1、Storage of FFPE Specimens
1.1、切片保存温度对提取的RNA质量有很大影响
1.2、化学修饰或RNA片段化影响生成的cDNA长度
1.3、切片暴露在空气或光下会影响提取的RNA质量，蜡块保存则不会
1.4、逆转录引物的选择影响RT-PCR的检测结果
2、Influence of Fixation Time
2.1、固定时间长不直接导致RNA片段化，但RT-PCR产物的长度受影响
2.2、固定时间长可能致蜡块保存时RNA片段化速度更快
3、Influence of Specimen Size
3.1、固定时组织厚，包埋后提取的RNA质量差，效果显著
4、RNA from Pathology Samples
4.1、真实手术病理FFPE样本的RNA质量比预期差，可能与样本自溶解相关
4.2、随机引物和oligo-dT混合引物相比单独的oligo-dT引物反转后的目标cDNA更丰度，用另外的反转录酶的结果也类似
To assess the effects of storage time and conditions on the quality of RNA from FFPE samples, freshly prepared paraffin blocks containing formalin-fixed specimens from different rat tissues (liver, kidney, heart, brain, lung and spleen) were stored at different temperatures (room temperature (20–25℃), 4℃ and 37℃).
Fig. 1 shows that storage at different temperatures has a profound influence on the extent of RNA fragmentation. RNA isolated from FFPE specimens within 1–3 days after embedding was surprisingly intact, with RNA Integrity Numbers (RIN [8]) around 7 or higher. Even after 1 year storage at 4℃, ribosomal RNA bands were still clearly visible in most RNA eluates, and RIN values around 5–6 could be obtained. In contrast, RNA from blocks stored at room temperature (20–25℃) or 37℃, did not show clearly distinct rRNA bands anymore, and the mean fragment length was well below 2000 and 100 nucleotides, respectively. Samples derived from different organs did not show any significant differences in the extent of fragmentation over time.
To determine in how far storage time and conditions affect usefulness of RNA isolated from FFPE material, one-step RT-PCR amplification was performed with primer sets directed against the rat HPRT gene, producing amplicons of different lengths from about 100 to 1100 nucleotides. Notably, even with RNA isolated from FFPE sections within three days after embedding, amplification of fragments up to 700 nt was successful (Fig. 2), whereas longer amplicons (1100 nt) could not be obtained. All amplicons were produced with similar efficiency using RNA isolated from RNAlaterstabilized tissue (not shown). Since BioAnalyser data (Fig. 1) clearly show that RNA up to 4.8 kb in length (size of the 23S rRNA band) can easily be isolated from these FFPE samples, and assuming that the same is true for mRNA, the limiting factor in this case is not RNA length per se. Instead, cDNA synthesis from the isolated RNA is apparently limited by the extent of chemical modification of RNA by formaldehyde during tissue fixation. The RNA purification method used in this study includes a heating step specifically designed to reverse formaldehyde crosslinking, which improves CT values in realtime RT-PCR by as much as 5–6 cycles compared to samples isolated
without this step (data not shown).
Exposure of individual FFPE sections to light and air had a negative influence on RNA quality, but storage of entire FFPE blocks either in the open or protected from air and light did not influence the effect of storage time on RNA quality, as long as the uppermost section was discarded at each time point, and sections not directly exposed to air were used for RNA preparation (data not shown).
In real-time PCR applications, amplicons are usually shorter than 200 nt. However, in cases where oligo-dT priming is used for cDNA synthesis, the distance between the 3' end of the mRNA and the 5' end of the PCR forward primer becomes the limiting factor, because shorter cDNA products may not contain the entire region to be amplified. Consequently, RT-PCR amplicons should be within ,500 nt or less from the mRNA 3' end when oligo-dT priming is used for cDNA synthesis. This is not a concern when random or gene-specific primers are used for cDNA synthesis. Similar considerations regarding cDNA length and distance from the 3' prime end apply to other workflows that rely on cDNA synthesis, such as microarray hybridizations, in cases where oligo-dT priming is used.
Another critical factor for nucleic acid quality from FFPE samples are the conditions used for formalin fixation. It is crucial to keep the time between sample acquisition and fixation as short as possible in order to avoid tissue autolysis and nucleic acid degradation by endogenous nucleases. In addition, overfixation can become an issue, particularly if fixation proceeds for considerably longer than 24 hours, resulting in more irreversible crosslinks [3–5]. There are also reports in the literature that fixation by formaldehyde, particularly over extended time periods, results in increased RNA degradation [9].
