Validation of the Agilent BioCel Automated Platform for Single Cell - - PowerPoint PPT Presentation

Validation of the Agilent BioCel Automated Platform for Single Cell Genomic Analysis

Mark Novotny, Project Lead Joyclyn L. Yee-Greenbaum Jeffrey S. McLean Shino Ishii Mary-Jane Lombardo Todd Hughes Robert Vandenberg Roger S. Lasken (PI)

Single Cell Genomics

• What is it?

The study of a genomic sequence obtained from a single cell

• Why is it useful?

Most bacterial cells are uncultivable

• What benefits can be achieved by automation?

High-throughput access to numerous unknown genomes

Mitch Leslie Science 7 January 2011:

• Vol. 331 no. 6013 pp. 24-26

News and Views Nature Biotechnology 24, 657 - 658 (2006). Single-cell genomics Clyde A. Hutchison, III1 & J. Craig Venter1 Zhang, K. et al. Nat. Biotechnol. 24, 681–687 (2006). Raghunathan, A. et al. Appl. Environ. Microbiol. 71, 3342–3347 (2005). Penny Chisholm (Broad), Ramunas Stepanauskas (Bigelow Lab), Tonya Wyoke (JGI)

• Single cells labeled with

16S rRNA gene probes

• Isolation by micromanipulation
• DNA amplified by MDA

Isolation of FISH Probed Bacterial Cells by Micromanipulation

Lasken R, Raghunathan A, Kvist T, Ishøey T, Westermann P, Ahring B, Boissy R In Whole Genome Amplification: Methods Express Edited by Hughes S., Lasken, R. Oxford: Scion Publishing Ltd. 2005

DNA Amplification by Multiple displacement amplification (MDA) Genotype and Sequence

Funded by DOE, 2001, at Molecular Staging, Inc.

Genomic DNA

>109 fold amplification Raghunathan, A., Ferguson, H.R., Bornarth, C.J., Driscoll, M., and Lasken, R.S. Applied and Environmental Microbiology (2005) Vol. 71, 3342-3347

Noise B.subtilis Noise E.coli Noise B.subtilis Noise B.subtilis Noise E.coli Noise E.coli

Flow cytometry

Single Cell Sequencing

Genomic DNA

Multiple Displacement Amplification (MDA)

• 1. Random hexamer primers
• 2. Phi29 DNA polymerase

Strand displacing

• 3. Isothermal reaction (30 C)

Dean FB, Nelson JR, Giesler TL, Lasken RS (2001) Genome Res. 11, 1095-9 Dean FB, Hosono S, Fang L, et al. (2002). Proc Natl Acad Sci USA;99:5261-6

12.0 kb Alkaline agarose

TempliPhi, GenomiPhi (GE Healthcare/Amersham) REPLI-g (Qiagen)

gDNA

600x, kmer=55

Great MDA 98% genome in contigs top contig 67k N50 68k

Single Cell MDAs Vary in Quality

Illumina Sequencing Redundancy Poor MDA top contig 14k N50 45

Single Cell Genomics Pipeline

Discovery platform for producing sequenced genomes from single cells

11 Major Steps in the Process

1. Single cell Fluorescence Activated Cell Sorting (FACS flow cytometry) 2. MDA 3. 16S or 18S PCR or qPCR 4. PCR analysis: melt curve assay 5. PCR hit picking 6. PCR SAP/Exo cleanup 7. Sanger sequencing 8. 16S or 18S characterization 9. MDA hit picking and re-amplification

• 10. Whole genome sequencing
• 11. MDA archival

High Throughput Fluorescence Activated Cell Sorting on BD FACS Aria II

T4 phage Shewanella sp. Sybr Green fluorescence Forward scatter

• 488nm 100mW laser
• FSC-PMT detector
• SSC PMT detector
• 3 color detection (2 Green, 1 Red)
• One 384-well plate with 384 single cell in 8-10mins
• Capable of detecting up to 70,000 events/sec
• >97% accuracy of 1 bead/well
• Sort cells into 2uL of TE (10:0.1)

Marine

• rganisms

Stained Unstained

FACS: Scripps Pier Seawater 30µm Filtered DNA Stained with SYBR Green I

Marine Single Cell Genomics

2 papers accepted and 1 more planned

Genome SAR86-2, 50% SAR86-1, 50% SAR86-3, 10% Uncultured, highly abundant SAR86 clade SAR86-4, >95% in closure SAR86 single cell genomes

