Adapta&on and Design of Panels for the CyTOF December 5, 214 CyTOF User Mee&ng University of Virginia Mike Leipold
Two Different Approaches to Simultaneous Detec&on Fluorophores Elemental Tagging Fluorometer Induc&vely Coupled Plasma Mass Spectrometry (ICP- MS)
Two Different Approaches to Simultaneous Detec&on 8 Alexa Fluorophores Elemental Tagging 1 9 8 7 intensity 6 5 4 3 2 1 38 43 426 449 472 495 518 541 564 587 61 633 wavelength 656 679 72 725 748 771 794 LSR II CyTOF
Overview 1. CyTOF the machine, pros and cons 2. Panel development background, desired signals 3. Experimental procedures an&bodies, reagents 4. Running samples example data - LSRII vs CyTOF - Flu 21, 211
Part 1 The CyTOF
Single Cell Analysis - CyTOF TM Machine Flow Interface to ICP- MS - Nebulizer vaporizes Flow - Argon Plasma atomizes and ionizes - Mass analyzer Figure courtesy O. Ornatsky
Single Cell Analysis - CyTOFTM Machine Published in: Dmitry R. Bandura; Vladimir I. Baranov; Olga I. Ornatsky; Alexei Antonov; Robert Kinach; Xudong Lou; Serguei Pavlov; Sergey Vorobiev; John E. Dick; Scob D. Tanner; Anal. Chem. 29, 81, 6813-6822. DOI: 1.121/ac9149w Copyright 29 American Chemical Society
Display of Cell Events Monocyte? CD8+ T cell? B cell? - Push analogous to &me: 76,8 pushes/sec: 22 displayed here (1/4 of data)
Pros of CyTOF - Advantages: - Minimal "spectral" overlap higher dimensionality (more probes at once) - Quan&ta&ve broad dynamic range - Not light- or &me- sensi&ve - Minimal background - no analogy to autofluorescence: low/nil biological background for lanthanides
Cons of CyTOF - Disadvantages: - Destruc&ve: (currently) no way to recover interes&ng cells - Slower: limit of 1 cells/sec; prac&cal limit for best resolu&on oken in ~4 cells/sec range - Cell transmission efficiency: only ~2-3% of cells that enter machine get counted - Ion transmission efficiency: only ~1 in 1, ions that enter machine get counted
Cons of CyTOF - Disadvantages: - Postprocessing: cannot set on fly gates to get only live intact singlets - Events registered by CyTOF contains debris, doublets, etc - Might only get ~5% of total events as live intact singlets - No analogy to FSC or SSC: MUST have marker (M n+ ) for any ga&ng
Single Cell Analysis - CyTOF TM Machine - CyTOFv1 ~21- May 213 - Mass window ~93 units: eg, AW 13-195 - CyTOFv2 May 213- current - Mass window ~135 units: eg, AW 78-212 - More automated tuning, including recordkeeping - Improvements in ion op&cs - &ghter peaks, less "spillover"
Part 2 Panel Design
Relevant Issues to Panel Design 1. Background any non- desired signal - contamina&ng signals - an&body &ter (not discussed) 2. Desired Signal - an&body clone - metal label analogous to fluorophore choice
Relevant Issues to Panel Design Background Signals * Desired signal only at Mass "M" 1. M+16 oxide cannot eliminate, can only limit 2. Metal salt impuri&es M+1, M- 1, etc; Ln 3. Environmental contamina&on sample 4. Instrument * All poten&al sources of background, even spillover!
