Multiplex digital nucleic acid quantitation using molecular barcodes Paul Rasmussen Sr. Manager of Emerging Markets and Consumable Programs prasmussen@nanostring.com NanoString Confidential.

Agenda Platform Introduction Chemistry Overview Performance Application extensions miRNA and CNVs NanoString Confidential.

NanoString nCounter assay: Single reaction, up to 800 targets QPCR NanoString Sensitivity Microarrays/NGS . 1 10 100 1000 10,000 Multiplexing # of Transcripts NanoString Confidential. 3

The nCounter platform facilitates powerful research Over 100 papers have been published using the nCounter platform as of June-2012, at a rate more than doubling yearly >20% are published in Science, Nature, Cell, or PNAS Publications span most major disciplines in molecular biology Cancer, Immunology, Stem Cells, Systems Biology, Agriculture Driven by performance with FFPE and remarkable precision Access previously unusable samples Observe biology not previously possible Publication rate underestimates utilization of the platform by large pharma and industry. Often less motivated to publish

The nCounter Analysis System: Two fully automated instruments nCounter Prep Station nCounter Digital Analyzer Fully Automated sample processing Up to 800 genes per sample 12 samples processed in one cartridge Up to 4 cartridges per day Fully automated imaging and counting Up to 6 cartridges (72 samples) per day 24 hour unattended processing Simple data output 5 5

Agenda Platform Introduction Chemistry Overview Performance Application extensions miRNA and CNVs NanoString Confidential.

Each Barcode Attached to an Individual RNA 7

Two Probe Assay Target Specific Capture & Reporter Probes are created to bind to the mRNA transcript Biotin Target Specific Capture Probe Target Specific Reporter Probe NanoString Confidential. 8

Both probes must hybridize Target Specific Capture & Reporter Probes are created to bind to the mRNA transcript Target Specific Capture Probe Target Specific Reporter Probe 9

nCounter CodeSet Pre-mixed sets of all probes and controls Capture Probes System Controls Target specific Capture probes and reporter probes are combined to create your custom CodeSet. To reiterate, NanoString builds the barcodes and assigns each gene sequence a unique barcode. Your CodeSet can contain target specific capture and reporter combinations sufficient for 10-576 genes of your choice in a single tube. Both capture probes and reporter probes are in excess to drive the hybridization reaction to completion. Each CodeSet is synthesized with dozens of internal controls included that allow for assessing the performance of each step in the process and for quantification of your endogenous transcripts. There is no need to waste extra reagents or sample material running extra controls like you would have to using qPCR, they are already included in each codeset. Customers may opt to have additional endogenous genes included for normalization. Sales Reps - Discuss the controls in depth later in the presentation during data. Reporter Probes NanoString Confidential. 10

The nCounter Assay: Three Simple Steps 5 min HANDS-ON 5 min HANDS-ON 5 min HANDS-ON Day 1 Day 2 AUTOMATED Day 2 AUTOMATED nCounter Prep Station nCounter Digital Analyzer Hybridize 1 Flexible sample requirements Only 4 pipetting steps No amplification 800 hybridizations in single tube Purify 2 Count 3 Sensitive Precise Quantitative Simple The protocol could not be more simple. You add buffer, your CodeSet and sample into a strip tube and hybridize overnight. Once hybridized, you simply load your hybridized samples, the nCounter Prep Plates and the nCounter Sample Cartridge on to the nCounter Prep Station. Each run processes a single sample cartridge, or 12 samples. This takes about 2.5 hours. After processing, you transfer your sample cartridge to the nCounter Digital Analyzer for imaging and data collection. The image processing and data collection take about 20 minutes per sample. You can load six cartridges on the analyzer and let it run overnight to process up to 72 samples unattended. Both the nCounter Prep Station and the nCounter Digital Analyzer have easy-to-use touch screens that walk you through each step of the process. Everything you need to process the nCounter gene expression assay is available in a master kit. This includes all reagents and consumable plasticware (comprised of 3 sub-kits that have different storage requirements.) NanoString Confidential. 11

