development and validation of a novel protein extraction methodology for quantitation of protein...

Journal of PathologyJ Pathol 2009; 217: 497–506Published online 1 December 2008 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/path.2504

Original Paper

Development and validation of a novel protein extractionmethodology for quantitation of protein expression informalin-fixed paraffin-embedded tissues using westernblottingNiroshini J Nirmalan, Patricia Harnden, Peter J Selby and Rosamonde E Banks*Clinical and Biomedical Proteomics Group, Cancer Research UK Clinical Centre, Leeds, UK

*Correspondence to:Rosamonde E Banks, Clinicaland Biomedical ProteomicsGroup, Cancer Research UKClinical Centre, Leeds Institute ofMolecular Medicine, St James’sUniversity Hospital, BeckettStreet, Leeds LS9 7TF, UK.E-mail: [email protected]

No conflicts of interest weredeclared.

Received: 23 August 2008Revised: 30 October 2008Accepted: 21 November 2008

AbstractThe development of efficient formaldehyde cross-link reversal strategies will make the vastdiagnostic tissue archives of pathology departments amenable to prospective and retro-spective translational research, particularly in biomarker-driven proteomic investigations.Heat-induced antigen retrieval strategies (HIARs) have achieved varying degrees of cross-link reversal, potentially enabling archival tissue usage for proteomic applications outsideits current remit of immunohistochemistry (IHC). While most successes achieved so far havebeen based on retrieving tryptic peptide fragments using shot-gun proteomic approaches,attempts at extracting full-length, non-degraded, immunoreactive proteins from archivaltissue have proved challenging. We have developed a novel heat-induced antigen retrievalstrategy using SDS-containing Laemmli buffer for efficient intact protein recovery fromformalin-fixed tissues for subsequent analysis by western blotting. Protocol optimizationand comparison of extraction efficacies with frozen tissues and current leader methodologyis presented. Quantitative validation of methodology was carried out in a cohort of matchedtumour/normal, frozen/FFPE renal tissue samples from 10 patients, probed by western blot-ting for a selected panel of seven proteins known to be differentially expressed in renalcancer. Our data show that the protocol enables efficient extraction of non-degraded, full-length, immunoreactive protein, with tumour versus normal differential expression profilesfor a majority of the panel of proteins tested being comparable to matched frozen tissuecontrols (rank correlation, r = 0.7292, p < 1.825e-09). However, the variability observed inextraction efficacies for some membrane proteins emphasizes the need for cautious inter-pretation of quantitative data from this subset of proteins. The method provides a viable,cost-effective quantitative option for the validation of potential biomarker panels through arange of clinical samples from existing diagnostic archives, provided that validation of themethod is first carried out for the specific proteins under study.Copyright 2008 Pathological Society of Great Britain and Ireland. Published by JohnWiley & Sons, Ltd.

Keywords: formalin-fixed paraffin-embedded (FFPE) tissue; western blotting; renal cellcarcinoma; tissue biomarkers; heat-induced antigen retrieval (HIAR); biomarker discovery;proteomics

Introduction

Advances in proteomic and mass spectrometry-basedtechnology have fuelled a vast expansion in trans-lational research directed at analysing global proteinexpression, particularly in the context of biomarkerdiscovery and validation [1]. The potential impact ofnew diagnostic, prognostic and therapeutic biomarkerson the management of clinical diseases cannot be over-estimated [2,3]. However, numerous technical, clinicaland analytical challenges need to be addressed beforethe full potential of these approaches is realized [4].

Many biomarker-based studies are heavily relianton the acquisition of clinical tissues from a widerange of diseases in the optimal frozen form [5].The challenges inherent in procuring/storing the largenumbers of frozen tissues requisite for proteomicbiomarker investigation, particularly in rarer clinicaldiseases, highlight the need for an alternative tissueresource [1,4,6,7]. Formalin-fixed paraffin-embedded(FFPE) tissue archives represent the largest clinicallyannotated human bio-specimen resource. However,extensive formaldehyde-induced cross-linking of pro-tein makes archival tissues intractable to analyses byroutine proteomic extraction and profiling methods

Copyright 2008 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.www.pathsoc.org.uk

498 NJ Nirmalan et al.

[8]. Recent advances in heat-induced antigen retrieval(HIAR) strategies developed initially for immunohis-tochemistry (IHC) have afforded varying degrees ofcross-link reversal, potentially enabling such tissueto be used for proteomic biomarker investigations[9–13]. However, this field still requires substantialprotocol refinement and standardization to ensure opti-mal extraction, identification and quantitation of pro-teins [7,14].

Early studies reported enhanced immunohistochem-ical staining for FFPE tissue blocks exposed to hightemperature in the presence of SDS-containing buffers[15]. Such concepts were subsequently successfullyapplied to extract DNA, RNA and, more recently, pro-teins from FPPE tissue [16–19]. The two principallines of investigation pursued for proteins were a high-throughput shotgun proteomics approach for trypticpeptides and a more challenging intact protein retrievalstrategy [6,10]. Most studies have employed the for-mer approach but determining differential expressionremains a challenge, due to the lack of a global quan-titative methodology [10,20–22].