MDA shallow sequencing deep sequencing

Single cell

de novo assembly annotation (closure)

Marine Single Cell Genomics

Uncultured SAR324 clade

MDA shallow sequencing deep sequencing

Single cell

de novo assembly annotation (closure)

SAR324-2, in assembly SAR324-1, 75% SAR324 single cell genomes SAR324-3, in assembly SAR324-4, in assembly Genome

Nycodenz

• r

Filtration

Oral cavity sample Cell fraction

EtOH fixation Staining Sorting

Single Cell Amplified Genomes From Human Oral Microbiome

B A

Stained Unstained

Automation Design Challenges

• 5,000 Single Cells processed per week
• <6 months to complete 100,000 cell screen
• 1 and only 1 cell per well
• 384-well format
• Integrated automation with 3rd party platforms
• 4C incubation and reaction temperature control
• Sophisticated timing mechanisms to meet assay

time restraints

• Isolate cells from complex mixtures
• Air (water impaction)
• Soil (Nycodenz)
• Seawater (filtering)
• Human gut (Nycodenz)
• Oral biome (Nycodenz/filtration)
• Non-contact pipetting of reagents for amplification
• Minimal waste from expensive reagents
• Filtered tips for cross-contamination control
• Adequate mixing of reagents during reaction assembly
• Pipetting of viscous solutions
• Contamination control
• Free DNA
• Well-to-well cross-contamination
• Amplicons-MDA, PCR, Sanger cycle sequencing
• Highest quality reagents
• Lot testing and validation
• Sealed amplification reactions
• Centrifugation during reaction assembly
• Turn-key operation with minimal staffing
• Icon-driven software with minimal training and skill required

Key Requirements

Etc., etc., etc.

Agilent Design: BioCel Single Cell Genomics

Process >14x384 well plates/week >5,000 single cells/week

BioRAPTR FRD Liquid Handler ABI 7900HT qPCR Direct Drive Robotic Arm Bravo Liquid Handler Liconic Chilled Incubator PlateLoc Plate Sealer Vspin Centrifuge Plate Hub Barcode Labeler Vworks Workstation PC Mecour Plate Tower

Agilent Design: BioCel Single Cell Genomics

BioCel Subsystems: BioCel 1200 and 2 Device Tables

~7’ L x 7’ W x 7’ H

BioCel Subsystem: Agilent DDR

• Fast
• Accurate
• Precise
• Compact
• Easy to program
• Efficient movement
• Large range of movement
• Portrait and landscape

BioCel Subsystem: Agilent Bravo

• Fast
• Accurate (750nL-50uL)
• Precise <5% CV
• Easy to program
• Efficient movement
• 384-tip pipeting
• Disposable Filter tips (no washing)
• Low retention in tips
• Hit picking
• Chilled platform
• Gripper
• Weigh station (reservoirs)
• Easy to clean/decontaminate

BioCel Subsystem: Beckman BioRAPTR

• Fast (30sec -1min/plate)
• Accurate (150nL-60uL)
• Precise <5% CV
• Non-contact pipeting
• Integration with BioCel
• Easy to program
• Controllable through Agilent VWorks
• Efficient movement
• Random access to wells
• 8-Tip dispenser head
• 8 Sealed Reagent Reservoirs
• 8 Removable Chillers (Cooling Nest)
• Volumes compatible (150nL-50uL)
• Low vol. retention in tips and lines
• Easy to clean/decontaminate

BioCel Subsystem: ABI 7900HT

• qPCR Taqman chemistries
• Syto9 melt curve capability
• Sybr Green qPCR
• 96-and 384-well formats
• Automation capable

(but not friendly)

• Custom compatible

with Vworks

BioCel Subsystems: Incubators and Temperature Controls

• BioRAPTR Cooling Nest
• Liconic and Liconic plate trays
• Polyscience chiller
• Chills Mercour tower
• Chills Cooling Nest
• Thermocube chiller
• Chills Bravo deck 3-plate platform

