MS proteomic profiling of formalin-fixed, paraffin-embedded tissues

J O U R NA L OF PR O TE O MI CS 7 4 ( 2 01 1 ) 1 0 1 5–1 0 2 1

available at www.sciencedirect.com

www.elsevier.com/locate/jprot

Technical note

Impact of fixation time on GeLC–MS/MS proteomic profiling of formalin-fixed, paraffin-embedded tissues Alessandro Tanca a , Daniela Pagnozzi a , Giovanni Falchi a , Grazia Biosa a , Stefano Rocca b , Gisella Foddai b , Sergio Uzzau a,c , Maria Filippa Addis a,b,⁎ a

Porto Conte Ricerche Srl, Tramariglio, Alghero (SS), Italy Dipartimento di Patologia e Clinica Veterinaria, Università degli Studi di Sassari, Sassari, Italy c Dipartimento di Scienze Biomediche, Università degli Studi di Sassari, Italy b

AR TIC LE I N FO

ABS TR ACT

Article history:

Formalin-fixed, paraffin-embedded (FFPE) tissue banks represent an invaluable resource for

Received 10 February 2011

biomarker discovery. Recently, the combination of full-length protein extraction, GeLC–MS/MS

Accepted 14 March 2011

analysis, and spectral counting quantification has been successfully applied to mine proteomic

Available online 23 March 2011

information from these tissues. However, several sources of variability affect these samples; among these, the duration of the fixation process is one of the most important and most easily

Keywords:

controllable ones. To assess its influence on quality of GeLC–MS/MS data, the impact of fixation

FFPE

time on efficiency of full-length protein extraction efficiency and on quality of label-free

GeLC–MS/MS

quantitative data was evaluated.

Fixation time

As a result, although proteins were successfully extracted from FFPE liver samples fixed for up

Formaldehyde

to eight days, fixation time appeared to negatively influence both protein extraction yield and

Proteomics

GeLC–MS/MS quantitative proteomic data. Particularly, MS identification efficiency decreased with increasing fixation times. Moreover, amino acid modifications putatively induced by formaldehyde were detected and characterized. These results demonstrate that proteomic information can be achieved also from tissue samples fixed for relatively long times, but suggest that variations in fixation time need to be carefully taken into account when performing proteomic biomarker discovery studies on fixed tissue archives. © 2011 Elsevier B.V. All rights reserved.

Fixing and embedding of surgical and bioptic tissue samples are routinely performed to enable the pathological characterization of clinical specimens. The criteria employed for the diagnosis of most oncological and degenerative diseases have been actually established in formalin-fixed and paraffinembedded (FFPE) tissue sections stained by hematoxylin and eosin. Thus, the ability to retrospectively analyze documented archival tumor cases, with known diagnosis, prognosis, response to therapy, and outcome, represents a significant

resource for biomarker discovery [1,2]. However, formalin fixation leads to significant cross-linking among proteins and other molecules in the tissue, hampering molecular investigation of these samples. Since 1991, immunodetection of antigens in FFPE tissue has been substantially improved thanks to the development of the antigen retrieval (AR) technique [3–5], consisting in heat-treating the sample in the presence of a suitable buffer, allowing the renaturation of the protein structure through possible breaking of the methylene

⁎ Corresponding author at: Porto Conte Ricerche Srl, Tramariglio, Alghero (SS), Italy. Tel.: + 39 079 998 526; fax: +39 079 998 567. E-mail address: [email protected] (M.F. Addis). 1874-3919/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jprot.2011.03.015

