Technical Note Sample Flexibility for Gene Expression Analysis with the ncounter Analysis System Sensitive Gene Expression Analysis From Crude Preparations and Fragmented RNA Cell Lysates Tissue Homogenates Whole Blood Lysate Without Globin Mitigation Fragmented RNA from FFPE Tissue The enzyme-free nature of the ncounter System enables sensitive gene expression analysis from crude preparations from various biological sources and from fragmented RNA. In contrast, other technologies such as microarrays require high-quality RNA to achieve acceptable results. By analyzing crude lysates on the ncounter System, scientists can bypass RNA purification issues that are unavoidable with other platforms. For example, RNA purification methods vary widely by tissue source and many sources present difficult challenges. Also, extracting RNA from a large number of samples can be tedious and costly in terms of materials and labor. Thus, by accurately measuring gene expression with minimal sample manipulation, the ncounter System offers great value to research and clinical laboratories. In this Technical Note, we describe how the ncounter Analysis System can analyze expression levels in crude preparations from various sources. We show that lysates from highly purified cells and from tissue homogenates can be analyzed without purification and result in excellent correlations to purified RNA from the same samples. We also demonstrate that the ncounter platform can measure transcript abundance with equal efficiency in RNA purified from blood or from whole blood lysate, without globin mitigation. Finally, we demonstrate that fragmented RNA extracted from formalin-fixed, paraffinembedded (FFPE) tissue produces excellent correlations with measurements made from freshly-prepared RNA from the same tissue. ncounter Analysis System Overview The ncounter Analysis System delivers direct, multiplexed measurement of gene expression, providing digital readouts of the relative abundance of hundreds of mrna transcripts simultaneously. The ncounter Analysis System is based on gene-specific probe pairs that are hybridized to the sample in solution. The protocol eliminates any enzymatic reactions that might introduce bias in the results (Figure 1, step 1). The Reporter Probe carries the fluorescent signal; the Capture Probe allows the complex to be immobilized for data collection. Up to 550 pairs of probes specific for a particular set of genes are combined with a series of internal controls to form a CodeSet. After hybridization of the CodeSet with target mrna, samples are transferred to the ncounter Prep Station (Figure 1, step 2) where excess probes are removed and probe / target complexes are aligned and immobilized in the ncounter Cartridge. Cartridges are then placed in the ncounter Digital Analyzer for data collection (Figure 1, step 3). Each target molecule of interest is identified by the color code generated by six ordered fluorescent spots present on the reporter probe. The Reporter Probes on the surface of the cartridge are then counted and tabulated for each target molecule. Methods and Results ncounter Analysis of Whole Cell Lysates The ncounter system does not use enzymes for any of the steps for measuring gene expression levels, raising the possibility that whole cell lysates could be used directly without further manipulation. To test whether a whole cell lysate could be used in the hybridization mixture, we obtained purified human umbilical cordvein cells (Cascade Biologics). Whole-cell lysates were prepared by resuspending approximately 10,000 cells per microliter of RLT buffer from the RNeasy Kit (QIAGEN GmbH). For comparison, we prepared total RNA from lysates described above using the RNeasy Kit, according to the manufacturer s instructions. One hundred nanograms of 1 2 3 Figure 1. The Nanostring technology.
