Hematology team member and researcher, Dr. Fabienne Desmots-Loyer at CHU – Hospital Pontchaillou, Rennes, France was performing RNA-seq of paired patient follicular lymphoma FFPE tumor samples in early 2019 for an important study she was assigned. Using the lab’s standard FFPE RNA isolation procedure, it was only possible to extract a few nanograms of mRNA from 100 ng to 1000 ng of total RNA. An additional challenge was that the RNA that was isolated had very low RIN scores. This wasn’t unexpected for FFPE samples, but the problem was that these samples were critical for the study. To maximize the chance of success, Dr. Desmots-Loyer decided to send her samples to a trusted sequencing service provider for library construction, NGS and bioinformatics analysis. Unfortunately, the sequencing service provider advised her that they weren’t able to create her libraries due to the poor RNA quality and low amounts of sample.
Running out of options, Dr Desmots-Loyer decided to try again and sent her samples to yet another sequencing service provider, only to receive the same feedback yet again that the samples were of poor quality and that there was insufficient RNA. Caught in a dilemma, what were her options? Were the FFPE samples unusable? Was there any way to extract enough intact mRNA from the samples for RNA-seq library construction? If not, what would be the impact for her study?
Dr. Fabienne Desmots-Loyer’s dilemma is not uncommon in the RNA-seq world when working with difficult FFPE samples. Challenges continue for both RNA-seq researchers and core labs, not only when it comes to library preparation from low-yield and highly degraded samples, but also with regards to consistently obtaining a high percentage of on-target reads from those libraries for gene expression analysis. These issues stem from the nature of the preservation process itself of FFPE samples, a process which was originally developed for long-term stability of tissue samples decades before the structures of DNA and RNA were elucidated. Several attributes of the FFPE fixation process have been recognized to affect RNA quality and quantity, including the post-mortem interval (PMI), cold ischemia (the time between biospecimen removal from the body and preservation), specimen size, tissue-to-fixative ratio, fixative age, fixation temperature, dehydration, decalcification, block size, section thickness and sample storage (1). However, despite typically poor RNA integrity and low yields, FFPE samples have proven to be critical as a source of molecular gene expression data for retrospective studies.
High levels of ribosomal RNA (rRNA) is an even more important factor that can further complicate RNA-seq for all sample types, making it harder to detect low-abundance mRNA transcripts, thereby wasting valuable sequencing resources. If rRNA is not thoroughly removed from total RNA samples, sequencing efforts will be negatively impacted due to sequenced rRNA, resulting in a low percentage of gene expression reads. Highly efficient and effective rRNA removal is essential for ensuring a high percentage of on-target mRNA reads. In fact, rRNA removal is likely to be the most underestimated step for optimizing RNA-seq. Most laboratories attempt to optimize RT-PCR instead, missing the fact that effective rRNA removal can greatly improve the percentage and consistency of their mRNA reads. Often surprised at the ultralow amounts of mRNA present in their tubes when near-complete rRNA removal is achieved, investigators typically must perform additional RT-PCR cycles before library construction.
Unfortunately, commercial kits and home-brew rRNA removal methods are lengthy, inefficient and require multiple steps, leading to increased workflow times and complexity, but often with less-than-satisfactory results. For FFPE and other low-yield samples, rRNA depletion methods that include multiple transfer steps of extracted total RNA between tubes can also result in additional loss of the already low mRNA amounts present in the sample. A new technology, the QIAseq FastSelect –rRNA HMR Kit, developed by QIAGEN for human, mouse and rat specimens, and other mammalian species, incorporates just a single 10-second addition step to samples before library prep. QIAseq FastSelect has also proven to be more effective at removing greater amounts of rRNA from FFPE total RNA than other methods, and in just one-sixth of the total time.
We asked Dr. Desmots-Loyer to further comment on her current research, her RNA-seq hurdles and to describe what finally moved her to try a new and different rRNA removal method and RNA-seq library prep strategy for her follicular lymphoma FFPE samples. Here’s what she had to say:
Tell me about yourself and your current research.
“In September 2008, I began my current position as a cancer molecular biology platform engineer in the laboratory of Prof. T. Fest, at the Hematology Department at Pontchaillou Hospital of Rennes. I am involved in translational projects and, in particular, the mutational monitoring of patients with diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL) by high-throughput sequencing. I have also been a part of Prof. Fest’s “differentiation B” team at INSERM U1236, which allowed me to focus on genetic and epigenetic dysregulation leading to loss of the proliferation-differentiation balance responsible for the emergence of FL. This is where RNA sequencing is used to better define the modification of transcriptomic signatures between healthy people and FL patients, but also to compare patients’ responses to new epigenetic therapeutic approaches.”