1016

J O U R NA L OF PR O TE O MI CS 74 ( 20 1 1 ) 1 0 1 5–1 0 2 1

bonds formed during the fixation process [6–8]. Several ARderived approaches have been developed in the last years which increase the amount of protein extractable from FFPE tissues [9–15]. The exploitation of such procedures has led to remarkable results, mostly by using gel-free shotgun proteomic approaches [16–18]. Recently, we reported an optimized extraction method enabling recovery of full-length proteins from FFPE tissues [19]; the successful application of gel-based proteome analysis approaches, such as 1-D PAGE followed by LC–MS/MS (GeLC–MS/MS) [19], 2-D PAGE-MS [20], and 2-D DIGE [21], was described using animal model tissues. However, an in-depth understanding of variables introduced by tissue fixation and processing is still far away. Particularly, the extent of variability due to one of the most important and experimentally evaluable pre-analytical factors, namely fixation time, and its qualitative and quantitative impact on protein extraction and gel-based proteomic profiling, are mostly unexplored, while some evidence has been gained in gel-free shotgun surveys exploiting direct trypsinization of FFPE extracts [14,16]. This is a key point for translational studies on FFPE samples, since the duration of tissue fixation is not highly standardized and may range from hours to several days, thus introducing a possible source of variability in proteomic data. In this work, the effect of fixation time on full-length protein extraction efficiency and on GeLC–MS/MS spectral counting data was evaluated on an animal model tissue. Moreover, the extent and nature of molecular modifications typically introduced by formaldehyde on amino acid residues were investigated. To evaluate the influence exerted by the duration of the formalin fixation process, an animal model was established, in order to maximize sample availability with limited ethical issues; furthermore, tissue specimens were collected from a single individual, in order to minimize biological variability. Dog liver was chosen as a tissue source because of the large animal size, allowing the collection of several replicates of the same tissue, and in view of the satisfactory and consistent results achieved in previous gel-based proteomic studies using FFPE liver extracts, in terms of extraction efficiency as well as pattern complexity and resolution [19,21]. Immediately after slaughtering, tissue cubes of about 1 cm3 were cut from the liver of a healthy dog, and multiple replicate samples were fixed in 10% neutral buffered formalin (4% formaldehyde, 0.03 M NaH2PO4, 0.04 M Na2HPO4, pH 6.8) at room temperature for 24, 48, 120, and 192 h, respectively. Then, tissue specimens were put in an automatic tissue processor (Pabisch, Top Processor LX 120/300) and treated under shaking with the following reagents for dehydration and embedding: formalin 40 °C 351 + 200 min, ethanol 70% 40 °C 200 min, ethanol 80% 40 °C 200 min, ethanol 90% 40 °C 130 min, ethanol 100% 40 °C 130 + 100 min, xylene 40 °C 2 × 100 min, and paraffin 59 °C 3 × 45 min. All samples used in this study had been generated as a result of routine surgical procedures performed in the Veterinary Faculty. Three series of ten sections per fixation time were independently subjected to protein extraction as described previously [19]. Triplicate series of ten FFPE tissue microtome sections, 10 mm thick, 80 mm2 wide, were placed in Eppendorf safe-lock tubes (Eppendorf, Hamburg, Germany) and deparaffinized by incubation at room temperature in xylene for 5 min. After each incubation, the tissue was

pelleted at 12 000 ×g for 3 min, and incubation/centrifugation steps were repeated two more times. The deparaffinized tissue pellets were then rehydrated with a graded series of ethanol (100%, 96%, and 70%), briefly air-dried in a fume hood, and weighed. Then, the tissue pellets were immersed at a 20% w/v ratio in 20 mM Tris HCl, pH 8.8, 2% SDS, 200 mM DTT, extraction buffer, and subjected to high-temperature extraction at 100 °C for 20 min, and subsequently at 80 °C for 2 h with shaking at 400 rpm in a Thermomixer (Eppendorf). Extracts were clarified for 15 min at 12 000 ×g at 4 °C, quantified by the EZQ Protein Quantitation Kit (Molecular Probes, Eugene, OR, USA), and stored at −20 °C until needed. Extraction yields were assessed for each technical replicate by measuring the micrograms of extracted protein per milligram of starting FFPE tissue. A dog liver archival specimen, which had been stored in formalin for about 5 years, was retrieved and used as “end point” control. As reported in Fig. 1A, proteins were extracted with satisfactory yields from tissues extracted for up to 8 days (192 h). On the other hand, extraction efficiency showed a negative correlation with fixation time (Pearson correlation coefficient, r = −0.84); moreover, the amount of protein extracted from the samples treated for 24 h and 192 h differed significantly (p < 0.05, ANOVA). The archival sample fixed for five years provided an extremely lower protein extraction yield (p < 0.001), confirming a time-dependent influence of fixation on protein extraction performance. Furthermore, when subjected to precipitation procedures (2-D Clean-Up Kit, GE Healthcare, following manufacturers' instructions) the protein pellet obtained from 5-years-fixed tissues displayed a viscous appearance upon resuspension with buffers (differently from proteins extracted from samples fixed from 24 to 192 h, which were similar in terms of visible appearance). This unusual physical behavior might be related to the possibly very high number of crosslinks established with time among tissue molecules. Interestingly, our data suggest, as expected, that increasing formalin fixation times are associated with an increasing difficulty in extracting proteins from FFPE tissues, and that very long fixation reactions make it impossible to obtain proteins which are viable to electrophoretic analysis. This is clearly consistent with formaldehyde reactivity, which involves the formation of reactive intermediates with a time-dependent introduction of intra- and interchain bonds [22–24]; therefore, a longer exposition to the reagent is expected to lead to a tighter net of molecular crosslinks. Subsequently, the first three fixation time points (24 h, 48 h, and 120 h) were selected for the proteomic analysis on the influence of fixation time, since they cover the time range usually adopted in standard hospital tissue fixation procedures. Twenty-five micrograms of each protein extract was separated by 1-D SDS-PAGE, according to Laemmli [25]. Gels were stained with Coomassie Brilliant blue G-250, and digitalized with an ImageScanner III (GE Healthcare, Little Chalfont, UK). As shown in Fig. 1B, the protein patterns corresponding to different fixation times exhibited a comparable (and remarkable, since dealing with FFPE samples) quality, both in terms of intensity and resolution. Pattern intensities were plotted against the running front and compared using QuantityOne software (Bio-Rad), using the following parameters for band detection: sensitivity, 10.000;