Figure 2. Correlation of three different concentrations of whole-cell lysate with total RNA prepared from the same cells. total RNA or lysate containing 10,000, 5,000 or 2,500 cells was incubated with a 96 human-gene CodeSet and processed according to the standard ncounter Gene Expression Assay Manual (Geiss et al, 2008). Counts were normalized to internal control spikes as described in the assay manual. We compared the counts from 100ng purified total RNA to counts from the 10,000-cell lysate (Figure 2). In both sample types, the ncounter system was able to quantify low- and high-expressing genes in a single hybridization and the results were highly correlated (R 2 =0.97). The slope of the linear regression (~ 1.3), indicates the difference in magnitude between counts in the samples and suggests that we recovered approximately 30 percent more counts from the whole cell lysate. To illustrate the consistency across different starting amounts of cell lysates, we performed identical comparisons with 5,000 cells and 2,500 cells. The correlation between counts obtained from these samples and total RNA remained as high as with the 10,000 cell sample (R 2 ~ 0.97). The slopes decrease linearly according to the decrease in starting cell lysate: the slope for 5,000 cells is half that of 10,000 cells (0.649), and the slope for the 2,500 cell sample was 0.298, approximately one-fourth that of 10,000 cells. Similar results demonstrating high correlations between lysate and total RNA preparations have been obtained using lysates from cells grown in culture (data not shown). Another common method of purifying RNA starts with cell lysis in Trizol (Molecular Research Center, Inc.). We compared ncounter results from the aqueous phase of a Trizol extract with results from total RNA purified using the Trizol method and again found them to be highly correlated (Figure 3). Because the cell density of the samples was not accurately quantified before preparing the extracts, the average counts are higher in the lysate sample; however, the correlation coefficient is 0.95. We have obtained similar results with other wholecell lysis and disruption solutions, including SV RNA lysis buffer (Promega), solution D (Chomczynski and Sacchi, 1987), and tissues preserved with RNAlater (Ambion, Inc.). These results show that the ncounter platform can quantify RNA expression levels in cell lysates and crude RNA preparations with equivalent results to purified RNA. ncounter Analysis of Tissue Homogenates Because the ncounter assay is an enzymefree solution-based hybridization, quantification of transcript levels should be insensitive to molecules present in extracellular matricies. To test this hypothesis, a mouse whole liver lysate was prepared by homogenizing mouse liver in QIAGEN RLT buffer. Half of this lysate was used directly for ncounter analysis, while total RNA was prepared from the other half before analysis. The lysate and purified RNA was hybridized with a 150 mouse-gene CodeSet, and processed and analyzed as described (Geiss et al, 2008). In both sample types, the ncounter system was able to quantify low- and highexpressing genes in a single hybridization (Figure 4). As with purified cells, the Figure 3. Correlation between counts from purified RNA with counts from the aqueous phase of a Trizol lysate of the same cells. Figure 4. Correlation of counts from whole mouse liver lysate with purified total RNA from the same tissue.
correlation between counts obtained from total RNA and the whole liver lysate was very high (R 2 =0.98). The slope in this experiment was high; since the absolute number of cells in the liver lysate was not determined when the experiment was set up, there was approximately 770 ng of RNA in the lysate sample used for the hybridization. Similarly high correlations between purified RNA and whole tissue lysates have been observed using mouse lung cancer xenografts (Malkov et al, 2009). High correlations between whole yeast lysates and RNA purified from the same cells have also been observed (data not shown). Together, these results demonstrate that the ncounter platform can be used to analyze gene expression levels in crude whole-tissue lysates without further manipulation. ncounter Analysis of Blood-Derived Samples Many research and diagnostic protocols analyze gene expression levels in samples derived from blood collections. However, since up to 75 percent of the mrna in a blood sample is beta-globin, measures are needed to ensure the nucleotides and enzymes required for copying and amplification are not exhausted by amplifying beta-globin. These globin mitigation strategies can skew the results of array- or PCR-based studies (Li et al, 2008, Liu et al, 2006). We therefore determined the performance of the ncounter platform by measuring expression levels in whole blood samples with and without globin mitigation. Blood samples from five individuals were collected directly into PAXgene (PreAnalytiX, GmbH) tubes and RNA was purified according to manufacturer s instructions. The purified RNA was split in half: one half was treated with a globin reduction protocol and the other half was untreated. One hundred micrograms of each sample type was hybridized overnight with a 152-gene CodeSet and processed on using the ncounter platform according to standard protocol. Counts were normalized to internal spikes. To correct for the total mrna content in each assay, we further normalized to the geometric mean of all the gene counts in the sample. Since