Conditions requisite for efficient protein cross-linkreversal and restoration of immunoreactivity havebeen identified as high concentrations of SDS andexposure to high temperature and pH [9]. However,the prolonged high temperature extraction regimesnecessary for cross-link reversal result in degrada-tion/heat induced scission of proteins, and initialattempts at extracting intact proteins from FFPE tis-sues have met with little success [8]. Using 2% SDSin RIPA buffer was reported to improve intact pro-tein extraction, enabling analysis of cyclinD1 andCDK2 from FFPE colorectal carcinoma tissue bywestern blotting [23]. However, comparison of gelpatterns for matched frozen and FFPE tissues var-ied widely, with the high degree of protein frag-mentation leading to intense lower molecular weightbanding [23]. Recently, Becker et al published a tar-geted quantitative approach using a commercial Qpro-teome kit (Qiagen, Germany) to extract full-length,

immunoreactive proteins from formalin-fixed tissues.The HIAR methodology, with immunoblotting andreversed-phase protein microarrays allowed quantifica-tion of HER2 in breast tissue, a key target for antibody-based treatment of cancer [6]. However, commercialrestrictions over buffer composition details togetherwith cost issues limit widespread use and limit proto-col refinement or development.

We have developed a novel cost-effective protocolbased on Laemmli buffer [24] for the efficient extrac-tion of full-length, non-degraded, immunoreactive pro-teins from FFPE tissue allowing biomarker validation.Method optimization, reproducibility and comparativeextraction efficacy in fresh frozen and FFPE tissuesare all described prior to illustration of quantitativevalidation of the methodology, using selected proteinsand examining differential expression in renal cell car-cinoma (RCC). Data comparing this methodology withthe current leader methodology are also presented [6].

Materials and methods

Reagents

General chemicals were purchased from Sigma (Poole,UK), VWR (Leicester, UK), and MP Biomedicals(London, UK). Complete mini protease inhibitortablets (Roche, UK); Hybond membrane (GE Health-care, UK); Super-Signal West Dura Substrate (Pierce,Rockford, USA); Qproteome FFPE Tissue Kit (Qia-gen, UK); EZQ Protein Quantitation Kit (Molecu-lar Probes, UK). Antibodies used are tabulated inTable 1 with optimal concentrations based on titrationsto determine linear range.

Processing of clinical samples

For protocol development, frozen and FFPE tis-sue from matched pairs of RCC and normal renalcortex collected specifically for proteomics were

Table 1. Primary and secondary antibodies used in the study

Antibodies Source Type Clone Dilution Size (kDa)

β-Actin Sigma Mouse AC15 1 : 400 000 42β-Dystroglycan BD Mouse 1 : 500 43Carbonic anhydrase IX Gift from E. Oosterwijk is acknowledged Mouse M75 16.25 ng/ml 50Fascin Santa Cruz Mouse 55k-2 1 : 1000 55GapDH Abcam Mouse mAbcam 9484 0.5 µg/ml 36Golgin Abcam Mouse 26 0.5 µg/ml 84Heat shock protein 70 Bioquote Mouse C92F3A-5 0.05 µg/ml 70Heat shock protein 60 Santa Cruz Goat 1 : 20 000 60NaK ATPase (α1-subunit) Novus Biologicals Mouse 464.6 0.4 µg/ml 115NaK ATPase (β1-subunit) Abcam Mouse 464.8 1 : 2500 35–55Thymidine phosphorylase BD Mouse P-GF-44C 1 : 1000 55β-Tubulin Abcam Mouse 3 µg/ml 55Ubiquitin core-1 Molecular Probes Mouse 16D10 1 : 8000 49Ubiquitin core-2 Molecular Probes Mouse 1312 1 : 8000 47Anti-mouse secondary Dako Cytomation Mouse–HRP NA 1 : 100 NAAnti-goat secondary Dako Rabbit–antigoat–HRP NA 1 : 2000 NA

J Pathol 2009; 217: 497–506 DOI: 10.1002/pathCopyright 2008 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Targeted quantitation of protein expression in formalin-fixed archival tissues 499

used (n = 3). Tissues excised immediately follow-ing nephrectomy were placed in ice-cold RPMImedium containing protease inhibitors. Tumour blocks(∼5 mm3) were selected from macroscopically viabletumour away from areas of haemorrhage. Tissueblocks were either frozen in liquid nitrogen forembedding in optimal cutting temperature embeddingmedium (OCT) or fixed overnight in 10% neutralbuffered formaldehyde, dehydrated in graded ethanols,followed by permeation in xylene and paraffin embed-ding. For subsequent validation, a further cohort ofmatched samples (normal renal/RCC tissue) from 10patients was used, ie 20 frozen and 20 FFPE blocks.The frozen blocks were from our research tissuebank and the FFPE blocks were from the correspond-ing patient samples held in pathology archives andincluded areas of haemorrhage, necrosis and periph-eral normal renal tissue. All samples had been storedfor <2 years to minimize possible variation introducedby long-term storage. Informed consent was obtainedfrom patients and the research was approved by thelocal research ethics committee.

