Fig 1.
Processing FFPE samples for the TempO-Seq assay.
(A) Examples of slides processed through the TempO-Seq FFPE assay. Left panel shows an H&E stained section, with the yellow box indicating the area of interest. Center left shows an expanded image of this area, demonstrating the mixed histology that would affect the data if the entire area were to be scraped and profiled. Areas identified from a stained tissue section can be scraped from an unstained, paraffinized adjacent section (right), or (if RNase-free reagents are used for staining) directly from a stained section (center right). The scraped areas in this case were ~1 x 5 mm, aligned with the focal histology of interest, and sufficient for gene expression profiling. (B) An area of interest is manually scraped from mounted FFPE sections. The tissue is added directly into 1X FFPE lysis buffer, overlaid with mineral oil, and then heated at 95°C for 5 minutes. FFPE Protease is added and the sample is incubated and manually homogenized. The processed lysate is then ready for input directly into the annealing step of the TempO-Seq assay. (C) Schematic of the TempO-Seq detector oligo annealing and ligation process.
Fig 2.
Correlation of gene expression from biological replicates of multiple tissue types.
(A) FFPE samples of three different tissue types from human, mouse, and rat were used as input into the TempO-Seq FFPE assay. Replicates were obtained by scraping adjacent areas from three different serial sections. R2 values were calculated by comparing gene expression of one section to the average of the remaining two. (B) Principal component analysis (PCA) of different tissue types of human (colon, pancreas, and prostate), mouse (breast, lung, and muscle), and rat (brain, kidney, and liver) was conducted. The first two principal components account for the majority of variance in the samples and clearly distinguish the different tissue types.
Table 1.
Coefficients of variation observed for genes expressing at a minimum level of 10, 50, or 200 counts.
Table 2.
Agreement between FFPE TempO-Seq human whole transcriptome assay data with the GTEx database rankings.
Table 3.
FFPE TempO-Seq counts for genes recognized in GTEx database as pancreas-specific or pancreas-expressing/prostate and colon non-expressing.
Fig 3.
PCA analysis of within and between patient variability.
Duplicate biological samples of matched normal and cancerous colon tissue from five anonymized patients were processed through the TempO-Seq FFPE assay. Fig 3 shows principal component analysis of all replicates, with normal tissue samples shown as circles, and cancer samples shown as triangles. Each patient is represented by one color. Despite normal biological diversity of expression, all normal samples cluster together, while cancer samples diverge widely.
Fig 4.
Expression variability in matched cancer vs. normal tissue.
Reproducibility of within-donor biological replicates (average of three technical replicates each) of (A) normal FFPE samples, patient 3 (brown symbols in Fig 3 PCA plot), (B) cancer FFPE samples, patient 3, (C) normal FFPE samples, patient 5 (purple symbols in the Fig 3 PCA plot), and (D) cancer FFPE samples, patient 5. (E) Comparison of normal and cancer samples from one patient (average of two biological replicates each). (F) Comparison of normal tissue between patients 3 and 5 (largest difference in normal tissue expression profiles), showing that differences due to normal gene diversity are far lower than those between cancer and normal. (G) Comparison between cancers of different pairs of patients (Patient 3 vs Patient 5) showing the range of differences. (H) Comparison between cancers of different pairs of patients (Patient 2 vs Patient 4), showing the range of differences that, again, exceeds those seen in normal-to-normal comparison range.
Fig 5.
Replicability of 10 mm2 and 2 mm2 inputs in the TempO-Seq FFPE assay.
Both (A) 10mm2 and (B) 2mm2 areas were scraped from rat liver. Biological replicates were generated by scraping the same area from adjacent tissue and gene expression correlation was calculated. (C) Gene expression correlation was calculated between 10 mm2 and 2 mm2 tissue sections from rat liver.
Fig 6.
Gene expression correlation in biological replicates of archived human FFPE tissue.
Human tissues were harvested in the year indicated and stored as FFPE blocks. Tissue sections were scraped and input into the TempO-Seq FFPE Human Whole Transcriptome assay. Replicates were obtained by scraping adjacent areas from three different serial sections. R2 values were calculated by comparing one section to the average of the other two.
Fig 7.
Gene expression correlation in biological replicates of rat FFPE tissues fixed for variable amounts of time.
(A) Rat liver tissue was harvested and fixed in 10% neutral buffered formalin for 24, 96, 192, and 384 hours. Samples were then moved into 70% ethanol and processed for FFPE and used as input into the TempO-Seq FFPE rat whole transcriptome assay. Replicates were obtained by scraping adjacent areas from three different serial sections. R2 values were calculated by comparing one section to the average of the remaining two. (B) Comparison between 24-hour fixation and 96 or 384 hour fixation. R2 values were calculated between averages of three biological replicates for each condition.
Fig 8.
Correlation of gene log2 fold changes between fresh and fixed MCF-7 and MDA-MDA-MB-231 cells, and correlation of TempO-Seq with RNA-Seq.
(A) Both MCF-7 and MDA-MB-231 cells from the same plate were split in half. One half was lysed directly and processed through the standard TempO-Seq human whole transcriptome assay (fresh lysate), and the other half was pelleted and fixed for FFPE embedding and sectioning to be processed through the TempO-Seq FFPE human whole transcriptome assay (fixed cell pellet). For each assay, differential expression was calculated between the two cell types. Log2 fold changes were then plotted for each assay to determine correlation of differential expression. (B) RNA-Seq was performed on RNA purified from fresh MCF-7 and MDA-MB-231 cells. TempO-Seq FFPE assay was performed on FFPE cell pellets produced from the same cell types. DESeq2 was used to determine statistically significant (padj <0.05) log2 fold changes between the different cell types and plotted to derive correlation of differential gene expression between the two methods.
Fig 9.
Gene expression correlation of human prostate FFPE samples that were unstained, deparaffinized, or deparaffinized and H&E stained.
Serial sections were scraped and used as input into the TempO-Seq FFPE assay for human whole transcriptome. Gene expression was compared among paraffinized (sectioned and mounted, but not otherwise treated), deparaffinized, and deparaffinized then stained conditions.