J O U R NA L OF PR O TE O MI CS 7 4 ( 2 01 1 ) 1 0 1 5–1 0 2 1

1017

Fig. 1 – Extraction yield, SDS-PAGE and GeLC–MS/MS label-free quantitative analysis. A) Histogram showing the amount of proteins extracted from tissue blocks subjected to increasing formalin fixation times. Asterisk indicates significant differences between 24 h and 192 h values (p < 0.05, ANOVA). Error bars indicate standard deviation of three technical replicates per time point. B) SDS-PAGE of proteins extracted from tissues fixed in formalin for increasingly longer times. Numbers indicate gel bands which were excised and subjected to GeLC–MS/MS analysis. Precision Plus All Blue (Bio-Rad) was used as protein standard (M, MWs on the left). C) Histograms showing values of unique proteins, SpC, UP, mean SC, MQR, and mean pI of identified proteins. Data were obtained after averaging out of three GeLC–MS/MS replicate datasets per fixation time point. Standard deviation among replicates is pointed out by error bars. Single asterisk indicates p < 0.05 between 24 h and 192 h values, while double asterisk indicates p < 0.1 between 24 h and 192 h values.

lane width, 2.400 mm; minimum density, 0.00%; noise filter, 4.00; shoulder sensitivity, 1.00; and size scale: 5. As depicted by the densitogram reported in Supplementary data S1, the three profiles were highly comparable. Samples were then analyzed following a GeLC–MS/MS labelfree workflow, as described previously [26]. Twenty-five micrograms of protein extract per fixing condition was separated by 1-D SDS-PAGE in technical triplicate. Four representative bands, corresponding to different MWs and signal intensities (as indicated by numbers 1–4 in Fig. 1B), were cut from each replicate. Bands were in-gel digested with trypsin as described previously [26], and derivative peptide mixtures were analyzed by LC–MS/MS. LC–MS/MS analyses of tryptic digests were performed on a Q-TOF hybrid mass spectrometer equipped with a nano lock Z-spray source, and coupled online with a capillary chromatography system CapLC (Waters, Manchester, UK), following previously reported procedures

[27]. Protein Lynx Global Server (Version 2.2.5, Waters) was used to analyze raw MS and MS/MS spectra and to generate peak lists which were introduced in Mascot Daemon software (version 2.3.0, Matrix Science, Boston, MA, USA) for protein identification. Multiple peak lists belonging to the same lane (or, when necessary, to the same sample) were merged selecting “Merge MS/MS files into single search” option. SwissProt (version 57.15) was used as sequence database. Search parameters were as follows: fixed modifications carbamidomethyl (C), variable modifications pyro-Glu (Nterm Q) and oxidation (M), peptide tolerance 30 ppm, MS/MS tolerance 0.4 Da, charge states + 2, + 3, and + 4, and enzyme trypsin, allowing up to 1 missed cleavage. Mammalia was used as taxonomy. Because of the limited presence of dog sequences in available databases, cross-species identification was carried out. When different homologue proteins were present in the protein list of the same sample, the