the samples were from five different individuals, the average counts of each gene in the five globin reduced samples was compared to the average counts of the five untreated samples. Globin mitigation did not change the pattern of gene expression as measured by the ncounter System (Figure 5). The correlation between samples with and without globin mitigation was very high (R 2 =0.99), similar to that observed in technical replicates. Interestingly, we found that the endogenous gene count averages obtained after globin mitigation were approximately 3-fold higher per 100 ng input than in the unmitigated sample. This might reflect the fact that such a large fraction of total RNA from blood is beta-globin and ribosomal RNA sequences. Once these sequences are removed, the effective concentrations of all other sequences increase and might be more similar to other tissues. Nevertheless, after normalizing for mrna content in samples with and without globin mitigation the gene counts were very similar. This demonstrates that globin mitigation has no effect on the ability of the ncounter system to quantify mrna content, even at very low expression levels. Although beta-globin transcripts have no effect on the measurement of unrelated sequences in an ncounter assay, users should be cautioned against having probes that recognize beta-globin in their CodeSet. The abundance of beta-globin transcripts is so great that the probe-transcript hybrids will saturate the imaging surface and interfere with the ability to count unrelated transcripts. Therefore, NanoString has developed methods for attenuating such highly expressed genes, so the assay performance is unaffected (see the Technical Note: Attenuation Strategies for Highly Expressed Genes). Researchers can contact NanoString for assistance with attenuation strategies for measuring transcript levels of beta-globin or other highly-expressed genes. Next, we determined whether as observed in other cell and tissue types, unpurified RNA from blood collections could be used in the ncounter assay. We therefore collected blood from one individual in two PAXgene tubes. The tubes were spun to separate the stabilized RNA from the rest of the cellular debris. In one tube, the resulting precipitate was resuspended in Solution D (Chomczinski and Sacchi, 2007) and this resuspended material was used in ncounter assays as lysate. Total RNA was purified from the second PAXGene collection tube using the protocol supplied by the manufacturer (QIAGEN). Lysate or purified RNA was hybridized with a 95- gene CodeSet in triplicate, and the samples were processed on the ncounter System Figure 5. Correlation of counts from total RNA from blood before and after beta-globin mitigation. Figure 6. Correlation of counts from total RNA purified from PAXGene-stabilized blood with counts from crude blood lysate.
according to standard protocol. Counts were normalized to spike-in controls as previously described. The counts were further normalized to the geometric mean of three housekeeping or internal reference genes (B2M, ACTB and POLR1B) that were present in the CodeSet so the counts in the two sample types could be directly compared. As with samples derived from other cell and tissue types, we were able to detect a wide range of expression levels in a single hybridization. The correlation between the two samples was extremely high (R 2 =0.99). We have obtained similar results by resuspending the stabilized nucleic acid pellet in QIAGEN RLT buffer; moreover, we found that a QIAshredder (QIAGEN GmbH) step to shear genomic DNA is not needed. Therefore, for the ncounter assay, it is not necessary to purify RNA from blood collected in PAXgene tubes - the stabilized cell pellet can simply be resuspended in a guanidinium-based buffer and then hybridized with the CodeSet. ncounter Analyses of FFPE Samples Many clinically important samples have been archived as Formalin-fixed, paraffinembedded (FFPE) blocks, and in some cases sectioned and preserved on slides. These treatments introduce modifications to the nucleotides and fragment the RNA molecules, making gene expression analyses of archived samples challenging. Nevertheless, the enzyme-free nature of the ncounter platform can overcome these obstacles and facilitate the analysis of RNA extracted from archived FFPE samples provided the block is in good quality and the RNA is not severely degraded. To illustrate this, we describe the outcomes of ncounter analyses of gene expression levels in total RNA prepared from frozen pieces of human heart tissue versus RNA extracted from matched cryropreserved and FFPE slices of the same heart (Biochain Institute Inc.). Total RNA was purified from a flash-frozen piece of heart using the QIAGEN RNeasy Kit according to manufacturer s instructions. RNA was extracted from cryoropreserved sections using the QIAGEN Fibrous Tissue RNeasy Mini-Kit according to manufacturer s instructions. RNA was extracted from FFPE slices using the QIAGEN RNeasy FFPE Kit, again according to manufacturer s instructions. One hundred nanograms of RNA from each sample was hybridized to a 96-plex human gene CodeSet, processed on the ncounter platform, and normalized according to the standard protocol. To assess the quality of the RNA, we ran total RNA purified from fresh tissue and extracted from cryropreserved and FFPE slices on an Agilent Bioanalyzer (Figure 7). To quantify the degree of RNA degradation, we measured the percentage of RNA molecules which size is between 50 