Development of Efficient Protein Extraction Methods for Shotgun

Proteome Analysis of Formalin-Fixed Tissues

Xiaogang Jiang,, Xinning Jiang, Shun Feng, Ruijun Tian, Mingliang Ye,*, and Hanfa Zou*,

National Chromatographic R&A Center, Dalian Institute of Chemical Physics, Chinese Academy of Sciences,
Dalian 116023, China, and School of Medicine, Suzhou University, Suzhou, Jiangsu 215007, China

Received October 10, 2006

There are vast archives of formalin-fixed tissues spanning many conceivable conditions such as different
diseases, time courses, and different treatment and allowing acquisition of the necessary numbers of
samples to carry out biomarker discovery study. However, the conventional protein analysis approach
is not applicable for the analysis of proteins in the formalin-fixed tissue because the formalin fixation
process resulted in the cross-linking of proteins, and thus, intact proteins cannot be efficiently extracted.
In this study, several protocols were investigated to extract proteins from formalin-fixed mouse liver
tissue for shotgun proteome analysis. It was found that incubation of tissue in a lysis buffer containing
6 M guanidine hydrochloride at high temperature led to the highest protein yield and the largest number
of proteins identified. The peptides and proteins identified from formalin-fixed tissue were first
comprehensively compared with those identified from frozen-fresh tissue. It was found that a majority
of peptides identified from fixed tissue were unmodified and proteome coverage for the analysis of
fixed tissue was not obviously compromised by the formalin fixation process. Valuable proteome
information could be obtained by shotgun proteome analysis of formalin-fixed tissue, which presents
a new approach for disease biomarker discovery.

Keywords: formalin-fixed tissue protein extraction shotgun proteome tandem mass spectrometry