BioCel Subsystems: Vspin, Vcode, and PlateLoc

• Vspin
• Meets g requirements
• Fast
• Counterbalanced
• PlateLoc plate sealer
• Compatible with all plates
• Seal works with ABI 7900HT
• Warms up fast
• Vcode barcode labeler
• JLIMS compatible
• Thermostable @ -80C

BioCel Subsystems: Plate Hub, Barcode Reader, Lid Tower, Lid Tool

• Plate Hub
• Capacity for 14 plate runs
• Holds tip boxes
• Holds Framestar and Greiner plates
• Barcode Reader
• Vworks compatible
• Lid Tower
• Holds lids for 14 plate runs
• High performance and precision
• Low error rate
• Compatible with Greiner low profile lid
• Lid Tool

ABI 9700 Thermocyclers

• Dual 384-well ABI 9700’s
• 12 total in the lab
• 24 plate capacity
• MDAs, PCRs, Lysis

BioCel Software: VWorks

• Controls all major components and subsystems
• Incredibly sophisticated event-driven software
• Efficient task control with error recovery
• Icon-driven programming
• Ease of re-programming
• Tons of features

BioCel Software: JLIMS

Register Sample

Register Samples (either pre-barcoded or requiring new barcoding) with the sytem Generate plate barcodes

Cell Sorting

Prepare the MDA plate

MDA Hit Plate

Create the MDA Hit Plate

Generate Plate Barcodes MDA Prep MDA Dilution 16s PCR Melt Curve Analysis 16s Analysis

Create the hit plate based on Melt Curve scores Run 16s Analysis pipeline

• Custom JCVI LIMS Software
• Web driven
• Designed for Sanger Pipeline at JTC
• Modified for Single Cell Pipeline
• Currently evaluating

(not implemented)

SCGP-Design Checklist

• 5,000 Single Cells processed per week
• <6 months to complete 100,000 cell screen
• 1 cell and 1 cell only
• 384-well format
• Integrated automation with 3rd party platforms
• 4C incubation and reaction temperature control
• Sophisticated timing mechanisms to meet assay

time restraints

• Isolate cells from complex mixtures
• Air (water impaction)
• Soil (Nycodenz)
• Seawater (filtering)
• Human gut (Nycodenz)
• Oral biome (Nycodenz/filtration)

       

• Non-contact pipetting of reagents for amplification
• Minimal waste from expensive reagents
• Filtered Tips for cross-contamination control
• Adequate mixing of reagents during reaction assembly
• Pipetting of viscous solutions
• Contamination control
• Free DNA (although cannot always be ruled out)
• Well-to-well cross-contamination
• Amplicons-MDA, PCR, Sanger cycle sequencing
• Highest quality reagents
• Lot testing and validation
• Sealed amplification reactions
• Centrifugation during reaction assembly
• Turn-key operation with minimal staffing
• Icon-driven software with minimal training and skill

required

                  Etc., etc., etc.

SCGP Process Validation

1. Single cell Fluorescence Activated Cell Sorting (FACS flow cytometry) a) FrameStar 2uL Plate Setup 2. MDA 3. 16S or 18S PCR or qPCR 4. PCR analysis: melt curve assay 5. PCR hit picking 6. PCR SAP/Exo cleanup 7. Sanger sequencing 8. 16S or 18S characterization 9. MDA hit picking and re-amplification

• 10. Whole genome sequencing
• 11. MDA archival



BioCel Validation: Step 1a: Framestar Plate Setup

• Created routine to fill 64 x 384-well plates with 2uL TE using 1 box of tips
• Pipettes TE from a trough on the Bravo deck
• Eliminates the need for BioRAPTR to fill empty plates
• Poses a risk to contaminating FACS sorting plates with gDNA on surfaces or amplicons

1. Single cell Fluorescence Activated Cell Sorting (FACS flow cytometry) a) FrameStar 2uL Plate Setup

 

BioCel Validation-BioRAPTR Tartrazine Assay

• Calibration: Curve Fit, Automated
• All 8 tips tested in triplicate 384-well plates
• Tartrazine Assay readout on Flexstation 3 fluorimeter
• Tartrazine Assay results: 5.5% CV or less @ 1,2,7, and 12uL
• Accurate 150nL to 12uL range tested
• Random Access Pipetting 12uL = good

