In this study, we have shown that FFPE tissues can be a valuable source in generating reproducible and biologically relevant proteomic profiles using reverse-phase protein arrays (RPPA).
We first evaluated a broad range of extraction conditions from FFPE tissues, including deparaffinization, rehydration, buffer components, pH conditions and temperatures. In the past, antigen retrieval techniques were used that mainly focused on applying high temperatures in appropriate buffers, facilitating the hydrolysis of formalin-induced crosslinks
. Subsequent studies hypothesized that the pH range may affect antigen retrieval and quality of immunostaining. In particular, the Tris–HCl buffer tended to produce better results at higher pH (pH = 8–9)
. Other authors have suggested that low pH around 4.0 can produce peptides resulting from acid-catalyzed aspartic acid cleavage, therefore lowering protein yield
. Addition of a reducing agent did not alter protein recovery significantly compared to pH
. Consistent with these studies, we observed that the highest protein yields were achieved using Tris–HCl containing 2% sodium dodecyl sulfate and 0.2 M glycine at pH 9.0 when heated at 100°C for 20 min, followed by incubation at 80°C for 2 h.
With respect to protein degradation, both Coomassie staining and western blot analysis revealed lower-weight products in the FFPE cell and xenograft samples, consistent with previous observations
[5, 22]. Since we initiated protein fixation immediately after harvesting, protein degradation may be the result of the duration of fixation, rather than solely due to handling prior to fixation. Our 16-hour fixation for cells and 24-hour fixation for xenografts may have led to challenges with isolating proteins from the fixed tissue, but this duration of fixation is commonly encountered in pathology laboratories. Novel extraction methods are continuously being developed to specifically address the duration of fixation in pathology laboratories, and it will be beneficial in the future, to adapt these protocols to being compatible with RPPA
. Besides the observed partial degradation, the quality of the signal was overall comparable to frozen extractions. This implies that partial protein degradation or protein modification may occur in FFPE samples, but not to the extent of significantly lowering protein yield or signal detection. This is an important point, since the downstream application, RPPA, represents a dot blot, and utilizes antibodies recognizing epitopes under denaturing conditions. This supports the notion that even partially degraded proteins will be readily detected using the RPPA approach, with little variation when compared to intact protein. Indeed, the dot blot approach of the RPPA may be even less influenced by protein degradation than the western blotting approach as RPPA only requires that the antibody epitope be intact and not the complete protein. This is supported by RPPA data which show that the average signal detected from the FFPE preparations was similar to that of the frozen counterparts (data not shown). In fact, the antibodies that exhibited marked difference between frozen and FFPE preparations across experiments (Table
2) did so without showing any bias towards a particular preparation. The observed differences in intensity for those markers between FFPE and frozen samples may be due to differential protein recovery rather than protein degradation, as suggested by previous mass spectrometry-based experiments
. This implies that some protein markers may perform better in FFPE compared to frozen or vice versa depending on the extent of protein recovery. Indeed, it may be appropriate to use other approaches such as clustering or correlations with clinical and other characteristics to define the “gold standard” rather than assuming that frozen samples represent the “truth”.
Previous studies have tested the applicability of RPPA on FFPE materials
[3, 24–27]. In this present study, we followed a systematic approach in evaluating feasibility, reproducibility and biological significance of RPPA applied on two distinct tissue types and for a wide range of protein markers. We first established a cell plug model, in order to eliminate experimental variability due to storage duration of FFPE tissue blocks, differences in fixing times, autolysis due to delayed fixing, individual differences in samples, anatomical or regional heterogeneity, and other possible sources of variability commonly occurring in human tissue FFPE preparations. The predictable patterns of protein phosphorylation according to pharmacological manipulations applied to cultured cells proved particularly useful in this study, in order to determine whether both FFPE and frozen preparations of treated cells were capable of reproducing the anticipated results. Changes in the AKT downstream effector phospho-S6 were inhibited or induced by inhibitors or growth factors to a similar degree in both FFPE and frozen cells. However, the increases induced by EGF in pAKT were not detected in FFPE samples suggesting degradation of pAKT. Nevertheless, the decrease in pAKT induced by LY294002 was faithfully captured in both frozen and FFPE preparations. A striking correlation of phosphorylation status between upstream activator and downstream effector was observed in tissues. In particular, factors in the same signaling pathway, AKT, 4EBP1 and S6, were coordinately phosphorylated in both FFPE and frozen preparations in the same patient samples interrogated (Figure
5A). Indeed, the correlation in patient samples was higher than that in the cell line samples. The usefulness of proteomic profiling of FFPE tissues in producing biologically meaningful information is further supported by the observation that global protein expression information from FFPE samples was capable of accurately categorizing renal tissues according to their underlying biology i.e. separating cancer from normal.
Using different types of tissues and cells in this study revealed a wealth of differences with respect to their proteomic profiles. As anticipated, protein markers performed markedly differently according to tissue type. Many of them were not detected according to tissue type, regardless of FFPE or frozen preparation status (Table
2). With respect to markers that demonstrated good correlation between FFPE and matched frozen samples, only 23 markers were common across all tissue sets (Figure
5A), as a testament to marked biological diversity of different cell lineages. This also suggests that it may be necessary to select different antibodies to interrogate the diverse array of tissue lineages as protein and particular phosphoprotein stability may be tissue specific.
With respect to the reproducibility of the approach, it is important to note that separate preparations, either FFPE or frozen, from the same tissue sample (replicates) performed similarly on RPPA (Figure
2). This questions the notion that tissue heterogeneity between replicate preparations may introduce critical variations in protein profiling preventing useful interpretation if laser-capture microdissection is not applied
. It is noteworthy that our replicate preparations were derived from adjacent serial sections or pieces, therefore limiting tumor heterogeneity. But unlike our replicate preparations from within the same sample, FFPE and matched frozen samples came from tissue pieces that were located a significant distance from one another, especially in the case of breast cancer samples. Thus, tissue heterogeneity may have contributed to discrepancies between FFPE and matched frozen RPPA results. This is consistent with recently published studies of spatial heterogeneity of DNA mutations
With respect to the applicability of RPPA in the clinical routine setting, it is encouraging that the RPPA-based measurements of PR correlated significantly with the immunohistochemical results of this clinically significant marker in breast tumors (Figure
FFPE tissues represent a valuable resource to conduct retrospective studies aimed to biomarker discovery and validation, in cancer as well as other diseases. The most effective approach may be to both discover and validate molecular markers on FFPE preparations to decrease variability potentially induced by tissue handling when comparing FFPE and frozen samples. In order to improve the ability to characterize proteins from tissues used for pathological evaluation, a number of steps could be considered: using a consistent and short period of fixation in formalin, limiting the duration of time between tissue collection and addition of formalin, using small tissue pieces where formalin will permeate more frequently, obtaining tissue from multiple parts of the tissue and the consideration of new fixation approaches that are under development. Clearly, FFPE tissue studies hold the promise of producing highly reproducible and meaningful data when linked to powerful high-throughput methodologies such as reverse-phase protein arrays.