Protein extraction from tissue

For each extraction, three serial 10 µm sections(∼25 mm2 each) of fresh normal renal/RCC tissuewere collected into an Eppendorf tube. Flanking 5 µmsections were routinely stained with haematoxylin andeosin (H&E) to verify tissue integrity. For fresh frozentissues, 150 µl Laemmli buffer (100 mM Tris–HCl,pH 6.8, 2% w/v SDS, 20% v/v glycerol, 4% v/vβ-mercaptoethanol) was added prior to heating at100 ◦C for 5 min [24]. Samples were cooled for 5 minon ice, centrifuged at 14 000 × g for 15 min and super-natants stored at −20 ◦C.

FFPE tissues were deparaffinized in 1 ml xylene(2 × 5 min), rehydrated in a graded ethanol series(100%, 90%, 70%, 5 min) and centrifuged to removeexcess ethanol. 150 µl freshly prepared Laemmlibuffer was added and tubes sealed with plastic clips(Qiagen, Germany) prior to heating at 105 ◦C for20 min. The samples were cooled for 5 min on iceand stored at −20 ◦C. Protein estimation was car-ried out using the EZQ Protein Quantitation Kit, withovalbumin used to generate standard curves. For par-allel comparison, extraction was also carried out withthe Qproteome FFPE Tissue Kit (Qiagen, Germany)according to manufacturer’s instructions and involvingextraction in proprietary buffer by HIAR at 100 ◦C for20 min, followed by 80 ◦C for 2 h. As the buffer com-ponents were unknown, compatability with the EZQprotein assay could not be determined and accord-ingly protein concentrations were determined using themicroBCA protein assay (Pierce, USA), as detailed inthe Qproteome kit.

Protocol optimization

Extraction buffers assessed in preliminary experi-ments included modified RIPA buffer pH 7.6 [23],

ACN–ammonium bicarbonate [25], Qproteome (Qia-gen) and Laemmli extraction buffers. Further studiesoptimizing temperature regimes (Figure 1), repro-ducibility, etc., were confined to the latter two buffers,which showed maximum efficacies.

To define technical reproducibility of methodology,serial 10 µm sections from a normal renal tissue FFPEblock were collected alternately into separate Eppen-dorf tubes to minimize variation due to tissue hetero-geneity. To compare the Laemmli and Qproteome pro-tocols, replicate extracts from the same normal renalFFPE block were loaded using a volumetric titrationequivalent to 2.5, 5 and 10 µg FFPE extract against a5 µg matched control of fresh tissue extract.

For the quantitative validation of methodology,tumour/normal differential expression profiles forseven proteins were compared by western blottingof extracts from frozen (n = 20) and FFPE (n = 20)matched normal and malignant renal tissue samplesfrom 10 patients with RCC. Proteins selected onthe basis of known degrees of differential expressionwere carbonic anhydrase IX (CAIX), GAPDH, thymi-dine phosphorylase, ubiquinol–cytochrome c reduc-tase core protein-1 and core protein-2, and HSPs60 and 70 [26–31]. Normalization of gel loadingbetween samples based on estimations of total proteinwas shown to be unreliable when probed with west-ern blotting, and an actin-based titration (describedbelow) was employed. The FFPE Laemmli workflowwas also tested on a wider panel of proteins, includ-ing β-tubulin, fascin, β-dystroglycan, golgin-84 andNaK-ATPase for assessment of extraction efficacy andspecificity.

Actin-based normalization

To ensure that equal concentrations of full-length pro-teins from fresh and FFPE extracts were loaded ontogels, a titration based on actin levels was carriedout for each patient sample. 5 µl fresh normal cor-tical/tumour extract was titrated against a doublingseries of matched FFPE extract and loading ratios forwestern blotting calculated from densitometric assays.Loading concentrations were within the linear densit-ometric range deduced for actin.

1D SDS–PAGE gel electrophoresis and westernblotting

Extracted lysates were resolved using 10% 1D SDS–PAGE in a Bio-Rad Mini Protean II system (Herts,UK) and transferred onto nitrocellulose [32]. Mem-branes were blocked and probed with antibodies priorto detection using the EnVision+ system (Dako, UK)[33], and developed using Super-Signal West-DuraSubstrate (Pierce) and Kodak Biomax film. Densito-metric scans of the blots as 12-bit images were anal-ysed with ImageQuant software (GE Healthcare, UK).For titration experiments designed to select the nor-malizing protein, blots re-probed for actin or tubulin



Figure 1. Representative example of the optimization of the FFPE Laemmli extraction protocol in normal renal corticalformalin-fixed tissue. Laemmli 1, no boiling; Laemmli 2, 105 ◦C for 10 min; Laemmli 3, 105 ◦C for 20 min; Laemmli 4, 105 ◦C for20 min, 80 ◦C for 20 min. Matched controls of FFPE tissue extracted using the Qproteome extraction buffer (Qiagen), 100 ◦C for20 min, 80 ◦C fo 2 h and fresh tissue are also shown. Sample loading was at 5 µg/lane. Western blots were probed with actin andHSP70 antibodies in duplicate experiments. Fresh renal cortical tissue extracted using Laemmli buffer at 100 ◦C for 5 min wasloaded at 2.5 µg/lane

were incubated for 30 min at 50 ◦C in stripping buffer(2% SDS, 100 mM β-mercaptoethanol, 50 mM Tris,pH 6.8) and washed overnight in TBS-T prior to re-probing.