Tartrazine+H2O Average STDEV Min Max CV Date water blank 47uL 0.111 0.005 0.102 0.125 4.190 11/30/2010 7uL+43uL 2.038 0.061 1.829 2.279 2.981 11/30/2010 12uL+38uL 3.459 0.056 3.296 3.824 1.610 11/30/2010 2uL+48uL 0.640 0.035 0.524 0.719 5.479 11/30/2010 2uL+48uL_1_new 0.644 0.027 0.518 0.709 4.250 11/30/2010 2uL+48uL_2_new 0.649 0.030 0.520 0.736 4.646 11/30/2010 2uL+48uL_3 0.658 0.022 0.571 0.731 3.271 11/30/2010 2uL+48uL_4 0.650 0.026 0.546 0.791 3.995 11/30/2010 2uL+48uL_5 0.656 0.021 0.561 0.740 3.270 11/30/2010 1uL+49uL_1_new 0.360 0.017 0.259 0.399 4.765 11/30/2010 1uL+49uL_2_new 0.369 0.017 0.292 0.423 4.549 11/30/2010 1uL+49uL_3_new 0.365 0.019 0.300 0.415 5.131 11/30/2010 1uL+49uL_4_new 0.370 0.014 0.317 0.419 3.872 11/30/2010 1uL+49uL_5 0.374 0.014 0.319 0.418 3.696 11/30/2010 2uL+48uL_6 0.592 0.021 0.522 0.657 3.469 12/1/2010 2uL+48uL_7 0.600 0.019 0.535 0.667 3.100 12/1/2010 2uL+48uL_8 0.601 0.025 0.507 0.681 4.232 12/1/2010 water blank 0.105 0.004 0.096 0.125 4.234 12/1/2010 2uL+48uL_Tip2_1 0.624 0.020 0.532 0.681 3.264 12/1/2010 2uL+48uL_Tip2_2 0.617 0.020 0.495 0.678 3.162 12/1/2010 2uL+48uL_Tip2_3 0.619 0.017 0.548 0.689 2.802 12/1/2010 2uL+48uL_Tip3_1 0.646 0.018 0.577 0.705 2.804 12/1/2010 2uL+48uL_Tip3_2 0.643 0.015 0.583 0.698 2.376 12/1/2010 2uL+48uL_Tip3_3 0.631 0.015 0.570 0.691 2.357 12/1/2010 2uL+48uL_Tip4_1 0.623 0.019 0.531 0.673 2.987 12/1/2010 2uL+48uL_Tip4_2 0.602 0.016 0.550 0.669 2.641 12/1/2010 2uL+48uL_Tip4_3 0.613 0.016 0.536 0.685 2.636 12/1/2010 2uL+48uL_Tip5_1 0.618 0.014 0.550 0.687 2.199 12/1/2010 2uL+48uL_Tip5_2 0.623 0.014 0.568 0.681 2.172 12/1/2010 2uL+48uL_Tip5_3 0.620 0.016 0.569 0.702 2.560 12/1/2010 2uL+48uL_Tip6_1 0.643 0.018 0.555 0.713 2.823 12/1/2010 2uL+48uL_Tip6_2 0.631 0.014 0.576 0.702 2.270 12/1/2010 2uL+48uL_Tip6_3 0.637 0.015 0.590 0.712 2.311 12/1/2010 2uL+48uL_Tip7_1 0.589 0.013 0.543 0.652 2.134 12/1/2010 2uL+48uL_Tip7_2 0.587 0.012 0.548 0.648 2.114 12/1/2010 2uL+48uL_Tip7_3 0.586 0.012 0.554 0.657 2.035 12/1/2010 2uL+48uL_Tip8_1 0.687 0.015 0.652 0.828 2.144 12/1/2010 2uL+48uL_Tip8_2 0.681 0.014 0.545 0.763 2.125 12/1/2010 2uL+48uL_Tip8_3 0.693 0.012 0.656 0.757 1.753 12/1/2010

EpMotion MDA dilution and 16S PCR Bravo dilution and 16S PCR, 1st experiment

Correct Incorrect

BioCel Validation: Bravo Cross-Contamination Experiments

• Checkerboard analysis of Bravo demonstrates improvement over EpMotion
• Viscosity of MDAs needed to be addressed on Bravo

