Transcriptome-Wide Gene Expression Profiles from FFPE Materials Based on a Nuclease Protection Assay Reveals Significantly Different Patterns between Synovial Sarcomas and Morphologic Mimickers

Simple Summary The following study aimed to validate and test the feasibility, reliability, technical applicability robustness, and reliability of a new commercially available transcriptome profiling assay, being performed on formalin-fixed, paraffin-embedded (FFPE) patient samples. The patients were suffering from either synovial sarcoma or spindle cell sarcoma, which are morphologically similar tumors, but differ in molecular characteristics. Transcriptome analysis of FFPE material is still challenging. However, it is the most available material in pathological routine and therefore valuable for translational research approaches. This new commercially available assay is based on a nuclease protection assay and has shown to be a feasible method for adequate transcriptome profiling with low sample input and therefore is suitable for further research of biomarkers. Abstract Background: Transcriptome profiling provides large data on tumor biology, which is particularly valuable in translational research and is becoming more and more important for clinical decision-making as well. RNA sequencing is considered to be the gold standard for this. However, FFPE material, as the most available material in routine pathology, has been an undefeatable obstacle for RNAseq. Extraction-free nuclease protection assays have the potential to be a reliable alternative method for large-scale expression profiling. The aim of this study was to validate and test the basic feasibility, technical applicability robustness, and reliability of the HTG transcriptome profiling (HTP) assay on clinical tumor samples. Methods: FFPE samples from 44 synovial sarcomas (SyS) and 20 spindle cell sarcomas (SpcS) were used. The HTP assay was performed on 10 µm thin FFPE slides. After nuclease protection in the HTG Edge Seq System, libraries were generated for sequencing on an Illumina NextSeq 500 platform. Fastq data were parsed and then analyzed by using the HTG analysis platform EdgeSeq REVEAL. Immunohistochemistry was performed to validate the expression of TLE1. Results: The technical application of the HTP Panel revealed robust and reliable results with 62 samples, and only 2 samples failed due to an incomplete digestion of gDNA. The analysis, performed at the analysis platform REVEAL, showed 5964 genes being significantly differentially expressed between SpcS and SyS. In particular, overexpression of the known marker TLE1 in synovial sarcoma could be recovered, which underlines the reliability of this system. Discussion: Transcriptome profiling gets more and more important for tumor research and diagnostics. Among other established technologies, the HTP Panel has shown to be a feasible method to get robust and reliable results. Thereby, this method needs very few sample-input by getting a success-rate of 96.88%, which indicates the upper average range, compared to other technologies working with FFPE tissue. Conclusion: The nuclease protection assay-based HTP Panel is a feasible method for adequate transcriptome profiling with low sample input and therefore is suitable for further research of biomarkers.


Introduction
Clinical evaluation of tumors creates and uses primarily formalin-fixed and paraffinembedded (FFPE) tissue [1]. Therefore, archives of FFPE tissue blocks represent an immeasurable resource for translational research. Analysis of molecular markers on FFPE materials, however, is challenging, since fixation treatment of the tissue causes degradation of nucleic acids [2] In particular, accurate transcriptome profiling on FFPE material has been an undefeatable obstacle for many years [3][4][5][6][7][8].
RNA sequencing (RNA-Seq) is the gold standard to measure whole transcriptomes of samples, but the reliability is limited with this method when FFPE samples are used. Recently, a new method for transcriptome analysis on FFPE tissue was established by HTG molecular, named the "HTG transcriptome profiling" (HTP). It is based on the Edge Seq System, which uses an extraction-free, nuclease protection assay. The analysis could be performed directly from FFPE slides in less than three days. Measuring 19,398 targets, this assay covers a large scale of relevant genes.
To test the robustness and reliability of this new method, we performed it on 64 patient samples, with 20 patients suffering from spindle cell sarcoma not otherwise specified (NOS) and 44 patients suffering from synovial sarcoma, which are characterized by the chromosomal translocation involving SS18 and either SSX1, SSX2, or rarely SSX4 [9]. These gene fusions orchestrate a specific transcriptional program, which determines morphologic appearance and biologic behavior of synovial sarcomas. There are, however, spindle cell sarcomas that share tumor morphology but behave clinically differently. Thus, we compared a genetically determined tumor entity with morphologic mimickers by applying a genome-wide transcriptional analysis.

Study Design
In this study, we aimed to test the robustness and reliability of the HTG transcriptome panel assay (HTG Molecular, Tucson, AZ, USA). Therefore, we used FFPE samples derived from 20 spindle cell sarcoma patients and 44 synovial sarcoma patients (see Figure 1). Patient samples exhibiting synovial sarcoma were pre-characterized by either Fluorescencein-situ-hybridization (FISH) or the Archer ® FusionPlex ® Sarcoma panel (Invitae, San Francisco, CA, USA), a targeted sequencing method using the RNA of patient samples.