1018

J O U R NA L OF PR O TE O MI CS 74 ( 20 1 1 ) 1 0 1 5–1 0 2 1

protein identity showing the highest number of spectral counts (SpC) was chosen, independently from the mammalian species. When comparing different replicates or samples, SpC belonging to putatively different homologue proteins were all assigned to the protein identity showing the highest number of SpC among the replicates/samples, independently from the mammalian species. False discovery rate (FDR) was determined per LC–MS/MS analysis using Automatic Decoy Search option provided by Mascot, and considering peptide matches above identity threshold. Mascot results were parsed with the IRMa toolbox (version 1.26.1), using the following inclusion criteria: protein significance threshold p < 0.05, peptide rank 1, and peptide average identity threshold p < 0.05, in order to keep FDR below 2%. Unique peptides (UP), sequence coverage (SC), SpC, total spectral queries, and pI values were reported as calculated by IRMa, as well as peptide sequence data. Skin keratins were excluded from the final protein list. Overall MS identifications are available in Supplementary data S2, as well as single peptide identification spectra (Supplementary data S3–S5). In order to assess robustness of the analytical workflow, its reproducibility was first assessed by considering proteins detected with more than 2 SpC. As a result, the percentage of protein identifications shared by three replicates was 49% (average of fixation times, CV = 6%), while the percentage of protein identifications detected in at least two replicates out of three was 90% (CV = 6%). Moreover, the correlation of spectral counting data between two technical replicates gave a mean R2 of 0.93. Representative Venn diagram (A) and correlation plots (B) for the 24 h condition are presented in Supplementary data S6. Therefore, label-free GeLC–MS/MS analysis of FFPE protein extracts showed a satisfactory method reproducibility if compared with previous studies [28]. In order to evaluate the influence of fixation time on MS data obtained with this method, the mean and standard deviation were then calculated among three technical replicates per time point for the following parameters: unique proteins identified, SpC, UP, and mean SC. Moreover, in order to measure how many of the overall spectral queries launched in Mascot had actually been matched with a known peptide sequence, the matched queries ratio (MQR) was defined as the ratio between identified SpC and total spectral queries. Finally, the mean pI of the identified proteins was also calculated. In fact, the amino acidic residues which exhibit the highest reactivity toward formaldehyde are mainly basic. Thus, it can be expected that longer formaldehyde incubation times enable formation of a higher number of methylene bonds, especially on basic residues, making it increasingly difficult to extract and MS-identify basic proteins compared to acidic ones, with a consequent decrease of the mean pI value. A decreasing trend was observed for all the parameters examined, as shown by the histograms in Fig. 1C; in fact, all six quantitative parameters showed a negative correlation with fixation time (r ranging from − 0.90 to −0.99). Moreover, the number of UP was significantly different between 24 and 120 h (p < 0.05); p-values for unique proteins, SpC, and MQR were slightly higher than 0.05 (0.063, 0.065, and 0.074, respectively). Although below significance threshold, the decrease of the mean pI across fixation times is in line with the above explained observations. Also after merging replicate data, a

negative correlation between the selected quantitative parameters and fixation time could be clearly observed (Supplementary data S7). MS data were then analyzed also at a single protein level. First, mean SpC were calculated among the three replicates per fixation time point; then, the 17 proteins which had been identified with an average of at least 2 SpC among fixation conditions were selected (Table 1). Interestingly, for all the selected proteins the mean SpC number detected in the 24 h samples was higher compared to the 120 h samples. The r values were lower than −0.75 for 11 out of 17 proteins, indicating an inverse correlation between fixation time and number of SpC detected per protein. Three proteins out of 17 showed an ANOVA p-value < 0.05 comparing samples fixed for 24 h and 120 h. Considering the average SpC among the 17 proteins selected across fixation conditions, r value was −0.95. Lysine percentage and pI of the mature protein form were calculated, but no association with correlation coefficients was detected. Summarizing, the number of identified peptides decreased significantly with increasing fixation times; moreover, we were unable to identify a progressively higher fraction of spectral peaks in samples subjected to increasingly longer fixation times (indicated by the reduction of MQR), although a comparable total peak number was observed among fixation times. In other words, as fixation time increases, the portion of tryptic peptides that can be successfully identified by MS decreases. A possible explanation of this phenomenon is that these tryptic peptides are being generated as a result of the digestion of two or more crosslinked proteins, with the formation of “hybrid” (and therefore unidentifiable) peptides, as previously illustrated [19,26].

