Biological and histological heterogeneity of tissue samples are challenging issues for the isolation and characterization of individual cell populations that typically comprise tissues such as epithelial, mesenchymal and immune cells. Techniques such as macrodissection produce samples consisting of a mixture of cell types. Laser capture microdissection (LCM) is a method to obtain samples containing one specific cell type from complex tissues based on phenotypical, cytological and histopathological features. Generally, stained tissue slices are observed in an inverted light microscope and under visual guidance, areas of interest are cut loose using a focused laser beam. Excised fragments are catapulted into a specialized tube and protein, RNA or DNA are subsequently extracted for molecular profiling techniques such as gene-expression analysis or proteomics. The extraction of RNA from fresh frozen prostate tissue is particularly challenging due to its rapid degradation by RNases and low quantities of material usually obtained through LCM.
In this study, we describe a protocol for the isolation of high quality RNA from fresh frozen prostate tissue including tissue handling, staining and RNA extraction kits of procured RNA (Fig. 1, Additional file 1).
To obtain high quality RNA from any type of tissue, optimization of steps that precede the LCM process are essential. In order to perform LCM on tissue slides, a histological stain is needed to discern the various cell types and structures present in the tissue. A widely used stain is haematoxylin, a staining method based on the reaction of haematein with DNA [1]. Metal ions in the DNA are used a mordant, giving a blue color to the nuclei. Cresyl violet or NissI staining uses a basic aniline dye that stains RNA, and thus the endoplasmic reticulum and ribosomes blue [2]. Others have compared these histological stains for other tissue types and cell lines, but not microdissected prostate cancer (PCa) tissue [3, 4]. We have compared the RNA quality of microdissected PCa tissue stained with haematoxylin or cresyl violet (Fig. 2a). Consecutive PCa tissue sections were stained with haematoxylin (n = 5) and cresyl violet (n = 5) and matching tumor regions from the same patient were isolated through LCM. As a control, RNA was isolated from one complete tissue slice immediately after cutting. The quality of RNA was subsequently measured with the Agilent Bioanalyzer as RNA integrity number (RIN).
The quality of RNA was similar for cresyl violet (median RIN 7.4) and haematoxylin (median RIN 7.6, p = 0.71) stained sections. However, the morphology of the tissue was much better visualized in cresyl violet treated slides (Fig. 2b). Studies have demonstrated that cresyl violet staining is also preferred for RNA isolation in breast and endometrium [5, 6]. Therefore, we applied cresyl violet staining for all further optimization steps.
RNA quality is dependent on RNA extraction kit
There are several commercial RNA extraction kits available that allow the isolation of RNA from tissue such as RNeasy® Micro (RNeasy), miRNeasy Mini (miRNeasy), Arcturus® Picopure® RNA isolation kit (Picopure), mirVana™ miRNA isolation kit (mirVana) and RNAqueous®-Micro (RNAqueous). Several studies have compared different RNA extraction kits on samples other than microdissected tissue [7–9]. Additionally, the preparation and process of LCM has been optimized for various tissues, including prostate [3, 10, 11]. However, to our knowledge there have been no reports on which RNA extraction kit is most suited for the isolation of RNA from microdissected (prostate) tissue. Therefore, we sought out to determine which RNA extraction kits can provide the best quality from fresh frozen prostate tissue fragments isolated through LCM.
RNA was extracted in parallel from microdissected PCa and benign regions from three patients in duplicate using each of the 5 RNA isolation kits (total of 60 LCM samples). Matching tissue regions (PCa and benign) were microdissected from the prostate tissue with a similar surface area (7 mm2). Additionally, RNA was isolated from a whole tissue slice (10 µm) from the same patient in duplicate with each kit, as a control for potential LCM quality loss (n = 30). PCa cell lines PC3, LNCaP and DU145 served as a control for high quality RNA control during each isolation round (n = 20) [12]. The duration of the LCM was kept within a maximum of 1 h to minimize RNA degradation, although RNA quality remained stable for at least 2 h (data not shown).