Introduction However, the formalin fixation process leads to the formation

Tissue-based proteome studies promise to decipher pro-
teome complexities of the tissue microenvironment and deliver
biomarker information with appropriate pathologic and his-
tologic relevance.1-5 Mass spectrometric profiling of complex
cellular proteome obtained from diseased tissue has previously
been demonstrated with frozen cancer tissue.6-9 Though fresh
and /or frozen tissue samples represent attractive samples for
biomarker discovery, the use of frozen tissue for such analysis
has some disadvantages. Biomarker discovery typically requires
analysis of dozens of different samples; however, the collection
of enough numbers of these tissue samples is very difficult
especially for human tissue. The frozen tissue also requires
specialized equipment for storage, which is expensive and not
suitable for long time storage. In contrast, formalin fixation of
tissue is a standard processing methodology practiced in
medical laboratories worldwide resulting in a highly stable form
of tissue that is easily stored due to its inherent stability at room
temperature. This routine process provides an easily stored
archive of tissue that is physiologically/pathologically well-
defined.
* Authors for correspondence. National Chromatographic R&A Center,
Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian
116023, China. Prof. Dr. Hanfa Zou: tel, +86-411-84379610; fax, +86-411-
84379620; e-mail, hanfazou@dicp.ac.cn. Dr. Mingliang Ye, tel, +86-411-
84379620; fax, +86-411-84379620; e-mail, mingliang@dicp.ac.cn.
Dalian Institute of Chemical Physics, Chinese Academy of Sciences.
Suzhou University.
1038 Journal of Proteome Research 2007, 6, 1038-1047
Published on Web 02/01/2007
of a net of covalent cross-links between side chains of proteins
by methylenic bridge formation. Therefore, methods for analy-
sis of proteins in formalin-fixed tissue are limited to immuno-
histochemistry (IHC),10 a technique that provides the intrac-
ellular localization of a protein. Unfortunately IHC requires a
priori knowledge of individual proteins being analyzed, and it
is not a high-throughput protein analysis approach for large-
scale analysis. Conventional protein analysis approaches are
not applicable to formalin-fixed tissue because the intact
proteins cannot be efficiently extracted. Development of strate-
gies to permit utilization of the universal formalin-fixed
specimens will be important in leveraging the application of
powerful mass spectrometry based proteomic approaches into
the investigation of archival experiment/clinical specimens.
The shotgun proteomics strategy, based on digesting proteins
into peptides and sequencing them using tandem mass spec-
trometry and automated database searching, has become the
method of choice for identifying proteins in most large-scale
studies.10-13 Several groups have reported the application of
the shotgun method to analyze the proteome of formalin-fixed
specimens. Prieto et al.14 and Hood et al.15 reported the use of
commercial Liquid Tissue buffer to extract proteins from
formalin-fixed, paraffin-embedded (FFPE) tissues for shotgun
proteome analysis. Hundreds of proteins from formalin-fixed
prostate cancer tissue were successfully identified, including
several known prostate cancer markers such as prostate-
10.1021/pr0605318 CCC: $37.00 2007 American Chemical Society
Shotgun Proteome Analysis of Formalin-Fixed Tissue
specific antigen, prostatic acid phosphatase, and macrophage
inhibitory cytokine-1.15 Shotgun proteomics was also applied
to the analysis of FFPE cell-block of a human lymphoma cell
line; the results were comparable to those obtained with lysates
from a fresh specimen of the lymphoma cell line.16 Recently,
the extraction of proteins from formalin-fixed tissue was
conducted by incubation of these tissue sections in lysis buffer
containing 2% SDS at high-temperature.17 Hwang et al.18
developed a methodology termed direct tissue proteomics, and
it was also concluded that the extraction condition, lysis buffer
containing 2% SDS at high temperature, was a better protocol
to extract proteins. Although only a few papers published the
shotgun method used to analyze the proteome of formalin-
fixed tissue, these studies indicated the possibility of using the
archival collected formalin-fixed tissues for discovery-driven
biomarker research.
The most important issue in shotgun proteome analysis of
formalin-fixed tissue is to efficiently extract proteins or peptides
from the fixed tissues. Except Hood et al., who used commercial
Liquid Tissue buffer and no detailed information regarding the
composition was given, other researchers mainly used lysis
buffer containing 2% SDS to extract proteins from formalin-
fixed tissue. However, the SDS concentration in protein extract
must be reduced before enzymatic digestion because trypsin
can only tolerate less than 0.1% SDS.19 The SDS concentration
could be reduced simply by diluting the protein extract more
than 20-fold. This approach is simple, while some hydrophobic
proteins may precipitate because of the low SDS concentration.
Most SDS can be removed by dialysis or precipitation, but it is
time-consuming and the recovery is relatively low.20 It is also
well-known that the presence of SDS in the sample is deleteri-
ous to reversed-phase separation. Therefore, it is preferable to
develop protocols without utilizing SDS.
In this report, several different methods using lysis buffer
without SDS were investigated for the preparation of protein
samples from formalin-fixed tissues of mouse liver. We found
that efficient extraction of proteins from formalin-fixed tissue
could be achieved by incubation of the tissue in lysis buffer
containing 6 M guanidine-HCl at high temperature. In previous
reports, proteins identified from formalin-fixed tissue were
classified by Gene ontology, and this found proteins which had
a broad range of molecular functions and arose from every cell