BioCel Validation: Bravo Optimization of Viscous MDA Pipetting

• lderia

• lderia

• idace

• lderia

• lderia

• lderia

• lderia

• lderia

• lderia

• lderia

• lderia

• lderia

• lderia

• nadac

• myc

• nadace

• nadace

• coccace

• nadace

• nadace

Correct Incorrect

16S PCR Replicate 1 16S PCR Replicate 2

BioCel Validation: Reagent QC for Automated PCR

• 384-well checkerboard process reveals contamination
• Green wells are sequences from PCR reagents alone

BioCel Validation-Bravo Viscosity Test and Hit Picking

BioCel Validation-HMP Fecal Sample-1 Plate

Automation tasks to be validated in single 384-well plate run:

 

1. Single cell Fluorescence Activated Cell Sorting (FACS flow cytometry) 2. MDA 3. 16S or 18S PCR or qPCR 4. PCR analysis: melt curve assay 5. PCR hit picking 6. PCR SAP/Exo cleanup 7. Sanger sequencing (outsourced to JCVI-JTC) 8. 16S or 18S characterization 9. MDA hit picking and re-amplification

• 10. Whole genome sequencing (outsourced to JCVI-JTC and others)
• 11. MDA archival

Human fecal sample (from Marty Blaser lab, NYU)

• SybrGreen staining
• FACS single cell sorting

from multiple gates

• BSL2 facility

aerosol containment

Flow Cytometry of Bacteria

Example of Taxonomic Analysis: OTUs

Single cell Clostridiales MDAs fall into several taxonomic groups

Red - 16S fecal sequences (Eckburg) Blue – 16S from sequenced genomes

BioCel Validation-Step 8: 16S Characterization of Fecal Plate

• 119 single cell amplified bacterial genomes obtained
• Confirmation that bacteria were derived from GI tract

100/119 have >99% identity to known 16S rRNA gene sequences (Eckburg fecal 16S library, Science, 2005) Taxonomic diversity observed: Firmicutes (Clostridium sp., Eubacterium sp., Lactobacillus sp.), Bacteroidetes (B. fragilus group, Flavobacteriales), Proteobacteria, Verrucomicrobia

• Confirmation that many genomes are from novel uncultured species

21%: >99% identity to sequenced genomes 57%: 90-99% identity to sequenced genomes 22%: 80-90% identity to sequenced genomes

BioCel Validation-HMP Fecal Sample-1 Plate

1. Single cell Fluorescence Activated Cell Sorting (FACS flow cytometry) 2. MDA 3. 16S or 18S PCR or qPCR 4. PCR analysis: melt curve assay 5. PCR hit picking 6. PCR SAP/Exo cleanup 7. Sanger sequencing (outsourced to JCVI-JTC) 8. 16S or 18S characterization 9. MDA hit picking and re-amplification

• 10. Whole genome sequencing (outsourced to JCVI-JTC and others)
• 11. MDA archival

       

BioCel Validation-Step 9 and 11: MDA Hit Picking, Re-amplification and Archiving

• Hit Pick MDAs from original plate

according to their 16S OTU’s

• Protocol validated during SAT/FAT
• Meets requirements
• Slow, but works
• Not currently validated with real MDAs,

but based on hit pick colormetric data, no problems are anticipated

• Matrix 2D barcoded tubes currently in use

for hand picked MDA archival

BioCel Validation-HMP Fecal Sample-1 Plate

1. Single cell Fluorescence Activated Cell Sorting (FACS flow cytometry) 2. MDA 3. 16S or 18S PCR or qPCR 4. PCR analysis: melt curve assay 5. PCR hit picking 6. PCR SAP/Exo cleanup 7. Sanger sequencing (outsourced to JCVI-JTC) 8. 16S or 18S characterization 9. MDA hit picking and re-amplification

• 10. Whole genome sequencing (outsourced JCVI-JTC and others)
• 11. MDA archival

       

X X

Jonathan Badger JCVI | T

• dd DeSantis LBL | Dirk Gevers Broad | Anthony Fodor UNC | Ashlee Earl Broad

PHYLOGENY-BASED SELECTION OF STRAINS FOR THE HMP REFERENCE GENOME COLLECTION

100 MOST BACTERIA!

HMP: Most Wanted Phyla

HMP: Body Region OTU’s

Oral Microbiome: Oral Cavity OTU’s

BioCel SCGP Multi-Plate Validation: Human Microbiome Oral Samples

Total of 35 plates run on human oral microbiome Full automation scaling from 6, 6, 11, 12 plates on the BioCel SCGP How’d we do?