SELDI–TOF analysis

For comparison of extraction efficacy in fresh andformalin-fixed renal tissue using the Laemmli proto-col, 100 µl extract was acetone-precipitated and thepellet resuspended in 100 µl 50 mM NH4HCO3. Sam-ples were digested in trypsin (Promega, UK) at anenzyme : substrate ratio of 1 : 50, at 37 ◦C overnight.The digest was dried in a speedVac (Thermo Electron,UK) prior to resuspension in 100 µl 5% acetonitrile,0.05% TFA. 1 µg protein extract was spotted on to aNP20 chip (Ciphergen Biosystems Inc) and overlaidwith 2 µl matrix (50% acetonitrile, 0.05% TFA, satu-rated sinnapinic acid) for acquisition of SELDI–TOFspectra in a Ciphergen ProteinChip System.

Results

Method optimization for extraction and westernblotting of FFPE proteins

Comparison of HIAR regimes and buffer combina-tions found optimum intact protein extraction withstandard Laemmli buffer as assessed by western blot-ting (Figure 1). Maximal total protein concentrationswere observed when β-mercaptoethanol was added

just prior to use. In contrast to previous reports rec-ommending extended heat treatment [9], this bufferpermitted contraction of the HIAR regime to 20 minat 105 ◦C. Extension of this by a step of 20 min at80 ◦C also resulted in good extraction but the formerwas selected, as some variation was observed betweenproteins in their ability to withstand high temperaturesfor longer periods (Figure 1). The contracted HIARregime minimized heat-induced degradation of pro-teins, generating comparable gels patterns in frozenand FFPE tissues, although the latter appeared tobe masked by heavy background staining (Figures 1,2B) which was not removed by acetone precipitation(Figure 2B).

For analysis of reproducibility, three independentreplicates of FFPE normal renal tissue extractionproduced protein concentrations of 2.01, 1.8, and1.78 µg/µl and when probed for β-actin in equalvolume loads showed similar densitometric results(mean % volume 3062.33, SD 118.7; Figure 2A).

Comparison of extraction of frozen and FFPEtissues

Densitometric quantitation of FFPE extracts using theLaemmli protocol showed an average 63% (SD ±4.72) extraction efficacy compared to matched freshfrozen tissues in three independent technical repli-cates when titrated using actin (Figure 3A, Table 2A).SELDI–TOF profiles of tryptic digests showed largelycomparable, although not identical, patterns (Figure



Figure 2. (A) Actin western blots of three technical replicate samples (FFPE A, B and C) extracted from a normal renal corticaltissue block, collected specifically for proteomics (3 × 10 µm, 25 mm2). Uniformly intense blotting patterns with comparabledensitometric ratios were observed for the replicates (mean % densitometric volume of 3062.33, SD ± 118.7). Samples wereextracted in 150 µl of the FFPE Laemmli extraction buffer and equal volumes (5 µl) electrophoresed on 10% acrylamide gels forwestern blotting. Densitometric data were generated using ProQuant software on blots scanned in the Personal Densitometer SI.(B) Coomassie-stained gel images of matched fresh and FFPE extracts before and after acetone precipitation

Figure 3. (A) Comparative actin western blots of extracts of matched renal cortex from fresh and FFPE tissues (3 × 10 µm,25 mm2). Densitometric data from a triple replicate experiment comparing the Laemmli FFPE protocol with fresh tissueextract loaded at 5 µg/lane are presented in Table 2A. (B) Corresponding SELDI–TOF spectra derived from tryptic digests ofacetone-precipitated fresh and FFPE extracts

Table 2A. Comparison of the extraction efficacy of Laemmli protocol to frozen and FFPE tissue extracts. Results are thedensitometric volumes of actin and extraction (%) of FFPE relative to frozen tissue

Experiment 1 Experiment 2 Experiment 3 Average SD

Frozen 3446.46 3314.47 4033.57 3598.16FFPE 2367.46 2034.1 2413.88 2271.81Extraction (%) 68.69 61.37 59.844 63.30 4.729

3B). In standard 150 µl Laemmli buffer extractions(3 × 25 mm2, 10 µm sections) from FFPE renal tis-sue, ∼2 µg/µl protein was obtained. Extracts fromcomparative volumes of fresh frozen tissue showedprotein concentrations of ∼1.4 µg/µl. Interestingly,western blotting experiments with on-gel loading nor-malized to total protein concentrations gave lowerdensitometric readings for FFPE extracts comparedwith fresh tissue extracts — this discrepancy is likelyto be due to a small degree of residual cross-linking and/or protein loss due to fragmentation/chain

scission. Protein estimations on FFPE extracts beforeand after centrifugation using 3 kDa cut-off filters(data not shown) showed no significant differences,suggesting minimal heat-induced fragmentation.