compartment. But the identified peptides and proteins were
not comprehensively characterized. In this study, the peptides
and proteins identified from formalin-fixed tissue and frozen
fresh tissue were first comprehensively compared in term of
some physicochemical properties such as amino acid composi-
tion, hydrophobicity, pI, and so on. We found a majority of
the resulting peptides were unmodified, and slight difference
was observed between the proteins identified from formalin-
fixed tissue and frozen fresh tissue.
Experimental Section
Materials. Magic C18AQ (5 m, 100 pore) was purchased
from Michrom BioResources (Auburn, CA). All the water used
in the experiment was purified using a Mill-Q system (Millipore,
Bedford, MA). Dithiothreitol (DTT) and iodoacetamide were all
purchased from Sino-American Biotechnology Corporation
(Beijing, China). TPCK-trypsin, guanidine hydrochloride, and
sodium dodecyl sulfate (SDS) were obtained from Sigma (St.
Louis, MO). Tris was from Amersco (Solon, OH). Formic acid
was obtained from Fluka (Buches, Germany). Acetonitrile (ACN,
research articles
HPLC grade) was from Merck (Darmstadt, Germany). Cyanogen
bromide (CNBr) was purchased from Shenyang Chemical
Factory (Shenyang, China). C-57BL/6J mouse was from Dalian
Medical University. The mouse liver was fixed in 10% formalin
for 2 weeks,21-23 and the fresh mouse liver was stored at -80
C.
Protein Extraction and Proteolysis. Protein concentration
was determined by the Bradford assay. Unless otherwise stated,
the resulting tryptic digests were desalted with a C18 solid-
phase cartridge. Five different protocols were used to extract
proteins from mouse liver tissues; the first protocol was applied
to both frozen fresh tissue and formalin-fixed tissue, and the
remaining four protocols were applied only to formalin-fixed
tissue. The detailed procedures were as follows: (1) For the
protocol using 6 M guanidine-HCl without heating, mouse liver
tissue, either frozen or formalin-fixed tissue, was homogenized
in lysis buffer (40 mM Tris, 6 M guanidine-HCl, and 65 mM
DTT, pH 8.2) and then sonicated for 180 s followed by
centrifugation at 25 000g for 1 h. The supernatant was collected
as protein sample A for frozen tissue and sample B for formalin-
fixed tissue. The protein samples were reduced by DTT and
alkylated by iodoacetamide. Then the solutions were diluted
to 1 M guanidine-HCl, and the pH values were adjusted to 8.1.
Finally, trypsin was added (trypsin/protein, 1:50) and incubated
at 37 C for 20 h. (2) For the protocol using 2% SDS with
heating, formalin-fixed mouse liver tissue was homogenized
in lysis buffer (40 mM Tris and 2% SDS, pH 8.2), and then
sonicated for 180 s. Then the mixture was incubated at 100 C
for 20 min and 60 C for 2 h. After centrifugation at 25 000g for
1 h, the supernatant was collected as protein sample C. The
protein sample was reduced by DTT and alkylated by iodoac-
etamide. Then the solution was diluted to 0.1% SDS, and the
pH was adjusted to 8.1. Finally, trypsin was added (trypsin/
protein, 1:50) and incubated at 37 C for 20 h. The tryptic digest
was enrichment with an SCX trap column. (3) For the protocol
with direct digestion of tissue homogenate, formalin-fixed
mouse liver tissue was homogenized in lysis buffer (40 mM Tris,
6 M guanidine-HCl, and 65 mM DTT, pH 8.2) and then
sonicated for 180 s. The proteins in the homogenate were
reduced by DTT and alkylated by iodoacetamide. Then, the
solution was diluted to 1 M guanidine-HCl, and the pH was
adjusted to 8.1. Trypsin was added and incubated at 37 C for
20 h. The resulting tryptic digest was collected after centrifuga-
tion. The resulted sample was referred as sample D. (4) For
the protocol using 6 M guanidine-HCl with heating, formalin-
fixed mouse liver tissue was homogenized in lysis buffer (40
mM Tris, 6 M guanidine-HCl, and 65 mM DTT, pH 8.2) and
then sonicated for 180 s. The mixture was incubated at 100 C
for 30 min. After centrifugation, the supernatant was collected
as protein sample E, and the pellet was also collected for further
processing. The preparation of tryptic digest of sample E was
as described above. (5) For the protocol with CNBr treatment,
the pellet from the previous protocol was mixed 90% formic
acid and cyanogens bromide as reported.24 Briefly, 90% formic
acid was added to the pellet and incubated for 5 min at room
temperature. Then CNBr was added to the concentration of 1
g/mL, and the solution was incubated overnight at room
temperature in the dark. On the following day, the pH was
adjusted to 8.5. The sample F was lyophilized and then was
redissolved in the buffer containing 40 mM Tris and 6 M
guanidine-HCl. From this point forward, the sample was treated
identically to other protocols. The schematic of the protocols
was shown in Figure 1.
Journal of Proteome Research Vol. 6, No. 3, 2007 1039
research articles
Figure 1. Summary of protocols for protein extraction.
Table 1. Protein Concentration of Samples A, B, C, and E (n )
3)
sample A B C E
Protein 17.7 ( 1.2 0.82 ( 0.13 6.4 ( 0.66 10.9 ( 0.87
concentration
(mg/mL)
HPLC and Mass Spectrometry. Columns were packed using
a homemade pneumatic pressure cell at constant nitrogen gas
pressure of about 580 psi with a slurry packing method.25 For
the preparation of analytical column, one end of a 75 m i.d.
fused silica capillary was first manually pulled to a fine point
of 5 m with a flame torch. The C18 particles were then
packed until the packing section reached the length of 12 cm.
The HPLC-MS/MS system consisted of a quaternary Surveyor
pump, a Surveyor autosampler, and an LTQ linear ion trap
mass spectrometer equipped with a nanospray source (Thermo,
San Jose, CA). The two buffer solutions used for the quaternary
pump were 0.1% formic acid (mobile-phase A) and 99.9% ACN/
0.1% formic acid (mobile-phase B). The temperature of the ion-
transfer capillary was set at 200 C. The spray voltage was set
at 1.8 V, and the normalized collision energy was set at 35.0%.
An automated gain control function was used to manage the