SCGP Process Wells % # Single cell wells Status as of 8/24/11 Notes Total oral microbiome single cells to be analyzed 12,144 100% 12144 Design Completed 33 single cell plates, 1 partial plate, 1 NTC plate In process on SCGP (La Jolla)

• 0%

Completed Completed SCGP (La Jolla) 12,144 100% 12144 Completed In process 16S Sanger (JTC-Maryland)

• 0%

Completed Completed 16S Sanger (JTC-Maryland) 12,144 100% 12144 Completed BLAST Complete, 16S IDs obtained 12,144 100% 12144 Data Analysis Completed SCGP Oral Microbiome Analysis Results (as of 8/24/10) # Single cell wells Successful Oral Single Cell MDAs 3,231 26.6% 12144 Sequenced Oral Single Cell MDAs* (Oral genomes already sequenced) 2,621 21.6% 12144 Novel** (Oral Microbiome db BLAST) 610 5.0% 12144 Novel-Good*** (Oral Microbiome db BLAST, good chromatogram) 509 4.2% 12144 100 Most Wanted Matched 99% **** 139 1.1% 12144 100 Most Wanted Matched 99% **** and non-redundant 11 0.1% 12144 100 Most Wanted Matched 97% ***** 303 2.5% 12144 100 Most Wanted Matched 97% ***** and non-redundant 16 0.1% 12144 Total Oral Single Cell MDAs already sent to sequencing centers 181 (JTC - Methe) Total Oral Single Cell MDAs ready to be sent to sequencing centers 328 * Sequenced Oral Single Cell MDAs defn: BLAST >97% identity to the sequenced genomes database ** Novel defn: BLAST >97% identity to the oral 16S database and <97% identity to the sequenced genomes database *** Novel-Good defn: BLAST >97% identity to the oral 16S database and <97% identity to the sequenced genomes database and good chromatagram quality **** 100 Most Wanted Matched 99% defn: Defined as "Novel-Good", BLAST >99% identity to the 100 Most Wanted database, >100 bp hit ***** 100 Most Wanted Matched 97% defn: Defined as "Novel-Good", BLAST >97% identity to the 100 Most Wanted database, >100 bp hit Oral Microbiome Single Cell Genomics Pipeline (SCGP) Summary Table as of 8/24/11

BioCel Validation: Summary

Checklist: 8 processes completed, validated, and operational

• 1. Single Cell Fluorescence Activated Cell Sorting (FACS flow cytometry)
• 2. MDA
• 3. 16S or 18S PCR or qPCR
• 4. PCR Analysis: Melt Curve Assay (TBD)
• 5. PCR Hit Picking
• 6. PCR SAP/Exo cleanup
• 7. Sanger Sequencing (outsourced to JCVI-JTC)
• 8. 16S Characterization
• 9. MDA Hit picking and re-amplification

10.Whole genome sequencing (outsourced to JCVI-JTC and others) 11.MDA Archival

       

Acknowledgements

FUNDING This work was supported by the Alfred P. Sloan foundation, grant NIH-2 R01 HG003647 from the National Human Genome Research Institute. Genome Sequencing Centers (GSC) award funded in whole or part with federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services under contract number HHSN272200900007C. HMP project supported by Award Number U54 AI-084844-01 administered by the National Institute of Allergy and Infectious Diseases on behalf of the NIH Roadmap Human Microbiome Project. JCVI Staff Joyclyn L. Yee-Greenbaum Jeffrey S. McLean Shino Ishii Mary-Jane Lombardo Jonathan Badger Dana Busam (JTC) Tamara Ryabtseva (JTC) Indresh Singh Barb Methe Roger S. Lasken (PI) Agilent Staff Robert Vandenberg Todd Hughes Noël Ruppert Steve Lappin Ginger Cooper Bill Rust Russ Berman Denise Williams Barb Nowaczyk Derek Young Hanzel Lawas

BioCel in Action: Video Questions and Discussion