Table 1 – GeLC–MS/MS protein identifications (mean SpC > 2). Accession no. P60712 P49822 P19483 P56480 P31327 P00367 Q53VB8 Q93077 P62807 Q5E9F8 P60529 P60524 O35244 P61284 Q3T0R1 Q32PD5 Q8WNN6 Overall

Protein name Actin, cytoplasmic 1 Serum albumin ATP synthase subunit alpha, mitochondrial ATP synthase subunit beta, mitochondrial Carbamoyl-phosphate synthase [ammonia], mitochondrial Glutamate dehydrogenase 1, mitochondrial Ferritin light chain Histone H2A type 1-C Histone H2B type 1-C/E/F/G/I Histone H3.3 Hemoglobin subunit alpha Hemoglobin subunit beta Peroxiredoxin-6 60S ribosomal protein L12 40S ribosomal protein S18 40S ribosomal protein S19 Superoxide dismutase [Cu–Zn]

SpC SpC SpC 24 h 48 h 120 h 8 3 9

6 4 10

3 3 7

8

9

6

10

6

4

3

3

3

6 4 10 4 42 73 4 4 5 5 6 12

4 4 9 1 37 64 3 5 3 4 3 10

5 2 6 2 33 58 3 3 2 3 2 9

J O U R NA L OF PR O TE O MI CS 7 4 ( 2 01 1 ) 1 0 1 5–1 0 2 1

As a further investigation, the molecular features of peptides belonging to proteins identified from tissues subjected to different fixation times were assessed. First, the percentage distribution of amino acids within peptides across time points was determined, and neither evident nor consistent trends could be observed (Supplementary data S8). Peptide sequences corresponding to all identified SpC were also investigated for terminal lysine to arginine ratio (K/R ratio). In fact, a possible variation of this value in fixed tissue extracts could be expected, in keeping with the different degree of reactivity toward formaldehyde of these amino acid residues [22]. As a result, an increasing trend was measured (r = 0.99), with near-significant differences between 24 and 120 h (p = 0.07, ANOVA). Previous data, obtained upon tryptic digestion of the whole protein extract, demonstrated a decrease of K/R ratio comparing fresh-frozen to FFPE proteomes [16], or a substantial equivalence between lysine- and arginine-terminating peptides [10,29]. On the contrary, in a previous MS examination of paired FFPE and fresh-frozen tissues subjected to the same extraction method used in this study [19], we saw a slight increase of the K–R ratio on the ten main identified proteins within the proteome (1.11 in fresh-frozen and 1.36 in FFPE). Thus, differences in K- or R-terminating peptides would appear to depend on the extraction procedure used, and therefore on the degree of

1019

reversal of formaldehyde-induced bonds (particularly, of crosslinks involving lysine and arginine residues). With the purpose of studying possible modifications induced specifically by the fixation process, MS peak lists were then re-analyzed by Mascot, selecting new customized modifications based on findings about formaldehyde reactivity with proteins. As reported by Metz et al. [22,24], the reaction of formaldehyde with proteins (namely toward arginine, cysteine, histidine, lysine, and tryptophan residues, and toward the protein N-terminus) leads to the formation of stable crosslinks with several amino acids (particularly arginine, asparagine, glutamine, histidine, tryptophan, and tyrosine), via reactive intermediates, such as Schiff bases. Based on the number of reacting formaldehyde molecules, the observable mass shift may be equal to +12 Da (1CH2O) or + 24 Da (2CH2O). Thus, two new variable modifications, accounting for +12 and +24 Da shifts on all nine possible reactive residues, were created into Mascot Daemon and used to carry out a new analysis of GeLC–MS/MS data. All MS/MS spectra of peptides indicated to bear modified residues were manually inspected. As a result, no +24 Da modifications were detected, while spectra examination supported the dependability of several + 12 Da modifications, with a general reduction across fixation times (16, 14, and 9 modified peptide sequences for 24, 48, and 120 h, respectively). Specifically,

Fig. 2 – Modification analysis. MS/MS spectrum of a Tyr-modified peptide (DSYVGDEAQSKR, MH+theor = 1354.6; MH+exp = 1366.6). In the y-series, a mass increment equal to 12 Da is carried by the y10 component and lost by the y9 component.