RNA-Isolation and Quantification for Archer ® FusionPlex ® Analysis
FFPE sections were stained with hematoxylin and eosin (H&E) and regions of interest were marked by two experienced pathologists for subsequent macrodissection. Macrodissection was performed on 2-6 10 µm thick FFPE sections and RNA was isolated by using the Maxwell ® RSC RNA FFPE Kit (AS1440, Promega, WI, USA), according to the manufacturer's instructions. RNA was eluted in 50 µL RNase-free water and stored at −40 • C.
Measurement of RNA concentration was performed via fluorometric quantification (Qubit, Thermo Fisher Scientific, Waltham, MA, USA) using the RNA High sensitivity assay kit according to the manufacturer's instructions. One µL of each isolated RNA sample was applied for measurement.

RNA-Isolation and Quantification for Archer ® FusionPlex ® Analysis
FFPE sections were stained with hematoxylin and eosin (H&E) and regions of interest were marked by two experienced pathologists for subsequent macrodissection. Macrodissection was performed on 2-6 10 μm thick FFPE sections and RNA was isolated by using the Maxwell ® RSC RNA FFPE Kit (AS1440, Promega, WI, USA), according to the manufacturer's instructions. RNA was eluted in 50 μL RNase-free water and stored at −40 °C.
Measurement of RNA concentration was performed via fluorometric quantification (Qubit, Thermo Fisher Scientific, Waltham, MA, USA) using the RNA High sensitivity assay kit according to the manufacturer's instructions. One μL of each isolated RNA sample was applied for measurement.

Archer ® FusionPlex ® Targeted RNA Sequencing
The detection of fusions in the tested samples was performed by using the Archer ® FusionPlex ® Sarcoma (ARSAR) Panel (Invitae, San Francisco, CA, USA). A total of 150 ng of each RNA sample were applied for cDNA synthesis. cDNA libraries were prepared via anchored multiplex PCR according to the manufacturers' instructions (Archer ® FusionPlex ® Protocol for Illumina ® , AB0005-8-reactions), leaving out the PreSeq RNA QC assay. The Archer ® FusionPlex ® Protocol for Illumina ® includes Archer ® Universal RNA Reagent Kit for Illumina ® (Catalog #SK-0093-8), Archer MBC adapters (Catalog #SA0043), and custom designed gene-specific primer (GSP) pool kit. The quality and quantity of cDNA libraries were assessed by using the D5000 Screen Tape assay on the Agilent 2200 TapeStation system (Santa Clara, CA, USA). Libraries were pooled (8-24 samples per run) and sequenced on an Illumina MiSeq ® system (SY-410-1003, Illumina, San Diego, CA, USA).

Archer ® FusionPlex ® Targeted RNA Sequencing
The detection of fusions in the tested samples was performed by using the Archer ® FusionPlex ® Sarcoma (ARSAR) Panel (Invitae, San Francisco, CA, USA). A total of 150 ng of each RNA sample were applied for cDNA synthesis. cDNA libraries were prepared via anchored multiplex PCR according to the manufacturers' instructions (Archer ® FusionPlex ® Protocol for Illumina ® , AB0005-8-reactions), leaving out the PreSeq RNA QC assay. The Archer ® FusionPlex ® Protocol for Illumina ® includes Archer ® Universal RNA Reagent Kit for Illumina ® (Catalog #SK-0093-8), Archer MBC adapters (Catalog #SA0043), and custom designed gene-specific primer (GSP) pool kit. The quality and quantity of cDNA libraries were assessed by using the D5000 Screen Tape assay on the Agilent 2200 TapeStation system (Santa Clara, CA, USA). Libraries were pooled (8-24 samples per run) and sequenced on an Illumina MiSeq ® system (SY-410-1003, Illumina, San Diego, CA, USA).

Fluorescence In Situ Hybridization
Subsequently, FISH-probe was applied on sections and covered with cover glasses. For detection of translocations involving the FUS gene, the ZytoLight ® 2C SPEC FUS Dual Color Break Apart Probe was used (ZytoVision GmbH, Bremerhaven, Germany, Z-2130-50). Fixogum (Marabu, Tamm, Germany) was used to seal the cover glass. Slides were placed in a hybridizer (Agilent Dako, Santa Clara, CA, USA). Denaturation was performed at 75 • C for 10 min followed by hybridization at 37 • for 17-18 h.
DAPI counterstain was applied to dry specimens. Slides were covered with a cover glass and subsequently, signal enumeration was performed on the microscope (Leica DM6 B, Leica Microsystems CMS GmbH, Wetzlar, Germany.

