1. Bead Normalization of Mass Cytometry Data and Mass-Tag Cellular Barcoding Rachel Finck May 7, 2014

2. Measure by TOF Stimulate cells in vitro Crosslink proteins Stain with isotope tagged Abs Nebulize single-cell droplets Ionize (7500K) Permeabilize cell membrane E. Simonds CyTOF Acquisition Decisions are Hidden in the Details

3. 5 2 5 9 5 0 5 2 1 2 4 3 1 2 1 0 1 2 1 0 1 1 4 1 2 3 2 1 Ion cloud splits into pushes before reaching the detector Each pulses, a.k.a. ‘push’, includes an entire mass spectrum A cell’s ion cloud will span ~20-60 pulses Single-cell events are interpreted by a post-processing peak detection step The data recorded for a particular mass in a given push is indicative of the number of ions reaching the detector during that metal’s time window within the pulse Ion Intensities and Pulse Counts of a Single Cell Push

4. Ease of Quality Assessment/Control Can Be Obfuscated by High-Dimensional Analyses vs. SPADE ViSNE Wanderlust CD3+ T cells CD4+T cells CD8+T cells

5. Ease of Quality Assessment/Control Can Be Obfuscated by High-Dimensional Analyses vs.

6. Outline –Normalization of Mass Cytometry Data with Bead Standards Review of normalization algorithm Demonstration of normalization software with updated beads –Mass-Tag Cellular Barcoding Dose response to inhibitors Single-cell deconvolution of barcoded populations “Doublet-free” barcoding scheme

7. Goal: Reliably compare mass cytometry data across patients, conditions, tissues, etc. Problem: Drifts in mass cytometry instrument sensitivity over time due to cellular debris, fluctuations in plasma temperature, and calibrations. Solution: Normalization using internal bead standards measured concurrently with cell samples. Normalization of Mass Cytometry Data

8. Before Normalization Signal Drift Across Multiple Conditions Measured in Human Bone Marrow Has Been Corrected Using Unstimulated Controls After Normalization Bendall, Simonds, et al., Science, May 2011 time

9. Internal Bead Standards Generalize Normalization Capabilities La139 Pr141 Tb159 Tm169 Lu175 Desirable Features for Internal Standard Independent of biological sample Not subject to variable staining Can be directly mixed with sample to monitor instrument sensitivity at all times beads were gift from DVS Sciences Metal-embedded polystyrene bead Span instrument mass and dynamic ranges Unique 5-element signature allows antibodies to be tagged with bead elements Ce140 Eu151 Eu153 Ho165 Lu175

10. cells beads Measure by TOF Beads can be spiked into standard mass cytometry protocol Stimulate cells in vitro Crosslink proteins Stain with isotope tagged Abs Nebulize single-cell droplets Ionize (7500K) Permeabilize cell membrane E. Simonds

11. Beads are Identifiable from Cells Even When Measured on Overlapping Channels No reduction in number of parameters since beads and element-tagged antibodies can be simultaneously measured on the same channel

12. Bead Smoothing Removes Local Variance 0 6000 8000 1000 0 1200 (sec)

13. 0 6000 8000 1000 0 1200 (sec) Bead Smoothing Removes Local Variance

14. Fitted Slopes Define a Correction Function 0 6000 8000 1000 0 1200 (sec)

15. Fitted Slopes Define a Correction Function 0 6000 8000 1000 0 1200 (sec)

16. Fitted Slopes Define a Correction Function 0 6000 8000 1000 0 1200 (sec)

17. x Fitted Slopes Define a Correction Function 0 6000 8000 1000 0 1200 (sec)

18. Fitted Slopes Define a Correction Function 0 6000 8000 1000 0 1200 (sec)

19. 4.9-fold mean range 1.3-fold mean range Normalization Reduces Bead Intensity Variation 0 6000 8000 1000 0 1200 (sec)