First, we compared RIN values of RNA isolated with the RNeasy (median RIN 7.2), miRNeasy (median RIN 6.6), Picopure (median RIN 6.0), Mirvana (median RIN 6.5) and RNAqueous (median RIN 6.3) extraction kits, respectively (Fig. 3a). The main difference between the RNeasy and miRNeasy kits is the option in the latter to enrich for micro RNAs in the sample. The RNeasy and miRNeasy kits gave the best quality RNA and led to successful RNA isolation of every sample. Both the mirVana (2 out of 12 samples) and RNAqueous (3 out of 12 samples) kits failed more often to isolate RNA from microdissected tissue. RIN values of the RNeasy samples were significantly higher than those obtained with the Picopure (p = 0.01) samples, but the differences were not found to be significant with RNAqueous (p = 0.08), miRNeasy (p = 0.57) or mirVana (p = 0.34). However, the number of samples was too low to conclude that the RNeasy and miRNEasy kit are truly superior to the other tested RNA extraction kits. Additionally, we found that the quality of extracted RNA can vary between benign and PCa areas of the same slide (Fig. 3b). In the first patient, RNA extracted from benign tissue showed similar quality (median RIN 6.7) as that extracted from PCa tissue (median RIN 6.4, p = 0.84). However, in the second (RIN 6.0 vs 6.8) and third patient (RIN 7.5 vs 6.2), RNA isolated from benign tissue and PCa cells was significantly different (p = 0.04 and p = 0.02 respectively). This difference may be due to local changes in ribonuclease activity. Studies on the expression of RNases in tumors are contradictory, although it has been suggested that RNase activity decreases in rapidly growing tumors [13]. The amount of blood loss during prostatectomy and tumor stroma content have also been shown to affect RNA degradation [14]. However, RNA quality can also vary in general throughout tissue slices, independent of morphology.
The overall quality of RNA obtained with the RNeasy or miRNeasy kit from control cell lines (n = 9), whole tissue sections (n = 12) and benign and PCa LCM material (n = 24) was compared (Fig. 4). As expected, the average quality of RNA obtained from PCa cell lines was significantly higher (median RIN 8.3) than from LCM material (median RIN 7.0, p < 0.001). However, the LCM samples still generated high quality RNA that was comparable to RNA from whole tissue sections from the same patients (median RIN 7.5, p = 0.09). Therefore, we demonstrate that high quality RNA can be isolated from microdissected tissue with only a small quality loss (±0.5 RIN) compared to whole tissue slices. Additionally, we found that the RNA quality of whole tissue sections was a good indication on the RNA quality of microdissected material taken from the same tissue sections. We would therefore recommend to test overall tissue quality prior to LCM. To confirm that the extracted RNA was reflective of the morphological origin (benign versus PCa), we performed amplification and quantitative PCR on RNA from normal (n = 3) and PCa (n = 3) LCM (RNeasy) samples with primers directed against the metabolic enzyme alpha-methylacyl-coenzyme A racemace (AMACR) and the serine protease hepsin. These proteins are frequently overexpressed in PCa [15–17]. AMACR and hepsin were upregulated 40- and 13- fold respectively in all three tested PCa samples compared to the respective benign tissue (results not shown). Overall, these results demonstrates that RNA isolated from LCM samples with our workflow provides high quality RNA, which could be used for further downstream analyses. A complete step by step protocol for the isolation of high quality RNA from prostate tissue is included as a supplementary file (Additional file 1).
RNA quality, not quantity, can be precisely measured with the Agilent Bioanalyzer
The Bioanalyzer is a very sensitive instrument that can measure picograms of RNA material. However, this also implies that contaminants can have a major impact on the final measurements. For example, we often encountered ‘ghost peaks’ in the electropherograms generated by the Bioanalyzer that would impede the interpretation of the electropherogram. Therefore, we tested the reproducibility of RNA quality and quantity measurements with the Bioanalyzer by comparing the results obtained in duplicate analyses of the same samples (Fig. 5). RIN values measured in the same sample with different microfluidic chips were strongly correlated (Fig. 5a, r = 0.89). However, the correlation between RNA quantities was low (Fig. 5b, r = 0.68). As the quantities of RNA samples are based on the ladder of each individual microfluidic chip, variations between ladder batches may attribute to the low correlation between quantity measurements. In conclusion, RIN values are consistently measured with a Bioanalyzer, but not RNA quantity.
The isolation of high-quality RNA from LCM material is a major challenge. In this study, we provide a working protocol for the isolation of high quality RNA from fresh frozen prostate tissue (Additional file 1). We found that the use of cresyl violet as a histological stains permits the isolation of high quality RNA from LCM prostate tissue with good discriminative tissue morphology. We also demonstrated that the applied RNA extraction kit can influence the quality of isolated RNA and that the RNeasy and miRNeasy kits consistently deliver high quality samples reflective of morphological origin. Furthermore, the RNA quality can vary within the same tissue slice. Finally, we showed that the Agilent Bioanalyzer can reproducibly determine RNA quality, but not quantity. This should be taken into consideration when samples are subsequently used for experiments that require precise input quantities.