To assess the influence of overfixation on RNA quality, we compared tissue samples fixed either overnight or for 72 hours in formalin, compared to RNAlater-stabilised tissue samples from the same animal. BioAnalyser data show that even after 72 h fixation in formalin, the RNA is only marginally fragmented, and ribosomal RNA bands are still largely intact (Fig. 3), thus clearly showing that the reaction with formaldehyde itself does not cause nucleic acid fragmentation. In contrast, RT-PCR results demonstrate that prolonged fixation aggravates the negative effects of formaldehyde modification on cDNA synthesis. Whereas amplicons up to 800 nt can be routinely amplified from samples fixed over night, the maximum amplicon length for samples fixed for 72 hours in our PCR system was around 400–600 nt in most cases (Fig. 3).
While overfixation did not directly result in increased RNA fragmentation, our preliminary data indicate that overfixation may result in faster RNA fragmentation during storage of FFPE blocks (data not shown).
With the exception of small biopsies, penetration of the tissue by the fixative is the rate-limiting step in fixation, with initial penetration rates for formalin around 1 mm per hour. Penetration rates decrease with depth. Thus, around 8 hours are required for penetration of 5 mm tissue [10]. Therefore, it is generally recommended to keep sample thickness around or below 5 mm for efficient and even fixation, in order to avoid the risk of overfixation at the periphery, and tissue autolysis near the center of the specimen [7,11].
We examined the effect of specimen size and thickness by comparing rat kidney and brain samples that were either fixed whole (about 1 cm thick), or cut into smaller pieces, around 3–4 mm thick (Fig. 4). Analysis of RNA purified from these FFPE samples within days after embedding showed strong RNA fragmentation when entire organs where fixed, compared to relatively intact RNA obtained from thinner pieces. The observed fragmentation also resulted in shorter maximum amplicon size in our RT-PCR system. In the electropherograms of RNA from the large specimens, a small amount of longer RNA is still visible (Fig. 4, arrows). Presumably the respective RNA originates from the outer parts of the specimens which were penetrated by the fixative more quickly. The larger amount of these longer RNA molecules in the brain sample is also reflected in longer maximal amplicon size of 600 nt, compared to 400 nt in kidney (Fig. 4).
The degradation effect observed here is surprisingly strong, considering that the thickness of the intact rat organs used was around 1 cm only. It is possible that the intact organs are more slowly penetrated compared to equally sized pieces cut from larger tissues.
To compare our results from FFPE specimens obtained from animal tissues under strictly controlled conditions with real-life pathology samples, FFPE specimens from surgically removed lung carcinoma were obtained (see Methods). The FFPE samples had been stored for 1, 2, 4, 7, and 10 years at ambient temperature prior to sectioning. RNA preparation was done using the same purification protocol as for the rat samples above. RNA yields from a single 10 mm section were between 2.5 and 16 mg total RNA, with no correlation between age of sample and RNA yield. RNA integrity was checked by capillary electrophoresis (Fig.5a). According to the electropherograms, mean size of RNA fragments isolated was between 220 and less than 100 nt, also with no correlation between age of sample and mean fragment size. Even for the 1 year-old specimens, RNA integrity was significantly inferior to the rat samples after 1 year storage at 37℃ (Fig. 5a, and Fig. 1). The reason for this is most likely tissue autolysis, because thickness of the cancer specimens used for fixation was generally more than 2 cm, which implies more than 24 hours for penetration of the sample by the formalin [10].
To determine the usefulness of such samples for molecular analysis, real-time RT-PCR was performed on the RNA samples. Different primer pairs directed against the human housekeeping gene TBP mRNA were used to amplify fragments at different distances from the 3' end (128, 227, and 428 nt). For cDNA synthesis, the QuantiTect RT kit (QIAGEN) was used, either with a primer mix composed of random and oligo-dT primers (as contained in the kit), or, for comparison, with oligo-dT primer. Using the primer mix, all of the 1- and 2-year old samples and one of the 4-year samples produced threshold cycle (CT) values of 29–35 (Fig. 5b), whereas the remaining samples gave higher CT values, and in some cases failed to produce any meaningful signals within 40 PCR cycles. Although no clear correlation between RNA fragmentation and CT value was observed, the most fragmented RNA sample (4yr –b, lane 6 in Fig. 5a) gave the poorest PCR results. In contrast, using oligo-dT priming for cDNA synthesis resulted in
roughly comparable CT values for the amplicon located at 127 nt from the 3' end, but significantly higher values for the more distant amplicons (Fig. 5b). Only 2 samples produced meaningful results consistently for the amplicon located 428 nt from the 39 end. Using a different reverse transcriptase for cDNA synthesis produced similar results (data not shown).