Capture & Reporter Probes nCounter Assay Hybridize CodeSet to RNA Remove excess reporters Bind reporter to surface Immobilize and align reporter Image surface Count codes mRNA Capture & Reporter Probes NanoString Confidential. 12

nCounter Assay Hybridized mRNA Excess Reporters Hybridize CodeSet to RNA Remove excess reporters Bind reporter to surface Immobilize and align reporter Image surface Count codes After the samples are hybridized, all of the sample processing steps are automated on the nCounter Prep Station. The Prep Station uses a magnetic bead purification to isolate only Reporter Probes that have found a target. The excess CodeSet is washed away. Hybridized mRNA Excess Reporters NanoString Confidential. 13

Hybridized Probes Bind to Cartridge nCounter Assay Hybridize CodeSet to RNA Remove excess reporters Bind reporter to surface Immobilize and align reporter Image surface Count codes Surface of cartridge is coated with streptavidin Hybridized Probes Bind to Cartridge NanoString Confidential. 14

nCounter Assay Hybridize CodeSet to RNA Remove excess reporters Bind reporter to surface Immobilize and align reporter Image surface Count codes Immobilize and align reporter for image collecting and barcode counting NanoString Confidential. 15

One coded reporter = 1 mRNA nCounter Assay Hybridize CodeSet to RNA Remove excess reporters Bind reporter to surface Immobilize and align reporter Image surface Count codes Image Surface One coded reporter = 1 mRNA NanoString Confidential. 16

Codes are counted and tabulated nCounter Assay Hybridize CodeSet to RNA Remove excess reporters Bind reporter to surface Immobilize and align reporter Image surface Count codes Code Gene Count x 3 y 1 z 2 Codes are counted and tabulated NanoString Confidential. 17

Simple read out of counts NanoString Confidential.

Agenda Platform Introduction Chemistry Overview Performance Application extensions miRNA and CNVs NanoString Confidential.

The nCounter Assay: Very Reproducible Reproducibility of NanoString Assay Technical Replicates R2 = 0.9999 Replicate 1 Counts H Replicate 2 Counts Data Courtesy of Dr. Roger Bumgarner NanoString Confidential. 20 20

Superior Precision in Site-to-Site Reproducibility Rapid, reliable, and reproducible molecular sub-grouping of clinical medulloblastoma samples Northcott P.E. et al., Acta Neuropathologica; November 16, 2011 Site 1 Site 2 Site 3 “We present an assay based on NanoString technology that is capable of rapidly, reliably, and reproducibly assigning clinical FFPE medulloblastoma samples to their molecular subgroup, and which is highly suited for future medulloblastoma clinical trials.” The other major advantage I mentioned earlier is nCounter’s reproducibility. These data are an excellent demonstration of this and were published last November by another group out of Toronto (at the Hospital for Sick Kids) The authors used nCounter to validate a gene expression signature for medulloblastoma, but then they took there analysis one step further. They shipped 48 RNA samples to 2 additional sites and performed an analysis of site-to-site reproducibility. The heat maps show expression levels and patient categorization from each site The scatterplot shows excellent correlation between the sites with R2 or .97 and .98 respectively when comparing Toronto to either of the 2 other sites R2 Site1 v Site 2 = 0.97 R2 Site1 v Site 3 = 0.98

Very good cross platform performance: TaqMan qPCR Log2 Fold Change 100ng BR/HR Total RNA nCounter Log2 Fold Change BR/HR TaqMan 6 NanoString reactions, 372 PCR reactions PCR data from: Canales et al. Evaluation of DNA microarray results with quantitative gene expression, Nature Biotechnology 24, 1115 - 1122 (2006) NanoString Confidential. 22 22

Very good cross platform performance: qPCR Khan et al., 2011

Very good cross platform performance: Affymetrix 6 calibration genes were selected based on their known expression profiles in each myeloid cell fraction (CD34+ (2), promyelocyte (5), neutrophils(2)), M3 AML subtypes (11) and other AML subtypes (17). Both NanoString and Affymetrix data are shown. 28 samples were run in total (in parenthesis above). Data normalized to the index group (shown by asterisk). Early Myeloid-specific heatopietic genes Promyelocyte-specific genes Late myeloid-specifici genes Take hime message- On both a signature wide (33 genes) and individual gene basis, NanoString showed good correlation to the initial microarray data Payton et al, High throughput digital quantification of mRNA abundance in primary human acute myeloid leukemia samples. J Clin Invest. June 2009