Relevant Issues to Panel Design Background Signals * Desired signal only at Mass "M" 1. M+16 oxide cannot eliminate, can only limit - Proper daily machine tuning - Lower AW Ln = worse
Daily Tuning Liquid Tuning Solu&on Current profile Make- up Gas profile
DVS Elemental Beads "Cells" - Polystyrene beads: contain nat. abund. Eu; or La, Pr, Tb, Tm, Lu - Act as "cells" run in acquisi&on mode (vs. tuning/solu&on mode) 1 4 (La139)Dd: (La139)Dd 1 3 1 2 1 1 Makeup Gas Flow Rates - Median Dual counts vs AW 2 MG =.74 15 1 1 1 2 1 3 1 4 (Pr141)Dd: (Pr141)Dd Median Dual Count 1 (Tb159)Dd: (Tb159)Dd 1 4 1 3 1 2 1 1 5 14 15 16 17 18 Atomic Mass 1 1 1 2 1 3 1 4 (Tm169)Dd: (Tm169)Dd
Effect of Tuning Beads - Oxida&on Linear Scale Median Dual counts Log Scale Median Dual counts Median Dual Count 2 15 1 La139 Pr141 Tb159 Tm169 Lu175 "Gd155" "Gd157" "Re185" Median Dual Count 1 1 1 1 1 La139 Pr141 Tb159 Tm169 Lu175 "Gd155" "Gd157" "Re185" 5.1.5.6.7.8.9 1. * Makeup Gas Flow Rate (L/min).1.5.6.7.8.9 1. * Makeup Gas Flow Rate (L/min) - At high MG flow rate, oxida&on can be significant (decreases M, increases M+16 spillover) - Even a few percent can be significant for high- abundance markers (eg, CD57, histone proteins, CD45, etc)
Relevant Issues to Panel Design Background Signals * Desired signal only at Mass "M" 2. Metal salt impuri&es a) Few lanthanides are naturally 1% monoisotopic - e.g., Nd144 signal in Nd145 salt - M+1, M- 1, etc b) Other lanthanide contamina&ons all Ln chemistry is similar - La139 most common contaminant
Purity of Isotopes Liquid Stock Solu&ons CyTOF Channel Single- Isotope Sample - - M+1 impuri&es some isotopes worse than others M+16 oxida&on impuri&es affects all isotopes; lower AW worst
BD Ig Κ Capture Beads and MAXPAR An&bodies M- 1 146 M M+1 Sm147 148 4 15 5 # Cells 3 2 1 # Cells 1 5 # Cells 4 3 2 1 1 1 1 2 1 3 1 4 IgD(Nd146)Dd: IgD 1 1 1 2 1 3 1 4 CD85j(Sm147)Dd: CD85j 1 1 1 2 1 3 1 4 Nd148(Nd148)Dd: Nd148 No "Sm146" - 146 is Nd * Spillover due to elemental and/or isotopic impuri&es, not strictly mass proximity
Major Contaminants/Spillovers: Percentage of M - M+1 and M+16 are usually the major spillovers - Only Fluidigm metals shown; non- Fluidigm metals can be worse * Ir191 contam?
Major Contaminants/Spillovers: Cell Data * Formerly, Fluidigm sold Dy162 and Dy164, but not Dy163 Dy purity issues Regular Phenotyping panel: CD45RA- Dy162 CD2- Dy164 * no Dy163 an&body in experiment, so not usually monitored
Major Contaminants/Spillovers: Cell Data CD45RA- Dy162 CD2- Dy164 * no Dy163 an&body in experiment! Magenta: both Dy162, Dy164 Blue: Dy162 only Red: Dy164 only 1 CD3+ 1 B cells % of Max 8 6 4 - no CD2 % of Max 8 6 4 - high CD2, high CD45RA 2 2 1 1 1 2 1 3 1 4 (Dy163)Dd 1 1 1 2 1 3 1 4 (Dy163)Dd
Relevant Issues to Panel Design Background Signals 3. Environmental contamina&on - Reagents PFA lots with micropar&cles, MeOH with free metal, Ar gas with Sn - Biological sourcing Iodine (free I vs IdU), Pt in cispla&n- treated donor samples - Lab dust - Barium, lead, etc dish soap, syringes, reagent sourcing, striker flint, etc
Relevant Issues to Panel Design Background Signals 4. Instrument Abundance Sensi&vity lek (M- 1) and right (M+1) leg of ion peak - CyTOFv1: - 1% - CyTOFv2: <.3% 4 # Cells 3 2 1 - Worse with high AW masses: peak less Gaussian, M+1 leg wider 1 1 1 2 1 3 1 4 (Sm147)Dd: CD85j
