove strip tube from magnet, briefly centrifuge, and return it to magnet.
  • j.
    Carefully remove and discard any remaining ethanol.
  • k.
    Air dry for 1–3 min until beads are no longer shiny.
  • l.
    Remove from magnet and immediately add 40 μL Buffer EB.
  • m.
    Resuspend beads by pipetting 15 times and incubate at room temperature for 5 min.
  • n.
    Place strip tube on magnet for 2 min until solution.
  • o.
    Transfer 39 μL into a 1.5 Eppendorf tube without carrying over any beads.

CRITICAL: To maximize recovery, perform washes with freshly prepared 80% ethanol.

• 24.Quantify both cDNA and HTO fractions with Qubit and cDNA fraction quality with Agilent Bioanalyzer or Agilent TapeStation.

Note: The quantity and average size of cDNA usually serve as summary indicators of the initial quality of the sample, the nuclear isolation, and FANS-assisted pooling. Excessive RNA degradation during any of these steps leads to reduced cDNA yields and shorter average cDNA size. In our experience, the cDNA concentration ranges from 10–40 ng/μL for endogenous transcript cDNA and 4–10 ng/μL for cDNA derived from HTOs.

Note: The expected average size of endogenous transcript cDNA can vary depending on the input tissue. The cDNA generated from subcutaneous WAT consistently exhibits a shorter size (∼1400 bp) compared to that from visceral WAT (∼1600 bp), in both human and mouse tissues (Figures 4A and 4B). When applying this protocol to mouse liver and brain, cDNA sizes of approximately 1,700 bp and 2,000 bp are observed (Figure 4C).

Note: HTO fraction will not produce a visible Bioanalyzer trace prior to PCR amplification during library construction.

Figure 4.

Bioanalyzer traces of cDNA from sorted nuclei isolated from various tissue inputs, including an example of an HTO library

Bioanalyzer trace showing cDNA fragments generated from sorted nuclei isolated from subcutaneous WAT (A), visceral WAT (B), and arcuate nucleus of the hypothalamus (C). (D) Bioanalyzer trace of HTO library.

mRNA and HTO library generation

Timing: 2 h

• 25.Follow the 10× Genomics protocol for the endogenous transcript cDNA fraction library construction (https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-single-cell-3-reagent-kits-user-guide-v-3-1-chemistry; step 3).

Note: For the HTO-derived cDNA fraction use the TotalSeq-A Antibodies and Cell Hashing protocol (https://www.biolegend.com/en-us/protocols/totalseq-a-antibodies-and-cell-hashing-with-10x-single-cell-3-reagent-kit-v3-3-1-protocol) with the following alterations.

• 26.
  Follow the TotalSeq protocol for HTO fraction library construction. To the Sample index PCR mix (step 3A), add:

  • a.
    50 μL 2× KAPA HiFi HotStart ReadyMix.
  • b.
    2.5 μL 10 μM SI PCR primer.
  • c.
    2.5 μL 10 μM TruSeq D70X_LONG.
  • d.
    20 ng HTO fraction.
  • e.
    Nuclease-free H₂O to bring total volume to 100 μL.
• 27.Amplify HTO with 11 PCR cycles.

Steps	Temperature	Time	Cycles
Initial Denaturation	98°C	2 min
Denaturation	98°C	20 s	11 cycles
Annealing	64°C	30 s
Extension	72°C	20 s
Final extension	72°C	5 min
Hold	4°C	forever

• 28.
  Perform 1.2X SPRI clean-up after amplification per manufacturer protocol:

