oming challenges of single-cell and spatial genomics experi
https://sail.mskcc.org/
Ronan Chaligne
[email protected]
11/02/2023
Single-cell genomics
Single-cell RNA sequencing (scRNA-seq) is probably of the most widely used single-cell sequencing approaches
The RNA […] amplification results in >106-fold amplification
of the original starting material, which is adequate for analysis-
e.g., use as a probe, making of cDNA libraries, etc.
Expression profiling of selected cDNAs.
Southern blot
Single-cell genomics
Single-cell RNA sequencing (scRNA-seq) is probably the most widely used single-cell sequencing approaches
Using our mRNA-Seq assay with only a single mouse blastomere,
we detected the expression of 75% (5,270) more genes than
microarray techniques and identified 1,753 previously unknown
splice junctions called by at least 5 reads.
Finally, for Dicer1/ and Ago2/ (Eif2c2/) oocytes, we found that
1,696 and 1,553 genes, respectively, were abnormally
upregulated compared to wild-type controls, with 619 genes in
common.
SAIL mission
- Benchmark new method / equipment
- Develop new assay / analysis pipeline
- Optimize existing protocols / analysis pipeline
- Guide / advice through your single-cell / spatial genomics project
Experimental Computational Biology
Ignas Brigita Roshan Andrew
Catherine Jo
Masilionis Meškauskaitė Sharma Moorman
Snopkowski Adams
Maria
Nikhita Meril Deena Tobias Rahul
Bikou
Pasnuri Takizawa Shefter Krause Saha
Single-cell multi-omics
From Nam, A.S., et al. Integrating genetic and non-genetic determinants of cancer evolution by single-cell multi-omics. Nat Rev Genet 22, 3–18 (2021).
Stand alone single-cell / nuclei RNA
Single Cell 3ʹ Reagent Kits v3.1
Single-cell / nuclei RNA
We need to have:
-Full length mRNA
-Efficient reverse transcription with template switch
-Intact cellular plasmic membrane
-As low as possible RNAse activity
Single-cell / nuclei RNA
Single Cell Isolation / Tissue preparation
RNAse activity in healthy mouse tissue
Fold increase
relative to brain
From https://www.thermofisher.com/
Single-cell nuclei ATAC
We need to have:
-Native chromatin structure
-Efficient tagmentation of gDNA
-No tagmentation of the mtDNA
“Multiome” by 10XG -> snRNA + snATAC
Single-cell nuclei RNA + ATAC
scATAC scRNA
To consider for single-cell genomics
Biological questions
Good / efficient logistics
Optimized single-cell Isolation / tissue preparation
To consider for single-cell genomics
Single Cell Isolation / Tissue preparation
A single-cell and single-nucleus RNA-Seq toolbox for fresh and frozen human tumors.
Slyper et al. Nat Med 26, 792–802 (2020).
scRNA-Seq and snRNA-Seq typically recovered similar cell types at varying proportions
scRNA-Seq: More immune cells
snRNA-Seq: More parenchymal (especially malignant) cells
Dissociation signatures is more present in scRNAseq
But …but present in nuclei as well > immune cells and stroma cells.
Likely a signature of immune activation.