Normalization of intact protein loadingNormalized protein loading for comparing tumour/normal expression profiles in fresh and FFPE extractswas achieved using an actin-based titration experi-ment. Initial optimization showed comparable repro-ducible results for both actin and tubulin. Although



labour intensive, necessitating separate titrations forevery patient sample screened, the method enablesreproducible loadings crucial for the quantitation ofprotein expression.

Comparison of the FFPE Laemmli and Qproteomeprotocols

Data comparing the extraction efficacies of theLaemmli and Qproteome buffers are depicted ina duplicate titration experiment comparing the twomethodologies to a fixed 5 µg load of matchedfrozen extracts (Figure 4A). Significant improvementsin extraction efficiency were reproducibly observedwith the Laemmli protocol, based on densitometricassessments of actin (Figure 4B, Table 2B).

Figure 4. (A) Comparative actin western blot images formatched renal FFPE tissues extracted using the FFPE Laemmliand Qproteome extraction protocols. Volumetric titrations foreach protocol are compared to a fixed frozen tissue load of5 µg/lane. (B) Deduced average fold changes for each methodcompared to the fresh frozen controls. Densitometric data forthe duplicate experiment are presented in Table 2B

Methodology validation — quantitation ofdifferentially expressed proteins in RCC tissue

For each patient sample screened, 5 µl fresh nor-mal/tumour extract was titrated against 2.5, 5 and 10 µlof the corresponding FFPE normal/tumour extractusing western blotting for actin (Figure 5A) and cal-culated normalizing ratios verified prior to probing forthe panel of proteins (Figure 5A, B). Fold changesin expression (tumour/normal) were calculated foreach protein in both fresh and FFPE tissue extracts(Figure 5c). Composite data for all 10 patient samples(70 data points) showed preservation in directionalregulation for most proteins analysed and a signif-icant correlation between the results obtained fromfrozen and FFPE tissue (r = 0.7292, p < 1.825e-09;Figure 6A). Some variation was noted in the intensi-ties of expression between fresh and FFPE tissue andpatient datasets with maximum and minimum variationare presented in Figure 6B.

Membrane protein extraction

In addition to the initial panel of proteins, the FFPELaemmli protocol was assessed on a wider groupof proteins. While the method offered a sensitiveand specific extraction approach for non-membraneproteins (tubulin, fascin), membrane protein detec-tion varied significantly. Single spanning integralmembrane proteins β-dystroglycan and CAIX weredetected reproducibly with isoform specificity. How-ever, golgin 84, a single spanning membrane proteinwith an extensive coiled coil domain, and the multi-spanning NaK–ATPase (α1- and β1-subunits) failedto be detected in FFPE tissue (Figure 7A). Furtherstudies examining NaK–ATPase subunits in frozentissue extracts before and after exposure to high tem-perature showed marked aggregation of the protein tohigh molecular weight species (Figure 7B).

Discussion

The broad objective of this study was to developan efficient and cost-effective workflow to extractfull-length, non-degraded, immunoreactive proteinsfor western blot assays from FFPE tissue, permit-ting sensitive quantitation of protein expression and

Table 2B. Comparison of the Laemmli and Qproteome protocols based on densitometric measurements of actin in extracts,expressed as extraction efficacy (%) compared with 5 µg fresh tissue controls

Laemmli Experiment 1 Experiment 2 Average SD

Dilution 1 (2.5 µg) 34.73 35.99 35.36 0.88Dilution 2 (5 µg) 65.30005 80.006 72.65302 10.39868Dilution 3 (10 µg) 161.7279 121.394 141.5609 28.52038

Qproteome Experiment 1 Experiment 2 Average SD

Dilution 1 (2.5 µg) 22.31555 18.0601 20.18783 3.009062Dilution 2 (5 µg) 45.86236 38.98869 42.42553 4.860421Dilution 3 (10 µg) 122.9963 116.2452 119.6208 4.773703



Figure 5. (A) Western blots of the actin titration experimentfor an individual patient sample. Fresh frozen normal or tumourextract (F; 5 µl) is titrated against a three-point doubling seriesof the corresponding FFPE extract to deduce normalizingratios. (B) Normalized loading for actin in western blots.(C) Differential expression profiles in fresh and FFPE tissueextracts probed with antibodies against a preselected panel ofRCC tissue markers. FN, fresh frozen normal; FT, fresh frozentumour; FFPE-N, formalin-fixed paraffin embedded normal;FFPE-T, formalin-fixed paraffin-embedded tumour

facilitating use of this resource in proteomics stud-ies [34]. We report the development and quanti-tative validation of an intact protein identificationstrategy for FFPE tissues enabling significant improve-ments in extraction efficacies, as compared withcurrently published leader methodology [6]. Experi-ments using the optimized FFPE Laemmli workflowdescribed (Figure 8) achieved up to ∼63% extrac-tion, as compared with matched fresh tissue and

significant improved extraction efficacies when com-pared with commercial reagents. Covalent cross-linkreversal and stabilization of extracted FFPE pro-teins from the powerful detergent/denaturant/reducingproperties of SDS/β-mercaptoethanol enabled the con-traction of the conventional heating regimes, min-imizing heat-induced scission and degradation ofproteins.