  • a.
    Add 120 μL SPRIselect reagent.
  • b.
    Incubate 5 min at room temperature.
  • c.
    Place strip tube on magnet and wait 5 min until solution is clear.
  • d.
    Carefully remove and discard supernatant.
  • e.
    Add 300 μL freshly made 80% ethanol to the tube without disturbing the beads and wait 30 s.
  • f.
    Carefully remove and discard the ethanol wash.
  • g.
    Add 200 μL freshly made 80% ethanol to the tube without disturbing the beads and wait 30 s.
  • h.
    Carefully remove and discard the ethanol wash.
  • i.
    Remove strip tube from magnet, briefly centrifuge, and return it to magnet.
  • j.
    Carefully remove and discard any remaining ethanol. Air dry for 1–3 min until beads are no longer shiny.
  • k.
    Remove from magnet and immediately add 26 μL nuclease-free water.
  • l.
    Resuspend beads by pipetting 15 times and incubate at room temperature for 5 min.
  • m.
    Place strip tube on magnet for 2 min until solution.
  • n.
    Transfer 25 μL into a 1.5 Eppendorf tube without carrying over any beads.
• 29.Quantify both cDNA and HTO libraries with Qubit. Assess their quality with Agilent Bioanalyzer.

Note: In our experience, the library concentration ranges from 10–30 ng/μL for endogenous transcript library and 1–3 ng/μL for HTO library. We expect that the average size for endogenous transcript and HTO libraries will be around 460 bp and 180 bp, respectively (Figure 4D).

Note: A clean HTO library will show up as a predominant single peak at around 180 bp. A smaller peak at around 140 bp is indicative of carryover primers from cDNA amplification. Excess primer carryover will interfere with quantification but will not cluster during sequencing.

Sequencing and analysis

Timing: 2–3 days

• 30.Pool the endogenous transcript cDNA and HTO libraries at a ratio of 85% endogenous transcript cDNA and 15% of HTO.
• 31.
  Sequence the pool using the regular 10× Single Cell 3′ Gene Expression read structure:

  • a.
    For 75 cycles kit: Read 1 = 28 bp, Index 1 = 8 bp, Index 2 = N/A, Read 2 = 55 bp.
  • b.
    For 300 cycle kit: Read 1 = 150 bp, Index 1 = 8 bp, Index 2 = N/A, Read 2 = 150 bp.
• 32.Sequencing depth should meet 10× Genomics recommended minimum of 20,000 mean reads per nuclei.
• 33.Input FASTQ files into the Cell Ranger pipeline (10× Genomics, v.3.1.0) for demultiplexing and alignment of sequencing reads to the mm10 or hg38 transcriptome and creation of feature-barcode matrices.⁷

Optional: We include intronic alignments for gene expression analysis.

Note:Use the “remove-background” module from CellBender to estimate and remove counts representing ambient RNA molecules.⁸

• 34.Perform sample demultiplexing via HTO sequences with the Cumulus sc/snRNA-Seq processing pipeline.⁹
• 35.Quantify HTO with the Cumulus Tool on Feature Barcoding to generate a cell-by-HTO count matrix.
• 36.Use the HTO count matrix, along with the gene count matrices generated via Cell Ranger to assign each cell barcode to its respective sample(s) with the demuxEM program.
• 37.Keep only the cell barcodes that are identified as singlets.

Expected outcomes

This protocol provides detailed procedures for performing multiplexed single-nucleus RNA sequencing (snRNA-seq) on fresh and flash-frozen white adipose tissue (WAT) from both mouse and human samples, encompassing both subcutaneous and visceral depots. Typically, processing 50–300 mg of adipose tissue yields a high-quality nuclear count ranging from 6,000 to 60,000. The integration of hashtag antibodies and FANS enables sample multiplexing, resulting in reduced batch-to-batch variability, enrichment of high-quality single nuclei, and the assurance of reaching the intended target representation for each individual sample.

Using this protocol, we successfully identify and recover all major adipose cell types, including adipocytes and increased numbers of vascular cells (Figure 5), compared to single-cell approaches (Drop-Seq) (Figure 5).^2,10 We find no discernible differences in cell type representation between the 10× snRNA-seq protocol we used in Emont et al.² and the flow cytometry-assisted, sample hashing protocol we present here, indicating that the inclusion of hashing and sorting procedures does not introduce any additional cell type-specific recovery biases (Figure 5).

Figure 5.