Major Contaminants/Spillovers: Percentage of M * CyTOF spillover - s&ll much less than Fluorescence spillover! * All propor&onal to signal at Mass "M"... 1. Match marker abundance with metal brightness - 1% of 3 is 3 (background) - 1% of 3 is 3 (probably not background) 2. Mutually exclusive lineages - T cell marker followed by B cell marker 3. Dump channels for most contaminated isotopes - analyze the "nega&ve" cells
Desired Signal Intensity- Choice of Metal Reminder: you cannot detect anything that doesn't have bound metal - No scaber to gate monocytes from lymphocytes, etc - Only about 3- fold difference in signal across lanthanides - Switching lanthanides might help only in borderline cases
Desired Signal Intensity- Choice of Metal 1 4 Lower sensitivity (139-15) # Cells 15 1 5 CD56(Yb174)Dd: CD56 1 3 1 2 1 1 1 1 1 2 1 3 1 4 CD3(Nd15)Dd: CD3 1 1 1 2 1 3 1 4 CD16(Sm149)Dd: CD16 2 1 4 Higher sensitivity (165-17) # Cells 15 1 5 CD56(Yb174)Dd 1 3 1 2 1 1 1 1 1 2 1 3 1 4 CD3(Tm169)Dd 1 1 1 2 1 3 1 4 CD16(Ho165)Dd
Desired Signal Intensity- Choice of Metal 1. "Dim" metals In, La, Pr, Nd, Sm - Bright/abundant markers (eg, CD57) - True bimodal (eg, CD27) 3 CD8+ 8 CD4+ 6 # Cells 2 # Cells 4 1 2 1 1 1 2 1 3 1 4 (In113)Dd: CD57 1 1 1 2 1 3 1 4 CD27(Sm152)Dd: CD27
Desired Signal Intensity- Choice of Metal 2. "Bright" metals Tb, Dy, Tm, Yb - Dim/rare markers - "Smear" distribu&on (eg, CD45RA, CCR7) - Small fold- change 1 3 1 2 1 1 TCRgd(Yb171)Dd: TCRgd14 TCRgd 3.29 # Cells 2 15 1 5 CD8+ 1 1 1 2 1 3 1 4 CD3(Nd15)Dd: CD3 1 1 1 2 1 3 1 4 CD45RA(Dy162)Dd: CD45RA
Desired Signal Intensity- Choice of Metal 3. Recommend bivariates - one bright, one dim - two mediums "smear", or subpopula&ons CD2(Dy164)Dd: CD2 1 4 1 3 1 2 1 1 B cells 48.2 (Er166)Dd: CD33 1 4 1 3 1 2 1 1 CD14- CD33-84.8 CD14+ CD33+ 14.9 1 1 1 2 1 3 1 4 CD19(Nd142)Dd: CD19 1 1 1 2 1 3 1 4 (Sm154)Dd: CD14
Desired Signals CD85j as a Borderline Case CD4+ CD8+ Monocytes 2 2 3 15 15 # Cells 2 1 CD85j- 96.9 CD85j+ 3.7 # Cells 1 5 CD85j- 82.1 CD85j+ 17.9 # Cells 1 5 CD85j- 3.85 CD85j+ 96.2 1 1 1 2 1 3 1 4 (Sm147)Dd: CD85j 1 1 1 2 1 3 1 4 (Sm147)Dd: CD85j 1 1 1 2 1 3 1 4 (Sm147)Dd: CD85j "nega&ve" subpopula&on "posi&ve"
Confirm New Markers With More Than One Donor TCRgd (Yb171)Dd: TCRgd 1 4 1 3 1 2 1 1 TCRgd 3.8 (Yb171)Dd: TCRgd 1 4 1 3 1 2 1 1 TCRgd.377 1 1 1 2 1 3 1 4 (Nd15)Dd: CD3 1 1 1 2 1 3 1 4 (Nd15)Dd: CD3 1 4 CD14+ CD33+ 16.9 1 4 CD14+ CD33+ 8.66 CD33 (Er166)Dd: CD33 1 3 1 2 1 1 (Er166)Dd: CD33 1 3 1 2 1 1 CD14- CD33-82.7 CD14- CD33-91.1 1 1 1 2 1 3 1 4 (Sm154)Dd: CD14 1 1 1 2 1 3 1 4 (Sm154)Dd: CD14
Part 3 Experimental Procedure
CyTOF Sample Info 1. Reagent purity is paramount - MilliQ water only - trace metal pure salts and buffers, if possible 2. All samples - fixed and permeabilized - Thorough fixa&on - Fresh PFA! 3. MilliQ water washes at the end - Proper fixa&on ensure intact cells - Minimizes buffer salt introduc&on into machine 4. Titrate all reagents for your assay condi&ons - Commercial - Every batch of in- house reagents re- verify before use
CyTOF Sample Info Star&ng Cell Count * Remember: only 2-3% cell transmission efficiency - also, processing losses Cells Plated Live intact singlets Titration of cell numbers - Plated vs Live Intact Singlets 2K 28K 15, 5K 1M 76K 175K Live intact singlets 1, 5, R square Live intact Singlets.9895 Yao et al: down to 1K (major only) HIMC: recommend at least 5K 5, 1,, 1,5, Cell #, plated Slope Y-intercept when X=. X-intercept when Y=. Live intact Singlets.1829 ±.188-14297 ± 69 78149