HTG Transcriptome Panel
Transcriptome analysis was performed by using the HTG Transcriptome Panel (HTG Molecular) according to the manufacturers' instructions. Approximately 11 mm 2 of tissue tumor area from each sample (FFPE slide of 10 µm thickness) were used to perform lysis reaction. Lysis reaction was performed by using the Lysis buffer A and included an additional DNase treatment step. After sample preparation, quantitative nuclease protection assay (qNPA) was performed on the HTG EdgeSeq Processor. In the next step, protection probes that hybridized to mRNA were amplified and tagged to generate sequencing libraries. After preparation of libraries, its concentration was determined by quantitative PCR using the Kapa Library Quantification Kit (Roche, Basel, Switzerland). Libraries were sequenced on an Illumina NextSeq 500 (Illumina, San Diego, CA, USA). Within 1 run, 24 sample libraries were sequenced by using the NextSeq 500/550 High Output Kit v2.5 (75 Cycles, Illumina, San Diego, CA, USA). Sequences were parsed directly from Fastq-files using the HTG EdgeSeq Parser software (HTG Molecular). Parsed data were analyzed via the Analysis platform Edge Seq REVEAL (HTG Molecular). Median normalization and subsequent differential expression (DE) analysis was performed. For DE analysis, the test method "DESeq2" was chosen. This differential expression analysis has been completed using the DESeq2 package (version 1.30.1, HTG molecular, Tucson, AZ, USA) available from Bioconductor. The DESeq2 package provides methods for estimating and testing differential expression using negative-binomial generalized linear models. Empirical Bayes methods are used to estimate dispersion and log2 (fold change) with datadriven prior distributions. See http://bioconductor.org/packages/release/bioc/vignettes/ DESeq2/inst/doc/DESeq2.html (accessed on 26 April 2022) for more information.
No pre-filtering is applied to the data prior to analysis. The DESeq2 model corrects for library size using the median ratio method from Anders and Huber (2010). Dispersions are estimated with the Cox Reid-adjusted profile likelihood method developed by McCarthy et al. (2012). Log2 fold change is estimated via Tikhonov/ridge regularization with a zero-centered normal prior distribution with variance calculated using the observed distribution of maximum likelihood coefficients (see DESeq2 documentation for details).
DESeq2 performs independent filtering on probes prior to applying the false discovery rate p-value adjustment in order to increase power. This will cause some probes to have no p-value.

Results
We analyzed the expression of around 20,000 gene targets in a cohort of 20 patients suffering from spindle cell sarcoma (SpcS) not otherwise specified (NOS) and 44 patients suffering from synovial sarcoma (SyS), exhibiting the classifying chromosomal translocation involving SS18 and either SSX1 or SSX2. The analysis was performed and visualized by using the Reveal software. The classifying chromosomal translocation involving SS18 and either SSX1 or SSX2 could also have been detected in each SyS sample. The samples have been pre-characterized by either FISH or targeted RNA Sequencing The first step of the expression analysis was a look at the QC metrics. These are helpful to evaluate if the samples failed the QC. The software generates four plots: QC0, QC1, QC2, and QC3, which are explained in Table 1. In our analysis, two samples failed the QC3, indicating an incomplete digestion of gDNA by DNase (Figure 2). These samples were excluded from further analysis. SpcS and SyS are morphologically similar to each other, regarding the H&E staining ( Figure 3A,B). However, the analysis of all tested samples revealed a distinct gene expression pattern, as shown in the principle component analysis (PCA) plot ( Figure 3C). SpcS and SyS are morphologically similar to each other, regarding the H&E staining ( Figure 3A,B). However, the analysis of all tested samples revealed a distinct gene expression pattern, as shown in the principle component analysis (PCA) plot ( Figure 3C).

Figure 2.
QC3-Plot of the HTP analysis. Two samples failed QC3 (red dots), which means that the digestion of gDNA did not complete, as they are above the dotted line meaning the threshold. SpcS and SyS are morphologically similar to each other, regarding the H&E staining ( Figure 3A,B). However, the analysis of all tested samples revealed a distinct gene expression pattern, as shown in the principle component analysis (PCA) plot ( Figure 3C).   Figure 4A,B. All genes shown in Figure 4A,B showed p-values < 0.001. Table 2 indicates fold-change and FDR-adjusted p-values of the top 10 differentially expressed genes.
Especially expression of TLE1 showed highly significant differences (6.14 × 10 43 ) between SyS and SpcS. Since TLE1 is a well-established marker for SyS, it is an interesting observation that this marker is one of the mostly significant genes. For this reason, we immunohistochemically stained SyS and SpcS against TLE1 ( Figure 5B,C) and additionally generated a Bar plot, showing the expression of TLE1 in two samples of Sys and SpcS, respectively ( Figure 5A).  Especially expression of TLE1 showed highly significant differences (6.14 × 10 43 ) between SyS and SpcS. Since TLE1 is a well-established marker for SyS, it is an interesting observation that this marker is one of the mostly significant genes. For this reason, we immunohistochemically stained SyS and SpcS against TLE1 ( Figure 5B,C) and additionally generated a Bar plot, showing the expression of TLE1 in two samples of Sys and SpcS, respectively ( Figure 5A).  To visualize the top 10 differentially expressed genes in SpcS and SyS, a heatmap was created ( Figure 6). Yellow bars indicate SpcS samples, and blue bars indicate SyS samples. Red designates low expression, whereas blue specifies an elevated expression of genes. This heatmap visualizes the clear differences in gene expression of the 10 genes, including the established marker TLE1 for SyS. To visualize the top 10 differentially expressed genes in SpcS and SyS, a heatmap was created ( Figure 6). Yellow bars indicate SpcS samples, and blue bars indicate SyS samples. Red designates low expression, whereas blue specifies an elevated expression of genes. This heatmap visualizes the clear differences in gene expression of the 10 genes, including the established marker TLE1 for SyS.