20. Normalization Aligns Surface Marker Histograms Day A Day B Day C Day D Day A Day B Day C Day D CD4 in T-cellsCD8 in T-cells CD14 in Monocytes CD38 in Monocytes HLA-DR in Monocytes CD19 in B-cells 61.7% 61.9.0% 62.2% 61.8% 62.4% 61.5% 61.9% 62.3% 97.1% 98.0% 97.7% 98.2% 97.3% 97.8% 97.5% 98.1% 36.5% 35.8% 35.4% 34.7% 37.2% 34.8% 34.1% 34.7% 87.0% 87.8% 89.2% 88.8% 87.2% 87.4% 88.3% 89.4% 92.8% 97.3% 97.9% 97.6% 93.8% 96.7% 97.1% 97.2% 97.5% 98.5% 98.7% 98.6% 97.5% 98.5% 98.6%

21. Internal Bead-based Normalization Corrects Intra-File Decay 73% of initial intensity 98% of initial intensity

22. Multivariate Relationships are Preserved by Normalization Before After Pairwise Correlations Between Surface Markers in a Single File

23. 4.1-fold mean range 1.2-fold mean range Normalization Reduced Non-Biological Variation in Primary Pediatric ALL Bone Marrow Samples

24. patient Normalization Reduced Non-Biological Variation in Primary Pediatric ALL Bone Marrow Samples

26. 1. Adjust gates. 2. Approve gates. 3. Set bead removal threshold. Normalization Software Example Using New Beads

27. Normalization Software Reports Bead Intensity Ranges

28. Relative Bead Metal Intensities Vary Across Calibrations and Instruments Instrument 1 Instrument 2 Day ADay BDay ADay B time

29. Normalization Summary the intensities of the normalization beads quantitatively monitor instrument sensitivity over acquisition time the effects of instrument variation on mass cytometry data are reduced by a correction derived from internal 5- metal bead standards normalization allows for the direct comparison of samples collected from multiple individuals and treated under different conditions

30. Multiplexed mass cytometry profiling of cellular states perturbed by small- molecule regulators Bernd Bodenmiller*, Eli R. Zunder*, Rachel Finck*, Tiffany J. Chen, Erica S. Savig, Robert V. Bruggner, Erin F. Simonds, Sean C. Bendall, Karen Sachs, Peter O. Krutzik and Garry P. Nolan Nature Biotechnology. 2012 September 10;30(9):858-67 Bernd BodenmillerEli Zunder

31. Cell Multiplexing/Barcoding Overview Advantages: 1.Uniform Staining 2.Reduced Antibody Consumption 3.Reduced Acquisition Time 4.Improved Singlet Detection Krutzik PO, Nolan GP. Fluorescent cell barcoding in flow cytometry allows high- throughput drug screening and signaling profiling. Nat Methods. 2006 May;3(5):361-8.

32. Combinatorial Cell Labeling by DOTA-Maleimide Mass-Tag Cell Barcode (MCB) Reagents

33. 7-metal Binary Cell Labeling Scheme for 96-well MCB Multiplexing 2 1 =2 2 2 =4 2 3 =8 2 4 =16 2 5 =32 2 6 =64 2 7 =128

34. Small Molecule Kinase Inhibitor Profiling Calculation of phosphorylation site dose responses to inhibitors after various stimulations in multiple cell types.

35. Small Molecule Kinase Inhibitor Profiling

36. 18,816 measurements from a single multiplexed tube 2,352 dose response curves

37. Dose Response Curve Website

38. Anastassiadis et al. Comprehensive assay of kinase catalytic activity reveals features of kinase inhibitor selectivity. Nat Biotechnol. 2011 Oct 30;29(11):1039-45. How Does Our 96-well MCB In Vivo Analysis Compare to In Vitro Kinase Inhibition Assays?

39. PBMC In Vivo IC50s Strongly Correlate with In Vitro Percent Inhibition Values Anastassiadis et al. Comprehensive assay of kinase catalytic activity reveals features of kinase inhibitor selectivity. Nat Biotechnol. 2011 Oct 30;29(11):1039-45. Davis et al. Comprehensive analysis of kinase inhibitor selectivity. Nat Biotechnol. 2011 Oct 30;29(11):1046-51.

40. U. Toronto / DVS Sciences Scott Tanner Olga Ornatsky Vladimir Baranov Dmitry Bandura Tad George Columbia Dana Pe’er Smita Krishnaswamy El-ad David Amir Prof. Garry P. Nolan www.stanford.edu/group/nolan Sean Bendall Erin Simonds Astraea Jager Kara Davis Wendy Fantl Eli Zunder Bernd Bodenmiller Rob Bruggner Karen Sachs Greg Behbehani