While the protocol presented here offers signifi-cant improvements for FFPE protein extraction, esti-mation of total protein concentrations in extractedFFPE tissue can be inaccurate, as both the reversedfull-length protein and small percentage of resid-ual cross-linked protein are included. In protocolsadopting the longer two-tiered heating regime, theerror is compounded by fragmented proteins. For nor-malizing between samples, the actin-based titrationused in this study, although labour intensive, provedeffective, although care must be taken with selec-tion of the normalizing protein to ensure that it doesnot vary in expression with disease state for exam-ple. Larger validation studies would benefit from areverse-phase protein microarray approach describedby Becker et al for normalization at higher through-puts [6,35].

Expression profiles for the selected panel of pro-teins in RCC tissue showed comparable expressionpatterns between fresh and FFPE tumour/normal pairswith preservation in directional regulation for mostproteins (up-regulation in RCC for thymidine phos-phorylase, carbonic anhydrase IX and GAPDH, down-regulation in RCC for ubiquinol–cytochrome core-1and core-2 proteins and HSP 60). HSP70 expres-sion showed variation in regulation between fresh andFFPE extracts with minimal fold changes, suggest-ing unchanged tumour/normal profiles. The variabilityin intensities of expression observed between somefresh and FFPE lysates could be due to the pres-ence of necrosis, haemorrhage and normal tissue in theFFPE blocks intended for diagnostic purposes, sinceminimal variation was seen in frozen tumour blocksspecifically collected for proteomics. Laser capturemicroscopy, needle microdissection or guided proteinextraction strategies could overcome the variabilityintroduced through tissue heterogeneity in archivalmaterial [13,36].

FFPE tissue blocks screened for other proteins,including β-tubulin, fascin and β-dystroglycan showedintense, specific banding with western blotting. How-ever, the effect of the extraction protocol on membraneproteins was variable. Although protein hydrophobic-ity and the number of transmembrane domains maycontribute to the degree of variation observed, our datasuggest that the -temperature extraction regime neces-sary for crosslink reversal also causes membrane pro-tein aggregation, corroborating previously publisheddata [37]. The results highlight the need for protocolrefinement and cautious interpretation of quantitativedata from this subset of proteins.



Figure 6. (A) Composite densitometric data for expression profiles of the seven pre-selected RCC tissue biomarker proteinsvalidated in 10 patients undergoing nephrectomy for RCC (70 data points). Directional regulation of expression is shown byplotting fold changes in relation to tumour tissue between FFPE and fresh frozen extracts. Statistical analysis using Spearman’srank correlation gave r = 0.7292 with p < 1.825e-09. (B) Expression profiles of fold changes of the two patient datasets withthe maximum and minimum variation in intensity of expression between fresh and FFPE tissues for the proteins evaluated.Tumour/normal fold change comparisons between the two FFPE and frozen tissue datasets showed an average minimum variationof 1.1 [SD ± 0.3 (R320)] and a maximum variation of 1.7 [SD ± 2.01 (R315)] compared to the ideal ratio of 1

Figure 7. (A) Variation of the extraction efficacies for membrane proteins using the FFPE Laemmli protocol. Blots probed forGolgin and NaK–ATPase (α1 and ß1 subunit) failed to show signals in the FFPE normal and tumour extracts (FFPE-N, -T) whilebeing detectable in the matched fresh frozen extracts (FN, FT). ß-Dystroglycan and the RCC-specific tissue marker protein CAIXare clearly detected in both fresh tumour tissue (FT) and formalin-fixed tumour tissue extracts (FFPE-T), with clear evidenceof isoform specificity in both extracts. (B) HIAR protocol-induced thermal aggregation observed with NaK–ATPase that couldaccount for the variation observed in membrane protein extraction

Further refinements may come from a detailed con-sideration of the sequential chemical processes result-ing in formaldehyde-fixation, allowing rational designof further reversal strategies [38]. Effects of pre-analytical variables associated with tissue harvestingand processing, including fixation length, processingprotocols, length and temperature of storage, need tobe investigated [39] so that data can be interpreted cor-rectly and standardization of processing/storage prac-tices for future FFPE tissue banks optimized [39–41].Parallel development of alternative fixatives may alsobe a possibility, although their acceptability for clin-ical practice would have to be extensively inves-tigated. The FFPE Laemmli workflow provides aviable route to achieve targeted quantitative validation

of novel biomarkers. Although a degree of protein-specific optimization is necessary to define optimalextraction/detection conditions prior to screening, themethod provides a powerful tool to investigate thearchival proteome. More importantly, it provides aninitial detailed framework to carry out the further pro-tocol refinement critically needed to expedite earlyphase research in archival proteomics.