PCA of signature on Affymetrix and NanoString Ultimately, the real litmus test is the signature’s ability to hold up in differentiating subtypes. The Principle component analysis above shows the original affy data (upper left), A subset of the original affy sample set on NanoString( upper right) and two validation cohorts on NanoString (bottom). Note that there is good separation of the M3 subtype (red) from M), M1, M2, and M4 subtypes (grey). Also in the lower left, note the yellow samples. These were PML-RARA positive M3 samples initially missed by traditional cytogenetics. The green sample was morphologically called M2, but contains the PML-RARA translocation, and is classified as such in the sample set. Payton et al, High throughput digital quantification of mRNA abundance in primary human acute myeloid leukemia samples. J Clin Invest. June 2009

Very good cross platform performance: RNA-Seq Sun et al. Integrated analysis of Gene Expression, CpG Island Methylation, and Gene Copy Number in Breast Cancer Cells by Deep Sequencing. PLoSone, Feb, 2011 NanoString Confidential.

What samples are you using: flexible options Total RNA and DNA (100ng/300ng/sample) Amplified RNA from Small Amount of Sample LCM and single cell (in progress) Whole Cell Lysates PaxGene Lysed Whole Blood Total RNA and DNA Extracted from FFPE Samples Crude Extracts from FFPE samples Plasma, Serum and other Biofluids

Formalin fixation inhibits qPCR much more than the nCounter platform Fold Decrease

Unparalleled Performance on FFPE Samples mRNA Transcript Quantification in Archival Samples Using Multiplexed, Color-coded Probes Reis, P.P. et al., BMC Biotechnology; May 9, 2011 nCounter® (r = 0.90) qPCR (r = 0.50) “… the probe-based NanoString method achieved superior gene expression quantification results when compared to RQ-PCR in archived FFPE samples. We believe that this newly developed technique is optimal for large-scale validation studies using total RNA isolated from archived, FFPE samples.” NanoString Confidential.

Sample flexibility: Cell Lysate and Matched Total RNA Measurements with crude whole cell lysates correlate extremely well with purified RNA NanoString Confidential.

Flexibility of sample input Total RNA vs Lysate Khan et al., 2011 NanoString Confidential.

PaxGene Blood Lysates vs. Purified RNA Measurements with unpurified PAXgene blood lysates correlate extremely well with purified RNA log2 counts from total RNA purified from blood log2 counts from blood lysate NanoString Confidential.

Agenda Platform Introduction Chemistry Overview Performance Application extensions miRNA, and CNVs NanoString Confidential.

nCounter miRNA Assays Human miRNA Panel Mouse miRNA Panel Rat miRNA Panel Detects: 800 human miRNAs 578 mature mouse miRNAs 423 mature rat miRNAs 3 nonhuman miRNAs (possible spike in controls) 33 mature murine-associated viral miRNAs All miRNA Panels Sample Types Supported: RNA from Fresh/frozen tissue, FFPE, Blood, Cells Sample Input Recommendation: 100 ng purified total RNA Linear Dynamic Range: 2106 counts Hands on Time: <2 hours

Current challenges to miRNA detection Short length overall low Tm prohibits concurrent binding, e.g. nCounter dual probe Highly Related Sequences Large sequence diversity leads to large Tm spread even though length distribution is fairly small NanoString Confidential.