bp and 300 bp. In the RNA sample prepared from frozen unsectioned heart tissue, 6% of the total RNA is between 50 and 300 bp; similarly, RNA recovered from frozen heart sections had 5% of RNA between 50bp and 300 bp and RNA recovered from FFPE had 36% of the RNA between 50 and 300 bp. Furthermore, in the FFPE-extracted sample, the ribosomal RNA peaks were missing, indicating that RNA in the sample was degraded. To determine the effects of this amount of degradation on the ncounter platform, we analyzed expression levels of 96 genes in these RNA samples (Figure 8), and compared counts obtained in flash frozen tissue to RNA extracted from cryropreserved heart slices or FFPE slices. When compared to total RNA, the expression levels in the RNA prepared from cryropreserved slices correlated well (R 2 =0.99). A similar correlation was observed in the RNA prepared from FFPE samples compared to total RNA (R 2 =0.97). However, there was a slight overall reduction in counts, illustrated by the reduced slope of the regression analysis, when comparing expression levels in RNA prepared from frozen slices with FFPE-extracted samples. This is likely due to the greater degree of RNA degradation in the FFPE-extracted samples. Since there is good correlation of expression among genes in the entire set, this reduction does not greatly affect analyses of fold changes in expression levels, especially among highexpressing genes (data not shown). Similar experiments with controlled degraded RNA suggests that good correlations with counts obtained from freshly prepared RNA can be obtained if at least 50% of the RNA is greater than 300 bp. Our results with cell and tissue extracts and unpurified RNA raised the possibility that the ncounter System could also accurately analyze FFPE samples without Figure 7. Bioanalyzer analysis of RNA extracted from flash-frozen (grey line), cryropreserved (green line) and FFPE (orange line) archived human heart samples. Figure 8. Comparison of counts of 96 genes in 100ng of total heart RNA versus 100ng total RNA prepared from frozen heart slices (green points) and total RNA prepared from FFPE heart slices (orange points).
RNA purification. To verify this, we deparaffinized FFPE slices of human heart, digested with Proteinase K, and used this crude extract in an ncounter assay with a different 96-gene CodeSet. For comparison, we performed the ncounter assay concurrently using purified total RNA extracted from the frozen tissue and the FFPE slices and normalized the counts to positive controls according to the standard protocol. The results from this experiment are shown in Figure 9. With this CodeSet, the correlation between RNA purified from flash-frozen tissue and FFPE tissue is 0.95. This result demonstrates that the correlation between intact and FFPE-extracted RNA is independent of the CodeSet used. Furthermore, the correlation between the counts obtained with the FFPE lysate and purified, intact RNA is almost identical to that seen with the FFPE-extracted and purified RNA (R 2 =0.95). Although we have not yet tested other tissues, these results suggest that crude RNA preparations containing degraded RNA from FFPE specimens can be used to quantify gene expression levels on the ncounter platform. Conclusions The enzyme-free nature of the ncounter System enables gene expression analysis from crude preparations and fragmented RNA from various sources. In this Technical Note, we demonstrated that data from cell and tissue lysates, whole blood lysate without globin mitigation, and crude FFPE extracts are highly correlated with data from matched, purified RNA samples. For analyzing large numbers of samples, eliminating the need for RNA purification provides significant savings in reagents, time and effort. That advantage, combined with its throughput of 72 samples per day and ability to multiplex over 500 probes in a single reaction, makes the ncounter System an ideal platform for gene expression analysis. References Chomczynski P, Sacchi M., Singlestep method of RNA isolation by acid guanidinium thiocyanate-phenol-chlororform extraction. Anal Biohem 162:156-159 (1987). Geiss GK et al., Direct multiplexed measurement of gene expression with colorcoded probe pairs. Nat Biotech 26(3):317-325 (2008). Li L, et al., Interference of globin genes with biomarker discovery for allograft rejection in peripheral blood samples. Physiol Genomics 32:190-197 (2008). Liu J, Walter E, Stenger D, Thach D Effects of globin mrna reduction methods on gene expression profiles from whole blood. J Mol Diag 8(5):551-558 (2006). Malkov VA et al., Multiplexed measurements of gene signatures in different analytes using the Nanostring ncounter Assay System. BMC Res Notes. 2:80 (2009). Figure 9. Counts from purified RNA extracted from FFPE human heart slices (green points) and a crude lysate from the same FFPE human heart (orange points) compared to total RNA purified from the same flash-frozen heart. NanoString Technologies, Inc. 530 Fairview Ave N, Suite 2000 Seattle, Washington 98109 (206) 378 -NANO (206) 378-6266 www.nanostring.com 2009 NanoString Technologies, Inc. All rights reserved. NanoString Technologies, NanoString, ncounter, and Molecules That Count are trademarks of NanoString Technologies, Inc., in the United States and/or other countries. Stratagene is a registered trademark of Stratagene. All other trademarks and/or service marks not owned by NanoString that appear in this document are the property of their respective owners. v.20091113