What is the current RNA research in your lab?
“Besides our involvement with DLBCL/FL research, we are also interested in mechanisms responsible for the occurrence and development of cancers from B-lymphocytes originating from the germinal center and, in particular, the occurrence of indolent non-Hodgkin’s lymphoma (follicular lymphoma) and aggressive lymphoma (diffuse large B-cell lymphoma). These two pathologies are very heterogeneous at the molecular level, and this heterogeneity is found in treatment responses.”
“This is where RNA sequencing can be useful, not only to better define the modifications of the transcriptomes between healthy people and FL patients but also to compare patients’ responses to new epigenetic therapeutic approaches. This should help to better stratify patients before therapy and administer the best treatments when available.”
What are some of the most difficult challenges you face in your lab when doing RNA library prep or RNA-seq?
“Most of the time, we work on large retrospectives cohorts of patients from a collection of FFPE samples with varying degrees of quality, depending on the storage conditions and preservation times. We are also working on very small numbers of sorted cultured cells from tonsil or lymph nodes samples, which make our investigation difficult and challenging for gene expression analysis.”
What was the RNA-seq problem you faced that led you to consider evaluating QIAseq FastSelect? What were your RNA samples, and what was their quantity and quality? Did you use commercial technologies/kits for rRNA depletion and RNA-seq library prep? What was the result?
“In this particular study, we were performing RNA-seq of paired tumor samples (before and after an HDACi treatment) from three follicular lymphoma patients. Using our standard FFPE RNA isolation and library prep procedure, we were only able to obtain a few nanograms of mRNA from 100–1000 ng of total RNA, which also was of inconsistent quality (RINs were very low compared to the expected values for these experiments). We asked two different RNA-seq service providers to perform RNA-seq library prep and sequencing for us. However, they were unsuccessful, stating that libraries weren’t achievable with our low-quality samples. However, QIAGEN then proposed a strategy using a new rRNA removal reagent addition (QIAseq FastSelect RNA Removal Kit) before performing library prep with their QIAseq Stranded Total RNA Library Kit, that could potentially work.
Using QIAGEN’s new QIAseq FastSelect RNA Removal Kit, which involved a single addition of the QIAseq FastSelect reagent to the samples before library prep, we were able to perform the entire workflow easily and very conveniently in our own laboratory. Additionally, QIAGEN’s Technical Support Team was great in providing me with hands-on advice. After two days of preparing libraries, we proceeded with sequencing on a NextSeq platform at CHU Hospital and did the bioinformatics analysis using a custom pipeline. After STAR alignment, we observed between 65% to 85% of reads uniquely mapped to the genome with 5% to 40% of reads assigned to genes from the annotation GRCh38-Ensembl 94. Unsurprisingly, two samples that showed only 5% of mapped reads were also the ones containing less starting material.”
“I would say that with our poor-quality samples and limited quantity of those samples, I wasn’t expecting the results we obtained. In the end, the results demonstrated that we were able to cluster treated patients’ samples together with samples collected at the diagnosis based on their transcriptomes.” – Dr. Fabienne Desmots-Loyer, Researcher, Hematology Team, CHU – Hospital Pontchaillou, Rennes, France “The workflow is very simple and fast. Without this technology, I would probably not try other procedures, and I really thank QIAGEN for their support.”
“I would say that with poor-quality samples and with the limited quantity of those samples, I wasn’t expecting the results we obtained. In the end, the results demonstrated that we were able to cluster treated patients’ samples together with samples collected at the diagnosis based on their transcriptomes. The preliminary analysis of differentially expressed genes between these two clusters (treated and non-treated) identified signatures related to B-cells with notable antigen presentation and B-cell development. We also confirmed the obtained results by quantitative PCR performed on a limited set of genes, which made us confident in QIAGEN’s technology.”
Based on the performance you experienced for your FFPE samples using QIAseq FastSelect, which types of RNA samples do you anticipate using with this kit moving forward? How do you think your future RNA-seq workflows and results will be affected?