miRNA Sample Preparation Basics

miRNA Sample Preparation Basics Hybridize bridge oligo to each miRNA target miRNAs

miRNA Sample Preparation Basics Bridge oligo specifically anneals to each miRNA target Unique miRtag for each miRNA species miRNAs

miRNA Sample Preparation Basics Bridge oligo specifically anneals to each miRNA target Unique miRtag for each miRNA species miRNA is covalently linked to miRtag via ligation miRNAs

miRNA Sample Preparation Basics Excess bridges and tags are removed

Probe Architecture Target Specific Capture & Reporter Probes bind to the chimaeric miRNA:miRtag molecule Biotin The target sequence for the Capture Probe and the Reporter probe are designed to be adjacent to one another along a transcript. Our probe design process picks probes with no gap or a one base pair gap in between the Capture Probe and Reporter Probe. Probes are selected to span exon-exon boundries if possible. Genomic DNA contamination has not been an issue in the nCounter system. The system never reaches denaturing conditions when used for RNA. Crossing exon boundaries also serves to eliminate the possibility of genomic contamination. Additionally the total RNA purification further reduces the potential. All things combined lead to no discernible genomic contamination in our hands. We have data further on in the presentation that shows empirical evidence of this (cell lysate slide). With regard to gene families, We default in selection to capture the most transcripts possible. If there are several transcripts selected that are closely related in a gene family, a single probe will be selected to represent as many transcripts as possible in that family. After the capture probe and reporter probe are bound to the mRNA, a stable tripartite structure is formed. Target Specific Capture Probe Target Specific Reporter Probe 41

nCounter miRNA Analysis Unambiguous Discrimination of miRNAs with Single Nucleotide Differences Multiplexed target profiling of miRNA transcriptomes in a single reaction Available for human, mouse, rat and drosophila High level of sensitivity, specificity, precision, and linearity nCounter miRNA kits enable precise analysis of comprehensive collections of miRNAs from human, mouse and rat. To illustrate our ability to distinguish miRNAs that differ by a single nucleotide, we looked at the let7 family of highly related miRNAs. We synthesized probes against the 8 different let7 family members, and examined each for their reactivity with designated target, as well as cross-reactivity against other let7 familiy members. Because we use a ligation-based method to extend the length of miRNAs, we can design the bridges and tags to exploit mismatches (they won’t ligate) and as a result there is little to no cross-reactivity,,

Dynamic range of the miRNA assay 100ng commercial purified RNA was analyzed with a single codeset detecting 676 miRNAs. Counts shown here are spike-normalized, singlet measurements. Counts were recovered over a dynamic range of 4-5 orders of magnitude. The two tissue types analyzed here are similar in origin (muscle-derived), so the high correlation (0.823) might reflect similar overall patterns of miRNA expression. Note howver that there are clearly differences in abundance of some miRNAs in the two muscle types.

Dynamic range of the miRNA assay hsa-miR-1 expression in different tissues Tissue-specific detection of a muscle-specific miRNA.

New miRNA assays for the nCoutner assay A la carte miRNAs Select from 20-50 miRNAs from our panels for focused profiling Workflow identical to standard miRNA assay Users specify housekeepers (stable miRNAs for normalization) miRGE assay Mixed mRNA and miRNA codesets 10-30 miRNAs from our panels Up to 200 mRNAs Workflow similar to our standard miRNA assay NanoString Confidential.

nCounter Copy Number Assay – Sample Preparation Perform nCounter Hybridization Prior to hybridization, genomic DNA is fragmented and denatured. After that, the hybridization steps are the same as for other nCounter assays. Genomic DNA Fragment ds genomic DNA using 4 base cutter (Alu 1) Average~ 500 bases Denature DNA @ 95C 46 NanoString Confidential. NanoString Confidential.

Variable copies of X chromosome Feasibility initially demonstrated with cell lines containing 1, 2, 3, 4 and 5 X chromosomes Requires fragmentation and denaturation of genomic DNA The accuracy (measured vs. expected values) obtained with the nCounter system is extremely high NanoString Confidential.

Calls agree with HapMap One copy No copies No copies HapMap CNV nCounter CNV NanoString Confidential.

Accuracy (concordance) Very good correlation with HapMap Experiment: 100 HapMap samples + reference assayed (600ng digest/300ng hyb) 20 CNV regions with 3 probes analyzed. Metric % of total Data points Call rate 94.8% 1902/2006 Accuracy (concordance) 94.2% 1791/1902 NanoString Confidential.

The importance of Karyotyping…. Normal Female HeLa NanoString Confidential.

Molecules That CountTM Thank You! Molecules That CountTM Paul Rasmussen prasmussen@nanostring.com 51