1020

J O U R NA L OF PR O TE O MI CS 74 ( 20 1 1 ) 1 0 1 5–1 0 2 1

+12 Da mass shifts were clearly observable on N-termini, lysine, tryptophan, and tyrosine. A representative MS/MS spectrum of a tyrosine-modified peptide is shown in Fig. 2. To date, the presence of this modification was demonstrated in model experiments (on synthetic peptides [22] and insulin [24]); as far as actual tissue protein extracts are concerned, such modifications have been previously demonstrated only through a MALDI imaging approach [30], but not through a LC–MS/MS approach [16]. In conclusion, these results demonstrate the ability to successfully achieve dependable proteomic information from tissue samples fixed up to several days. In addition, both protein extraction efficiency and ability to correctly identify proteins by MS appear to be significantly influenced by the duration of the fixation procedure. Finally, chemical modifications induced by formaldehyde on proteins were reliably detected in fixed tissue extracts. On the whole, these data highlight the importance of precisely standardize fixation procedures in pathology laboratories, as well as the significance of keeping as more homogeneous as possible the fixation time range when selecting the FFPE samples to be investigated, in order to maximize dependability of proteomic results. Moreover, further development of efficient and optimized protocols for protein extraction from FFPE tissues might also contribute to minimize molecular biases introduced by formalin fixation. Supplementary materials related to this article can be found online at doi:10.1016/j.jprot.2011.03.015.

Acknowledgments The authors wish to thank Dr. Roberto Tonelli for helpful suggestions about statistical analysis, Dr. Salvatore Pisanu for critical reading and suggestions on the manuscript, and Dr. Tonina Roggio for her valuable support. This work was supported by funding from Regione Autonoma della Sardegna.

REFERENCES [1] Fox C, Johnson F, Whiting J, Roller P. Formaldehyde fixation. J Histochem Cytochem 1985;33:845–53. [2] Hood BL, Conrads TP, Veenstra TD. Mass spectrometric analysis of formalin-fixed paraffin-embedded tissue: unlocking the proteome within. Proteomics 2006;6:4106–14. [3] Shi S, Key M, Kalra K. Antigen retrieval in formalin-fixed, paraffin-embedded tissues: an enhancement method for immunohistochemical staining based on microwave oven heating of tissue sections. J Histochem Cytochem 1991;39: 741–8. [4] Shi S-R, Cote RJ, Taylor CR. Antigen retrieval techniques: current perspectives. J Histochem Cytochem 2001;49:931–8. [5] Shi S-R, Shi Y, Taylor CR. Antigen retrieval immunohistochemistry. J Histochem Cytochem 2011;59: 13–32. [6] Sompuram SR, Vani K, Messana E, Bogen SA. A molecular mechanism of formalin fixation and antigen retrieval. Am J Clin Pathol 2004;121:190–9. [7] Yamashita S. Heat-induced antigen retrieval: mechanisms and application to histochemistry. Prog Histochem Cytochem 2007;41:141–200.