See also:
Van Den Brink et al. Nat Methods 2017
Machado et al. Cell Stem Cell 2021
Sample preparation logistic
Good logistic -> Better data
Sample preparation logistic
Pros Cons
Fresh suspension (Fresh tissue → Single cell suspension)
scRNA-seq quality Challenging logistics
Tumor collection in OR Dissociation & FACS sorting Encapsulation for single-cell seq
~ ½ hour ~ 3 to 5 hours ~ 1 hour CITE-seq amenable Enzymatic dissociation
Cell enrichment
Sample preparation logistic
Pros Cons
Fresh suspension (Fresh tissue → Single cell suspension)
scRNA-seq quality Challenging logistics
Tumor collection in OR Dissociation & FACS sorting Encapsulation for single-cell seq
~ ½ hour ~ 3 to 5 hours ~ 1 hour CITE-seq amenable Enzymatic dissociation
Cell enrichment
Fresh-frozen suspension (Fresh tissue → Single cell suspension → -160°C) scRNA-seq quality Survival after thawing
Tumor collection in OR Dissociation -160°C FACS sorting Encapsulation for single-cell seq CITE-seq amenable Enzymatic dissociation
~ ½ hour ~ 1 hour storage ~ 2 hour ~ 1 hour
Cell enrichment
Sample preparation logistic
Pros Cons
Fresh suspension (Fresh tissue → Single cell suspension)
scRNA-seq quality Challenging logistics
Tumor collection in OR Dissociation & FACS sorting Encapsulation for single-cell seq
~ ½ hour ~ 3 to 5 hours ~ 1 hour CITE-seq amenable Enzymatic dissociation
Cell enrichment
Fresh-frozen suspension (Fresh tissue → Single cell suspension → -160°C) scRNA-seq quality Survival after thawing
Tumor collection in OR Dissociation -160°C FACS sorting Encapsulation for single-cell seq CITE-seq amenable Enzymatic dissociation
~ ½ hour ~ 1 hour storage ~ 2 hour ~ 1 hour
Cell enrichment
Fresh-frozen tissue (Froze tissue → -160°C → Single nuclei suspension)
Tumor collection in OR -160°C Nucleus preparation Encapsulation for single-nucleus seq Easier logistics No cell enrichment
~ ½ hour storage ~ 1 hour ~ 1 hour Non bias dissociation No CITE-seq
RNA from nucleus
Benchmarking sample preparation
-Comparison of dead cell removal methods for single-cell RNA-seq
-Impact of nucleus isolation methods on single-nucleus RNA-seq
Benchmarking sample preparation
-Comparison of dead cell removal methods for single-cell RNA-seq
-Impact of nucleus isolation methods on single-nucleus RNA-seq
Comparison of dead cell removal methods for single-cell RNA-seq
Flow cytometry (100 uM nozzle)
DAPI / 7AAD negative
LeviCell EOS, LevitasBio
Physical, based on integrity of cell membrane
Dead Cell Removal Microbubble Kit, Akadeum
Molecular, microbubbles are coated with Anexin V, negative selection
EasySep Dead Cell Removal Kit, Stem Cells
Molecular, immunomagnetic negative selection using Anexin V
Dead Cell + Debris Removal Kit, Miltenyi
Molecular, immunomagnetic negative selection using Anexin V
Comparison of dead cell removal methods for single-cell RNA-seq
Lymph Node Cerebrospinal fluid
No or easy dissociation to single cell
Always processed as single-cell
Contain fragile sub-population
Comparison of dead cell removal methods for single-cell RNA-seq
Lymph Node Cerebrospinal fluid
No or easy dissociation to single cell
Always processed as single-cell
Contain fragile sub-population
LN Viability Total yield CSF Viability Total yield
FACS 92 % 71 % FACS 94 % 60 %
Levicell 98 % 73 % Levicell 87 % 84 %
Akadeum 94 % 90 % Akadeum 88 % 93 %
Stemcell 98 % 76 % Stemcell 90 % 86 %
Miltenyi 98 % 73 % Miltenyi 91 % 62 %
Collaboration with Chrysothemis Brown lab Collaboration with Adrienne Boire lab
Comparison of dead cell removal methods for single-cell RNA-seq
Lymph Node
Cerebrospinal fluid
Cells = Filtered cells using a 0.01 p-value cutoff
Avg. Genes = Avg. Number of genes with non-zero counts
All methods show good quality data Cell Fraction = Number Barcodes / Labelled Cells
Read Fraction = Reads in labelled Cells / Total Read