Acknowledgements

The financial support of Cancer Research UK and the researchsupport of the oncology and urology teams, particularly JoanneBrown and the patients at St. James’s University Hospital, aregratefully acknowledged.



Figure 8. Optimized workflow for the extraction of intact, full-length, immunoreactive proteins from formalin-fixedparaffin-embedded tissue using the FFPE Laemmli protocol

References

1. Cho WCS, Cheng CHK. Oncoproteomics: current trends andfuture perspectives. Exp Rev Proteom 2007;4:401–410.

2. Cho WCS. Contribution of oncoproteomics to cancer biomarkerdiscovery. Mol Cancer 2007;6:60.

3. Wilson JF. The promise of disease proteomics: faster detec-tion, diagnosis, and drug development. Ann Intern Med2004;140:317–319.

4. Rifai N, Gillette MA, Carr SA. Protein biomarker discovery andvalidation: the long and uncertain path to clinical utility. NatBiotechnol 2006;24:971–983.

5. Bodovitz S, Joos T. The proteomics bottleneck: strategies forpreliminary validation of potential biomarkers and drug targets.Trends Biotechnol 2004;22:4–7.

6. Becker KF, Schott C, Hipp S, Metzger V, Porschewski P, Beck R,et al. Quantitative protein analysis from formalin-fixed tissues:implications for translational clinical research and nanoscalemolecular diagnosis. J Pathol 2007;211:370–378.

7. Hood BL, Conrads TP, Veenstra TD. Mass spectrometric analysisof formalin-fixed paraffin-embedded tissue: unlocking theproteome within. Proteomics 2006;6:4106–4114.

8. Ahram M, Flaig MJ, Gillespie JW, Duray PH, Linehan WM,Ornstein DK, et al. Evaluation of ethanol-fixed, paraffin-embeddedtissues for proteomic applications. Proteomics 2003;3:413–421.

9. Shi SR, Liu C, Balgley BM, Lee C, Taylor CR. Protein extractionfrom formalin-fixed, paraffin-embedded tissue sections: qualityevaluation by mass spectrometry. J Histochem Cytochem2006;54:739–743.

10. Hood BL, Darfler MM, Guiel TG, Furusato B, Lucas DA,Ringeisen BR, et al. Proteomic analysis of formalin-fixed prostatecancer tissue. Mol Cell Proteom 2005;4:1741–1753.

11. Guo T, Wang WJ, Rudnick PA, Song T, Li J, Zhuang ZP,et al. Proteome analysis of microdissected formalin-fixed andparaffin-embedded tissue specimens. J Histochem Cytochem2007;55:763–772.

12. Crockett DK, Lin ZS, Vaughn CP, Lim MS, Elenitoba-Johnson KS.Identification of proteins from formalin-fixed paraffin-embeddedcells by LC–MS/MS. Lab Invest 2005;85:1405–1415.

13. Bagnato C, Thumar J, Mayya V, Hwang SI, Zebroski H, Claf-fey KP, et al. Proteomics analysis of human coronary atheroscle-rotic plaque — a feasibility study of direct tissue proteomics byliquid chromatography and tandem mass spectrometry. Mol CellProteom 2007;6:1088–1102.

14. Nirmalan NJ, Harnden P, Selby PJ, Banks RE. Mining thearchival formalin-fixed paraffin-embedded tissue proteome:opportunities and challenges. Mol Biosyst 2008;4:712–720.

15. Shi SR, Key ME, Kalra KL. Antigen retrieval in formalin-fixed, paraffin-embedded tissues: an enhancement method forimmunohistochemical staining based on microwave oven heatingof tissue sections. J Histochem Cytochem 1991;39:741–748.

16. Shi SR, Datar R, Liu C, Wu L, Zhang Z, Cote RJ, et al. DNAextraction from archival formalin-fixed, paraffin-embedded tissues:heat-induced retrieval in alkaline solution. Histochem Cell Biol2004;122:211–218.

17. Stanta G, Bonin S, Perin R. RNA extraction from formalin-fixedand paraffin-embedded tissues. Methods Mol Biol 1998;86:23–26.

18. Yamashita S. Heat-induced antigen retrieval: mechanisms andapplication to histochemistry. Progr Histochem Cytochem2007;41:141–200.

19. Layfield R, Bailey K, Lowe J, Allibone R, Mayer RJ, Landon M.Extraction and protein sequencing of immunoglobulin light chainfrom formalin-fixed cerebrovascular amyloid deposits. J Pathol1996;180:455–459.