Discussion
Transcriptome profiling makes it possible to give an overview of gene expression in a large scale and thus simplifies the search for molecular biomarkers. For this purpose, RNA sequencing is considered the gold standard [10]. However, this method has its limitations using FFPE tissue. Nonetheless, FFPE tissue is the only available material, as it is routinely used and could be easily stored at room temperature. For this reason, sequencing methodologies using FFPE tissue are constantly under optimization to improve analysis on these challenging samples. We tested the HTP panel from HTG for sensitivity, robustness, and reliability of transcriptome-wide gene expression analysis on

Discussion
Transcriptome profiling makes it possible to give an overview of gene expression in a large scale and thus simplifies the search for molecular biomarkers. For this purpose, RNA sequencing is considered the gold standard [10]. However, this method has its limitations using FFPE tissue. Nonetheless, FFPE tissue is the only available material, as it is routinely used and could be easily stored at room temperature. For this reason, sequencing methodologies using FFPE tissue are constantly under optimization to improve analysis on these challenging samples. We tested the HTP panel from HTG for sensitivity, robustness, and reliability of transcriptome-wide gene expression analysis on FFPE tissue.

Assay Performance
The extraction-free and nuclease-protection-based workflow simplified the practical handling. Additionally, with 11 mm 2 of a 10 µm tissue slide, very few sample inputs are needed. This is especially important for samples where only biopsy material is available.
Our analyses revealed genes showing distinct expression patterns in SpcS compared to SyS. Regarding Figure 3C, the expression patterns of SyS compared to SpcS differ significantly and the volcano plot ( Figure 3D) shows that 5964 genes are differentially expressed in these 2 entities. As we generally expected different gene expression patterns when comparing two different entities, these findings underline the robustness and reliability of the assay. In SyS, the most significantly result of differences in gene expression was shown for TLE1, being upregulated in SyS. This reflects the reliability of this assay, as TLE1 is an already established marker for SyS, being overexpressed in the nuclei of synovial sarcoma cells [11]. The immunohistochemical staining in Figure 5 is shown to validate the results gained by the novel nuclease-protection assay. As IHC is routinely performed in pathological diagnostic of SyS, we added this figure to underline the reliability of these results. In addition, a significantly different expression of other genes relevant for tumor progression could be found, e.g., Carnitine palmitoyltransferase 1 C (CPT1C), which promotes tumor growth and drug resistance, being reported for gastric cancer [12]. CPT1C was identified to be involved in fatty acid metabolism, due to its enzyme activity, which allows the entry of long-chain fatty acids [13]. Reilly and Mak reported unusual and enhanced expression of CPT1C in brain cancers and several sarcomas of soft-tissues and lung [14]. However, the detailed role or CPT1C in cancer remains indefinite [13]. Procollagen-lysine, 2-oxoglutarate 5-dioxygenase (PLOD) 3 was overexpressed in SpcS and is reported to promote tumor progression and poor prognosis in gliomas [15]. In addition, PLOD2 expression is elevated in SpcS. Qi and Xu reported PLODs to have enhancing effects of cell migration, invasion, and proliferation potentially by the modulation of collagen cross-link and maturation [16]. Our analyses have shown that the HTP panel is a suitable method to perform gene expression profiling on a large scale.  Table 3.
The HTP Panel of HTG Molecular has a moderate sample throughput (24 samples per 3 days), but the number of targets (19398) as well as the sequencing depth (8 M) is close to comparable methods such as BioSpyder TempO-Seq [17]. However, the costs per sample are very high in comparison to the other technologies. Nevertheless, its sensitivity and success rate of FFPE samples is the highest compared to others, considering the low RNA input needed for HTP. In addition, FFPE blocks used in this study were stored for up to five years. HTG Molecular itself reports robust results for FFPE material older than 10 years [10].