Metal Staining of Cells - DNA Intercalator - Labels any cell containing DNA regardless of whether any labelled an&bodies bind Schaferet al J.Organomet.Chem., 27, 692, 13 139 Ornatsky et al, Anal. Chem., 28, 8, 2539 2547
Chela&ng Polymer Structures X8 linear polymer with ~22 chela&on sites DN3 branched dendrimer, ~16 chela&on sites
An&bodies X8 vs DN3 Comparison
Live- Dead maleimide- DOTA cispla&n
Normaliza&on EQ Four- Element Beads - Polystyrene beads: contain elemental Ce, Eu, Ho, Lu - burn like cells, but are defined composi&on
Normaliza&on Beads - Long Run&me Decrease in Signal Intensity 1:33 AM 6:55 PM
Normaliza&on Affects All Signals Noise amplifica&on! * Global Mean as baseline, rather than Minimum or Maximum - limits noise amplifica&on Normaliza&on CANNOT correct for improper machine cleaning/tuning (undetectable analyte signals). Finck et al, Cytometry A, 213
Signal Decrease Over Very Long Run- Times - Single file: >2 million cells = >3 hr 98% of start 73% of start * Highlights need for moving window normaliza&on Finck et al, Cytometry A, 213
Effects of Normaliza&on - Control sample split in half, run 8.5hr apart; ra&o aker manual ga&ng Freq Parent Median Intensity * Freq Parent fairly robust; only small gains from normaliza&on * Median Intensity more affected by normaliza&on (phosphoflow...)
Part 4 Sample Running
Running Samples 1. Warm up machine ~15-2 min to create and maintain fully hot and stable plasma 2. Tune machine maximize M signal while minimizing M+16 oxide forma&on 3. Finish washing cells final wash in MilliQ water, then resuspension and dilu&on in MilliQ water. - residual buffer salts can affect Current value 4. Resuspend and filter immediately before injec&on - Reduces aggregates (clogs) 5. Appropriate dilu&on minimize doublets - Around 1 cell event/screen refresh; usually ~75K/mL on cell counter
Standard CyTOF Output Appearance 15 - Spike at ~ is legi&mate ( true zero ) - Flow data is usually distributed above and below zero due to comps and auto- fluorescence. # Cells 1 5 CD3-27.8 CD3+ 72.2 * FlowJo se~ngs must be changed for CyTOF data 1 1 1 2 1 3 1 4 CD3(Nd15)Dd: CD3
1 4 Intact cells 76.1 1 4 DNA1(Ir191)Dd: DNA1 1 3 1 2 1 1 DNA1(Ir191)Dd: DNA1 1 3 1 2 1 1 Intact singlets 71.6 1 1 1 2 1 3 1 4 DNA2(Ir193)Dd: DNA2 3 6 9 12 Cell_length: Cell_length CD33(Er166)Dd: CD33 1 4 1 3 1 2 1 1 Monocytes 3.7 Lymphocytes 69.1 DNA1(Ir191)Dd: DNA1 1 4 1 3 1 2 1 1 Live cells 96.1 1 1 1 2 1 3 1 4 CD14(Sm154)Dd: CD14 1 1 1 2 1 3 1 4 Dead(In115)Dd: Dead
CyTOF vs LSRII Representa&ve Examples B Cells CD3+ Cells CD3- Cells Fluorescence <PE-Cy7-A>: CD27 1 5 1 4 1 3 13.6 7.3 <APC-Cy7-A>: CD8 1 5 1 4 1 3 CD8+ 32.1 CD4+ 59.9 <FITC-A>: CD56 1 5 1 4 1 3 1 2 NK cells 19.5 9.51 69.8 1 2 1 3 1 4 1 5 <Am Cyan-A>: IgD 1 2 1 3 1 4 1 5 <PerCP-Cy5-5-A>: CD4 1 2 1 3 1 4 1 5 <Pacific Blue-A>: CD16 Mass CD27(Sm152)Dd: CD27 1 4 1 3 1 2 1 1 14.3 7.64 6.3 72 CD8(Nd144)Dd: CD8 1 4 1 3 1 2 1 1 CD8+ 3.3 CD4+ 61 CD56(Yb174)Dd: CD56 1 4 1 3 1 2 1 1 NK cells 2.1 1 1 1 2 1 3 1 4 IgD(Nd146)Dd: IgD 1 1 1 2 1 3 1 4 CD4(Nd143)Dd: CD4 1 1 1 2 1 3 1 4 CD16(Sm149)Dd: CD16
Flu 21-211 - CyTOF Phenotyping Panel - Collapse 6 tube LSRII panel into 1 tube CyTOF panel 23 markers CD14 CD33 CD3 CD4 CD8 TCRgd CD19 CD2 CD16 CD56 CD45RA CCR7 CD25 CD127 CD27 CD28 CD38 CD85j IgD HLADR CD24 CD94 CD161