Comparison of adipose tissue cell types across single cell/single nucleus RNA-seq techniques

Uniform manifold approximation and projection (UMAP) plot of cell types identified from equivalent adipose tissue samples inputs by Drop-seq single cell RNA-seq (left) or single nucleus RNA-seq on the 10× Chromium instrument using the Emont et al. (center) or this flow cytometry-assisted protocol with sample indexing (right).

In addition to increased super-loading nuclei yields and doublet removal as described previously,³ we observe the number of unique molecular transcripts (nUMI), a metric of transcriptome complexity, is higher in hashed snRNA-seq across all major adipose tissue cell types (Figure 6A), while mitochondrial gene percentage (mt.percent), a proxy indicator of ambient RNA in snRNA-seq, is substantially lower (Figure 6B). At the targeted 40,000 nuclei loading concentration, we generally observe a 20%–30% doublet rate.

Figure 6.

Comparison of per cell single cell/nucleus RNA-seq quality metrics across techniques

Comparing data from Drop-seq scRNA-seq, Emont et al. snRNA-seq, and this updated flow cytometry-assisted protocol, we observe the (A) Number of Unique Molecular Identifiers (nUMI) detected per nucleus, a measure of data complexity, is increased (40%–90%) in all cell types and (B) Percent mitochondrial reads (mt.percent), a correlate of ambient RNA, is reduced by >90%. Within each box, thick horizontal black lines denote the median values. Boxes extend from the 25th to the 75th percentile of each group's distribution of values; vertical extending lines denote the most extreme values within 1.5 interquartile range of the 25^th and 75^th percentile of each group; dots denote observations outside these ranges.

Limitations

As for most droplet-based single nucleus approaches, 1) there is gene dropout for genes with very low expression, given the limited starting material, 2) any cytoplasmic post-transcriptional regulation cannot be detected, and 3) the 3′ directed libraries provide limited ability to evaluate splice isoforms. 4) There remains the possibility of doublets/multiplets. Though mitigated by the use of hash antibodies to determine inter-sample multiplets prior to clustering, within-sample doublets still occur, especially if the number of hash barcodes used is low (see note above). Post-hoc doublet detection should still be performed.

More specifically for this protocol, one limitation is the added time to reverse transcription necessitated by hash oligo incubation time and the length of the FANS sort. For the adipose tissue depots shown, we are able to allow up to 2 h of sorting, but in tissues richer in RNases, the balance between the time of RNA degradation and time to QC nuclei needs to be determined. Another limitation is a starting input that is sufficiently large. Samples with limited nuclear yield increase time to RT-PCR due to lower nuclear concentration and are more affected by dead volume inherent to FANS. Further, in actively removing poor QC nuclei, there may be a bias induced in what kinds of cells are being assessed. Although we have not observed a cell type bias, we do note a reduction in the number of dividing cells. Lastly, this protocol utilizes access to multiple specialized instruments (gentleMACS Dissociator, cell sorter with low output volumes, and the 10× chromium device) that not only create a financial limitation but must be close geographically and readily available. Careful coordination and active management throughout this long protocol are required.

Troubleshooting

Problem 1

Low nuclei yield/aggregated nuclei during isolation.

Potential solution

Lower nuclei yield can result from both excessive or insufficient tissue homogenization (step 2) or inadequate size of the sample input. It is imperative to adequately homogenize tissue to release nuclei, while not over-homogenizing such that the nuclei are damaged, which is often detectable as increased clumping of nuclei. Adjustment of the gentleMACS programs to provide the right amount of homogenization may be necessary. Additionally, nuclei loss during filtration (steps 3a and 4a) can decrease yield, underscoring the importance of pre-wetting both the 40 μm and 20 μm filters to ease the passage of nuclei. Lastly, smaller samples may yield small or sometimes non-visible nuclei pellets. When removing supernatant after spin-down in a swinging-bucket centrifuge (steps 4c, 5b, and 8c), special attention should be given to avoid disturbing the pellet or inadvertently aspirating nuclei from the tube’s bottom when pellet is not visible.

Problem 2

Clog during FANS.