“The entire procedure, including the bioinformatics analysis part, will probably be used routinely in our lab. We are first testing this strategy on other types of samples that are not as damaged (RNA extracted from cultured cells with higher quantity but in very low quantity) but have only a low amount of starting material. One of the problems that remains is that, although we begin with identical RNA quantities (but differing qualities) for library preparation of each sample, we observed some variations in the number of reads assigned to a gene, which in terms of differential expression analysis between samples could be a problem. We are working on this problem by trying to adjust the starting sample quantity/sample quality ratio in order to standardize results. We also observed a high proportion of duplicate reads in some samples that could also be linked to the starting quantity and the number of cycles required for clean start library amplification. To try to address this, we will adjust the number of cycles used. With an adapted calibration step (quantitative PCR on a reference gene, for example) we will probably have a solid and reproducible RNA-seq workflow for samples of varying degrees of quality and quantity that we will use for future projects.”
What was the final outcome of the libraries you created and of your research studies when you used QIAseq FastSelect?
“We expect to write a publication on the transcriptomic signature of patients receiving epigenetic treatment. Indeed, we confirmed some very important genes that are re-expressed in the patients receiving treatment as published previously by us (2), but we also identified new target genes that could explain why only one of these patients had an objective response whereas the others remained non-responders. This is a very important result since the HDAC inhibitors are promising drugs for follicular lymphoma due to the high frequency of abnormalities in epigenetic factors in this lymphoma entity compared to all others entities.”
What did you learn about the importance of rRNA removal methods from this experience?
“I knew from other researchers that rRNA contamination was very problematic for performing bioinformatics analyses. In our case, with the QIAseq FastSelect strategy, we didn’t find many mitochondrial signatures or at least (analysis is still ongoing) this did not prevent the discovery of the expected B-cell signatures and other interesting signatures in the context of epigenetic treatment.“
Learn more about the QIAseq FastSelect –rRNA HMR Kit.
Need globin removal? Find out more about QIAGEN’s QIAseq FastSelect –Globin Kits here.
Interested in trying QIAseq FastSelect for yourself in your lab? Get started now! Request a free trial of QIAseq FastSelect –rRNA/Globin Kit (8) for evaluation.
Connect with our technical experts to discuss QIAseq FastSelect, ask questions or discuss custom solutions for your research.
“When I was a child, I wanted to help people and find a job connected to improving healthcare. I was thinking of possibly becoming a doctor or a nurse involved in humanitarian aid missions. Then, during my college studies, I became intrigued by cancer and AIDS and was drawn towards a profession involving research focused on the treatment of those diseases. In 2000, I received a Ph.D. degree in cellular and molecular biology from the University of Rennes in France. During my graduate studies, I identified a new stress marker that was induced during pathophysiological processes in the liver (GSTA4) and developed a lot of technical approaches in molecular biology. In 2001, I obtained a teaching position in biochemistry and molecular biology for one year at the University of Rennes, but I was very motivated to go for a research position in the USA. So, in 2002, I joined the Department of Genetics at St. Jude Children’s Research Hospital in Memphis, Tennessee, as a post-doctoral fellow for three years. During this period, I became very interested in and passionate about research thanks to St. Jude’s hospital environment, where patients, medical staff and researchers work together with the ambition to save children and teenagers suffering from aggressive cancers. My research field at that time focused on understanding processes involved in mouse tumorigenesis following inactivation of the genes involved in apoptosis. I returned to France in 2005 and was a post-doctoral fellow in Professor Michel’s team in the UMR6026-CNRS laboratory at the University of Rennes. There, I continued investigating the emergence of cancer, focusing on alteration of protein homeostasis by particular chaperones and co-chaperone proteins. In September 2008, I began my present position as a cancer molecular biology platform engineer in the laboratory of Prof. T. Fest, Hematology Department at Pontchaillou Hospital of Rennes. I am currently involved in translational projects and, in particular, the mutational monitoring of patients with diffuse large B-cell lymphoma and follicular lymphoma by high-throughput sequencing.”
References:
Bass, P.B., Engel, K.B., Greytak, S.R., and Moore, H.M. (2014) A review of preanalytical factors affecting molecular, protein, and morphological analysis of formalin-fixed, paraffin-embedded (FFPE) tissue: How well do you know your FFPE specimen? Arch Pathol Lab Med. 138 (11),1520.
Desmots, F. et al. (2019) Pan-HDAC inhibitors restore PRDM1 response to IL21 in CREBBP-mutated follicular lymphoma. Clinical Cancer Research 25(2). DOI: 10.1158/1078-0432.CCR-18-1153