Flu 21, Flu 211 Studies - On- going Flu vaccine studies U19 - CyTOF Phenotyping 21, 211 years - Day (pre- vaccina&on) only Study 15, 17, 18 - Day, 1, 4, 7 Study 23-211 - 21-211: >3 unique donors between- donor variability Age Band (y/o) <35 35-65 >65 Total Male 62 22 46 13 Female 84 47 65 196 Total 146 69 111 326
21 Flu Studies Consistency 284 Control - Same Draw, Same Year Replicates
Flu 211, 212 Studies Consistency 885 Control - Generally good year- to- year agreement from same 885 donor draw Naive CD8+ IgD+ CD27- B cells 8 1 Percentage of Parent 6 4 2 Percentage of Parent 8 6 4 2 211 212 Men 35-65 211 212 Men 35-65 Year Year
Flu 211, 212 Studies Consistency 885 Control - Some&mes greater variability in less- frequent popula&ons CD4+ Treg 1 8 Percentage of Parent 6 4 2 211 212 Men 35-65 Year
Flu 21, Flu 211 Studies Donor Variability 8 **** *** B cells total **** **** Percentage of Parent 6 4 2 Mann- Whitney two- tailed * <.5 ** <.1 *** <.1 **** <.1 Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65
Flu 21, Flu 211 Studies Donor Variability Mann- Whitney two- tailed * <.5 ** <.1 *** <.1 **** <.1 Percentage of Parent 1 8 6 4 2 **** CD4+ total **** *** Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65 Percentage of Parent 1 8 6 4 2 **** *** ** CD4+ Naive **** **** * Percentage of Parent 1 8 6 4 2 CD4+ Effector Memory *** * * Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65 Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65
Flu 21, Flu 211 Studies Donor Variability Mann- Whitney two- tailed * <.5 ** <.1 *** <.1 **** <.1 Percentage of Parent 8 6 4 2 * * CD8+ total **** *** * Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65 Percentage of Parent 1 8 6 4 2 **** **** **** CD8+ Naive **** **** **** Percentage of Parent 1 8 6 4 2 CD8+ Effector Memory **** **** * **** *** **** -2 Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65 Men <35 Men 35-65 Men >65 Women <35 Gender and Age Band Women 35-65 Women >65
Study 23-211 (Sm152)Dd: CD27 1 4 1 3 1 2 1 1 CD27+ CD38+.465 Day Day 1 (Sm152)Dd: CD27 1 4 1 3 1 2 1 1 CD27+ CD38+.338 1 1 1 2 1 3 1 4 (Eu151)Dd: CD38 1 1 1 2 1 3 1 4 (Eu151)Dd: CD38 (Sm152)Dd: CD27 1 4 1 3 1 2 1 1 Day 4 Day 7 CD27+ CD38+.566 (Sm152)Dd: CD27 1 4 1 3 1 2 1 1 CD27+ CD38+ 2.26 1 1 1 2 1 3 1 4 (Eu151)Dd: CD38 1 1 1 2 1 3 1 4 (Eu151)Dd: CD38
Conclusions 1. Proper tuning of the CyTOF ensures minimal sample oxida&on. 2. Metal purity (both elemental and isotopic) is highly important. 3. The choice of metal label is dependent upon marker abundance, modality, and resolu&on needed (fold- change). 4. All samples must be stained with metals to be detected by the CyTOF. Proper fixa&on is important, as are MilliQ water washes. 5. Head- to- head comparisons with LSRII assays are consistent. 6. The Flu studies are an example of how a CyTOF panel can be used to monitor the surface phenotype for hundreds of donors to detect varia&ons between gender, age- band, and &me- point.
Acknowledgements - HIMC staff - Henrik Mei, Meena Malipotlalla, Holden Maecker - External CyTOF: - Nolan lab: Sean Bendall, Erin Simonds, Eli Zunder, Bernd Bodenmiller - Davis lab: Evan Newell, Bill O'Gorman, Peber Brodin - Fluidigm/DVS Sciences - U19 NIH Grants and HIMC Service Center fees hbp://himc.stanford.edu