20. Prieto DA, Hood BL, Darfler MM, Guiel TG, Lucas DA, Con-rads TP, et al. Liquid tissue: proteomic profiling of formalin fixedtissues. Biotechniques 2005;38:532–535.

21. Guo T, Wang W, Rudnick PA, Song T, Li J, Zhuang Z,et al. Proteome analysis of microdissected formalin-fixed andparaffin-embedded tissue specimens. J Histochem Cytochem2007;55:763–772.



22. Palmer-Toy DE, Krastins B, Sarracino DA, Nadol JB, Mer-chant SN. Efficient method for the proteomic analysis of fixedand embedded tissues. J Proteome Res 2005;4:2404–2411.

23. Ikeda K, Monden T, Kanoh T, Tsujie M, Izawa H, Haba A, et al.Extraction and analysis of diagnostically useful proteins fromformalin-fixed, paraffin-embedded tissue sections. J HistochemCytochem 1998;46:397–404.

24. Laemmli UK. Cleavage of structural proteins during the assemblyof the head of bacteriophage T4. Nature 1970;227:680–685.

25. Hwang SI, Thumar J, Lundgren DH, Rezaul K, Mayya V, Wu L,et al. Direct cancer tissue proteomics: a method to identifycandidate cancer biomarkers from formalin-fixed paraffin-embedded archival tissues. Oncogene 2007;26:65–76.

26. Thongboonkerd V. Proteomic analysis of renal diseases: unrav-elling the pathophysiology and biomarker discovery. Expert RevProteomics 2005;2:349–366.

27. Banks RE, Craven RA, Harnden P, Madaan S, Joyce A, Selby PJ.Key clinical issues in renal cancer: a challenge for proteomics.World J Urol 2007;25:537–556.

28. Craven RA, Banks RE. Laser capture microdissection and pro-teomics: possibilities and limitations. Proteomics2001;1:1200–1204.

29. Craven RA, Hanrahan S, Totty N, Harnden P, Stanley AJ, MaherER, et al. Proteomic identification of a role for the von Hip-pel–Lindau tumour suppressor in changes in the expression ofmitochondrial proteins and septin 2 in renal cell carcinoma. Pro-teomics 2006;6:3880–3893.

30. Craven RA, Stanley AJ, Hanrahan S, Dods J, Unwin R, Totty N,et al. Proteomic analysis of primary cell lines identifiesprotein changes present in renal cell carcinoma. Proteomics2006;6:2853–2864.

31. Craven RA, Banks RE. Understanding and managing renal cellcarcinoma: can proteomic studies contribute to clinical practice?Contrib Nephrol 2008;160:88–106.

32. Towbin H, Staehelin T, Gordon J. Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. Proc Natl Acad Sci USA1979;76:4350–4354.

33. Banks RE, Craven RA, Harnden PA, Selby PJ. Use of asensitive EnVision+-based detection system for Western blotting:avoidance of streptavidin binding to endogenous biotin andbiotin-containing proteins in kidney and other tissues. Proteomics2003;3:558–561.

34. Pastwa E, Somiari SB, Czyz M, Somiari RI. Proteomics in humancancer research. Proteom Clin Appl 2007;1:4–17.

35. Becker KF, Schott C, Becker I, Hofler H. Guided proteinextraction from formalin-fixed tissues for quantitative multiplexanalysis avoids detrimental effects of histological stains. ProteomClin Appl 2008;2:737–743.

36. Patel V, Hood BL, Molinolo A, Lee NH, Conrads TP, BraistedJC, et al. Proteomic analysis of laser-captured paraffin-embeddedtissues: a molecular portrait of head and neck cancer progression.Clin Cancer Res 2008;14:1002–1014.

37. Broughton J, Swennen D, Wilkinson BM, Joyet P, Gaillardin C,Stirling CJ. Cloning of SEC61 homologues from Schizosaccha-romyces pombe and Yarrowia lipolytica reveals the extent of func-tional conservation within this core component of the ER translo-cation machinery. J Cell Sci 1997;110:2715–2727.

38. Metz B, Kersten GFA, Hoogerhout P, Brugghe HF, Timmer-mans HAM, de Jong A, et al. Identification of formaldehyde-induced modifications in proteins: reactions with model peptides.J Biol Chem 2004;279:6235–6243.

39. Mischak H, Apweiler R, Banks RE, Conaway M, Coon J,Dominiczak A, et al. Clinical proteomics: a need to define thefield and to begin to set adequate standards. Proteom Clin Appl2007;1:148–156.

40. Qualman SJ, France M, Grizzle WE, LiVolsi VA, Moskaluk CA,Ramirez NC, et al. Establishing a tumour bank: banking,informatics and ethics. Br J Cancer 2004;90:1115–1119.

41. Shi SR, Liu C, Perez J, Taylor CR. Protein-embedding technique:a potential approach to standardization of immunohistochemistryfor formalin-fixed paraffin-embedded tissue sections. J HistochemCytochem 2005;53:1167–1170.


development and validation of a novel protein extraction methodology for quantitation of protein...

Documents