[8] Bogen SA, Vani K, Sompuram SR. Molecular mechanisms of antigen retrieval: antigen retrieval reverses steric interference caused by formalin-induced cross-links. Biotech Histochem 2009;84:207–15. [9] Ikeda K, Monden T, Kanoh T, Tsujie M, Izawa H, Haba A, et al. Extraction and analysis of diagnostically useful proteins from formalin-fixed, paraffin-embedded tissue sections. J Histochem Cytochem 1998;46:397–404. [10] Hood BL, Darfler MM, Guiel TG, Furusato B, Lucas DA, Ringeisen BR, et al. Proteomic analysis of formalin-fixed prostate cancer tissue. Mol Cell Proteomics 2005;4:1741–53. [11] Shi S-R, Liu C, Balgley BM, Lee C, Taylor CR. Protein extraction from formalin-fixed, paraffin-embedded tissue sections: quality evaluation by mass spectrometry. J Histochem Cytochem 2006;54:739–43. [12] Jiang X, Jiang X, Feng S, Tian R, Ye M, Zou H. Development of efficient protein extraction methods for shotgun proteome analysis of formalin-fixed tissues. J Proteome Res 2007;6: 1038–47. [13] 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 nanoscale molecular diagnosis. J Pathol 2007;211:370–8. [14] Xu H, Yang L, Wang W, Shi S-R, Liu C, Liu Y, et al. Antigen retrieval for proteomic characterization of formalin-fixed and paraffin-embedded tissues. J Proteome Res 2008;7:1098–108. [15] Nirmalan NJ, Harnden P, Selby PJ, Banks RE. Development and validation of a novel protein extraction methodology for quantitation of protein expression in formalin-fixed paraffin-embedded tissues using western blotting. J Pathol 2009;217:497–506. [16] Sprung RW, Brock JWC, Tanksley JP, Li M, Washington MK, Slebos RJC, et al. Equivalence of protein inventories obtained from formalin-fixed paraffin-embedded and frozen tissue in multidimensional liquid chromatography–tandem mass spectrometry shotgun proteomic analysis. Mol Cell Proteomics 2009;8:1988–98. [17] Ostasiewicz P, Zielinska DF, Mann M, Wiśniewski JR. Proteome, phosphoproteome, and N-glycoproteome are quantitatively preserved in formalin-fixed paraffin-embedded tissue and analyzable by high-resolution mass spectrometry. J Proteome Res 2010;9:3688–700. [18] Xiao Z, Li G, Chen Y, Li M, Peng F, Li C, et al. Quantitative proteomic analysis of formalin-fixed and paraffin-embedded nasopharyngeal carcinoma using iTRAQ labeling, two-dimensional liquid chromatography, and tandem mass spectrometry. J Histochem Cytochem 2010;58:517–27. [19] Addis MF, Tanca A, Pagnozzi D, Crobu S, Fanciulli G, Cossu-Rocca P, et al. Generation of high-quality protein extracts from formalin-fixed, paraffin-embedded tissues. Proteomics 2009;9:3815–23. [20] Addis MF, Tanca A, Pagnozzi D, Rocca S, Uzzau S. 2-D PAGE and MS analysis of proteins from formalin-fixed, paraffin-embedded tissues. Proteomics 2009;9:4329–39. [21] Tanca A, Pagnozzi D, Falchi G, Tonelli R, Rocca S, Roggio T, et al. Application of 2-D DIGE to formalin-fixed, paraffin-embedded tissues. Proteomics 2011;11:1005–11. [22] Metz B, Kersten GFA, Hoogerhout P, Brugghe HF, Timmermans HAM, de Jong A, et al. Identification of formaldehyde-induced modifications in proteins. J Biol Chem 2004;279:6235–43. [23] Toews J, Rogalski J, Clark T, Kast J. Mass spectrometric identification of formaldehyde-induced peptide modifications under in vivo protein cross-linking conditions. Anal Chim Acta 2008;618:168–83. [24] Metz B, Kersten GFA, Baart GJE, de Jong A, Meiring H, ten Hove J, et al. Identification of formaldehyde-induced modifications in proteins: reactions with insulin. Bioconjug Chem 2006;17: 815–22.

J O U R NA L OF PR O TE O MI CS 7 4 ( 2 01 1 ) 1 0 1 5–1 0 2 1

[25] Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970;227:680–5. [26] Tanca A, Addis MF, Pagnozzi D, Cossu-Rocca P, Tonelli R, Falchi G, et al. Proteomic analysis of formalin-fixed, paraffin-embedded lung neuroendocrine tumor samples from hospital archives. J Proteomics 2011;74:359–70. [27] Pisanu S, Ghisaura S, Pagnozzi D, Biosa G, Tanca A, Roggio T, et al. The sheep milk fat globule membrane proteome. J Proteomics 2011;74:350–8. [28] Gao BB, Stuart L, Feener EP. Label-free quantitative analysis of one-dimensional PAGE LC/MS/MS proteome: application on

1021

angiotensin II-stimulated smooth muscle cells secretome. Mol Cell Proteomics 2008;7:2399–409. [29] DeSouza LV, Krakovska O, Darfler MM, Krizman DB, Romaschin AD, Colgan TJ, et al. mTRAQ-based quantification of potential endometrial carcinoma biomarkers from archived formalin-fixed paraffin-embedded tissues. Proteomics 2010;10:3108–16. [30] Lemaire R, Desmons A, Tabet JC, Day R, Salzet M, Fournier I. Direct analysis and MALDI imaging of formalin-fixed, paraffin-embedded tissue sections. J Proteome Res 2007;6: 1295–305.