Potential solution

The presence of excessive cell debris and clumped nuclei in your preparation can lead to tubing and nozzle blockages in the FACS instrument during nuclei sorting/pooling (step 15). To prevent this issue, you can perform additional filtration using a 20 μm filter after the initial centrifugation (step 4c). This involves resuspending the pellet with 500 μL of NRB and rinsing the filter with 2 mL of NRB. If a clog occurs and there is no alternative FACS instrument with a 70 μm nozzle available, you can use a larger nozzle size to sort the target number of nuclei. However, this will require the nuclei suspension to be centrifuged and resuspended in 43.2 μL of nuclease-free water before adding the master mix ((https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-single-cell-3-reagent-kits-user-guide-v-3-1-chemistry; step 1.2b). It is important to note that this may result in lower nuclei recovery and potentially lower data quality.

Problem 3

Run out of sample before reaching the target number of sorted cells.

Potential solution

While it is preferable to maintain a balanced representation of hashtag antibodies by sorting an equal number of nuclei per sample, low sample input or issues in sample preparation preclude the possibility. There are two options: either proceed to 10× GEM generation without sorting the full 52,000 nuclei or sort additional nuclei from the rest of the samples to make up for the deficit. To best maintain consistent final reagent concentrations, we recommend sorting additional nuclei to make up the deficit.

Problem 4

Shorter than expected average size of endogenous transcript cDNA Bioanalyzer profile.

Potential solution

This issue is likely caused by RNA degradation at some stage of the protocol. The highest RNase activity occurs during tissue sample collection (before you begin steps 1 and 6). It is crucial to minimize the time taken to freeze the sample and to maintain an RNase-free environment as possible. For fresh tissue, prompt transfer to ice-cold PBS is recommended. For nuclei isolation, we use buffers containing 0.2 U/μL of RNase inhibitor, a concentration we have found sufficient to preserve RNA integrity. While it is possible to increase the concentration of RNase inhibitors in the buffers, their cost is significant, and we did not observe significant differences between using 0.2 U/μL and 1 U/μL of RNase inhibitors.

Problem 5

Low number of estimated cells after Cell Ranger analysis.

Potential solution

This issue can be caused by inclusion of poor-quality nuclei during FANS or loss of nuclei post FANS. Inclusion of poorer-quality nuclei may be detected as a higher-than-normal event rate, and gating thresholds should be checked and adjusted. To prevent loss of nuclei after FANS, minimizing pipetting and the use of low retention tips is recommended. Loss may also result from imperfect registration of the FANS output nozzle to the collection tube resulting in nuclei completely missing the collection tube or sticking to its side before they reach the collection buffer. We check plate registration prior to sorting by placing an adhesive film on the plate with collection buffer and positioning drop in the center of the designated collection well.

Problem 6

Poor hashtag antibody performance.

Potential solution

We assess the performance of hashtag antibodies bioinformatically using demuxEM software to determine the determination and counting of cell and hashtag barcodes and assignment to specific sample based on signal to noise ratios of each of the hashtags associated with a given cell barcode. Baked into this algorithm, poorly performing antibodies will show poor discriminatory power.

When poor hashtag performance is observed, we recommend: 1) To minimize cross-labeling, we incorporate a post-incubation spin-down (step 8c) to reduced unbound antibody carryover. It is a delicate balance to remove as much supernatant as possible without disturbing or inadvertently disturbing the nuclear pellet. Addition of dilutional washes using ST buffer + RNase Inhibitor can help further reduce carryover. 2) Attempt to reduce the FANS sort time to minimize time in which nuclei pooled from different samples release antibody that can rebind to other nuclei. 3) Ensure that preparations after filtering have minimal debris. We have found that both excess debris seems to act as a blocking agent, reducing initial antibody binding. This is especially true in white matter parts of the brain and spinal cord that contain myelin.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Linus Tsai (ltsai@bidmc.harvard.edu).

Technical contact

Questions about the technical specifics of performing the protocol should be directed to and will be answered by the technical contact, Linus Tsai (ltsai@bidmc.harvard.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

This study did not generate new or unique datasets or code.