Comparison of dead cell removal methods for single-cell RNA-seq
CSF
UMAP 2
UMAP 1
Comparison of dead cell removal methods for single-cell RNA-seq
CSF
UMAP 2
UMAP 1
Comparison of dead cell removal methods for single-cell RNA-seq
LN
UMAP 2
UMAP 1
Comparison of dead cell removal methods for single-cell RNA-seq
LN
UMAP 2
UMAP 1
Comparison of dead cell removal methods for single-cell RNA-seq
• FACS is harsher than any other method
• However, FACS has clearest distinction between empty and non-empty drops i.e. less ambie
• LeviCell seems to captures fragile cells best
• However, LeviCell also maintains most “apoptotic" cell (probably not a bad things)
Benchmarking sample preparation
-Comparison of dead cell removal methods for single-cell RNA-seq
-Impact of nucleus isolation methods on single-nucleus RNA-seq
Impact of nucleus isolation methods on single-nucleus RNA-seq
Impact of nucleus isolation methods on single-nucleus RNA-seq
Impact of nucleus isolation methods on single-nucleus RNA-seq
Cell type proportions in GBM
Collaboration with Kenny Yu
Impact of nucleus isolation methods on single-nucleus RNA-seq
Cell type proportions in CRC
Collaboration with Karuna Ganesh Lab
Impact of nucleus isolation methods on single-nucleus RNA-seq
Cell type proportions in mouse liver
Collaboration with Scott Lowe Lab
Impact of nucleus isolation methods on single-nucleus RNA-seq
●Singulator and Miltenyi* performed consistently well (* no data for CRC)
●There were some variation in cell types and proportions recovered between methods
●There are batch effects between methods
●Automated methods excel at parallelization
scRNAseq on fixed samples
snRNAseq from FFPE
Collaboration with S2 Genomics
GBM - Kenny Yu
Tabar lab
UMAP2
UMAP1
Also to consider for single-cell genomics
Capture rate:
~65% for 10X (vs ~25% for single cell sorted in plate)
Number of cells to target:
How big is your population of interest? (0.1%, 1%, 10%…)
Number of samples:
Model organism (go for multiplexing) vs Clinical samples (always more)
Sample QC:
Cell Viability
Cell concentration
Cell doublets / Clumps
Buffer composition (Avoid FBS when possible)
Getting more from Single-cell
“Rescuing” signal from scRNAseq
Enrich and detect mitochondrial DNA variants to
establish lineage relationships in primary human cells
Using scATAC-seq dataset
Using scRNA-seq dataset
From Ludwig et al., Cell 2019
Multiome on whole cells
DOGMA-seq
Scalable, multimodal profiling of chromatin accessibility, gene expression and protein levels in single cells.
Mimitou, E.P. et al. Nat Biotechnol 39, 1246–1258 (2021). https://doi.org/10.1038/s41587-021-00927-2
Spatial transcriptomics - In Situ Hybridization
Xenium by 10X
Xenium by 10X
Collaboration with Jan Remsik and Adrienne Boire Lab
- 2 Xenium slides
- 4 FFPE blocks with 3 brain slices each
- 60,000-120,000 cells captured per tissue
Xenium by 10X
Xenium by 10X
Xenium to Vizium (with CytAssist)
Vizium on FFPE
Complementary approaches
Vizium (+/- FFPE) Xenium (+/- FFPE)
Fixed scRNAseq
Genotyping with custom probes using probe-based scRNA
Collaboration with Raajit Rampal Meril Takizawa, SAIL
JAK2 V617F SNV detection
Sensitivity: 41.6%
False positive: 0.20% Approach can be expanded to Vizium and Xenium
Collaboration with Lowe lab
Is In Situ Hybridization ground truth?
We have open positions
Acknowledgements
Collaborators at MSK
Roshan Sharma
Andrew Moorman Adrienne Boire Lab
Ignas Masilionis Tobias Krause Karuna Ganesh Lab
Brigita Meškauskaitė Deena Shefter Christine Iacobuzio Lab
Catherine Snopkowski Jo Adams Scott Lowe Lab
Meril Takizawa Rahul Saha Kenny Yu Lab
Nikhita Pasnuri Tal Nawy Chrysothemis Brown Lab
Joan Massagué Lab
Dana Pe’er Lab
All the clinicians
All the patients and their families
The Alan and Sandra Gerry
Metastasis and Tumor Ecosystems Center