number of ions injected into the ion trap. One microscan was
set for each MS and MS/MS scan. All MS and MS/MS spectra
were acquired in the data-dependent mode. The mass spec-
trometer was set so that 1 full MS scan was followed by 10 MS/
MS scans on the 10 most intense ions. The dynamic exclusion
function was set as follows: repeat count 2, repeat duration
30 s, and exclusion duration 90 s. System control and data
collection were done by Xcalibur software version 1.4 (Thermo).
Peptides were eluted using a linear gradient of 5% mobile-phase
B to 35% mobile-phase B in 120 min, then 80% mobile-phase
B in an additional 15 min, all at a flow rate of about 200 nL/
min.
Data Analysis. Each digest was analyzed three times by LC-
MS/MS, and data analysis was based on the cumulative total
proteins identified in three reduplicate analyses. The acquired
MS/MS spectra were searched against Mouse International
Protein Index (IPI) database (v3.17) using the TurboSEQUEST
in the BioWorks 3.2 software suite (Thermo). Reversed se-
quences were appended to the database for the evaluation of
false-positive rate. Cysteine residues were searched as static
modification of 57.0215 Da. Peptides were searched using fully
tryptic cleavage constraints, and up to two missed cleavages
sites were allowed for tryptic digestion. The mass tolerances
were 2 Da for parent masses and 1 Da for fragment masses.
1040 Journal of Proteome Research Vol. 6, No. 3, 2007
Jiang et al.
Table 2. Total Numbers of Peptides and Proteins Detected
from Frozen Mouse Liver Tissues and Formalin-Fixed Mouse
Liver Tissues
proteins identified
number of with two unique number of
sample identified proteinsa peptides minimumb unique peptides
A 976 480 3207
B 130 57 352
C 820 395 2540
D 331 106 589
E 827 470 3005
F 526 202 1129
a
Accepted fully tryptic peptides only. Xcorr at least 1.9, 2.2, and 3.75 for
singly, doubly, and triply charged peptide ions, respectively. Cn g 0.15.
False-positive rate less than 5%. b In addition of a, proteins identified with
at least two peptides for each protein were accepted.
The peptides were considered as positive identification if the
Xcorr were higher than 1.9 for singly charged peptide, 2.2 for
doubly charged peptide, and 3.75 for triply charged peptides,
and Cn cutoff values were g0.15. False-positive rates (FPR)
were calculated by using the following equation, FPR ) 2n(rev)/
[n(rev) + n(forw)], where n(forw) and n(rev) are the number
of peptides identified in proteins with forward (normal) and
reversed sequence, respectively.19,26 False-positive rate less than
5% was obtained for the peptide identifications by using
theparameters mentioned above. Proteins were assigned sub-
celluar localization and molecular functions using GoMiner, a
Web-based application based on Gene Ontology (GO) (discov-
er.nci.nih.gov/gominer).27 To improve the confidence of the
protein identification, only the proteins identified by two
peptides per protein minimum were subjected to GO clas-
sification. Grand average of hydropathy (GRAVY) values for
each unique protein were calculated according to the method
of Kyte and Doolittle.28
Results
Extraction of Proteins with Different Protocols. Efficient
extraction of proteins is the most critical step for proteome
analysis.29 To investigate if the routine protein preparation
protocol could be utilized to extract protein from formalin-
fixed tissue, the same protocol was applied to the preparation
of protein samples for fresh-frozen tissue and formalin-fixed
tissue of mouse liver. Both tissues were homogenized in the
lysis buffer containing 40 mM Tris, 6 M guanidine-HCl, and
65 mM DTT at 4 C. The protocol was referred as using 6 M
guanidine-HCl without heating. After centrifugation, the su-
pernatants were collected, and the protein concentrations were
determined by the Bradford assay. As shown in Table 1, protein
concentration as high as 17.7 mg/mL was obtained for the
protein extract prepared from frozen fresh tissue (sample A),
while only 0.82 mg/mL was obtained for formalin-fixed tissue
(sample B). The protein yield for the fixed tissue sample was
about 20-fold lower than that of frozen tissue sample. The
resulting tryptic digests were also subjected to LC-MS/MS
analysis. The database search results were listed in Table 2.
The analysis of the sample from frozen fresh tissue yielded 3207
unique peptides, which matched 976 unique proteins. While
the analysis of the sample from the fixed tissue yielded only
352 unique peptides, which matched 130 unique proteins. For
more confident identification, the numbers of proteins identi-
fied by at least two peptides were 480 and 57 for the frozen
sample and formalin-fixed sample, respectively. These results
show that the conventional protocol cannot efficiently extract
Shotgun Proteome Analysis of Formalin-Fixed Tissue
protein from formalin-fixed tissue. Since proteins undergo both
intra- and inter-protein covalent cross-linking resulting from
formalin fixation, it is reasonable to assume that they exist as
extensive cross-linked protein networks leading to the difficul-
ties in their efficient extraction.
On the basis of these results, proteins in the formalin-fixed
tissue are difficult to extract by the conventional method
because they are trapped in the network of cross-linked
proteins. On the other hand, the cross-linked protein network
may be disrupted by trypsin digestion. Therefore, formalin-
fixed tissue homogenates were directly digested with trypsin
to improve the result of protein identification. After incubation
with trypsin for 20 h at 37 C, the tissue homogenate was
subjected to centrifugation, and the supernatant was collected
for LC-MS/MS analysis. After database searching, 589 unique
peptides were matched with MS/MS spectra, which resulted
in the identification of 331 unique proteins. Among the 331
unique proteins, 106 proteins were matched by two or more
peptides (Table 2). Compared with the conventional protocol
where only 57 proteins that matched two peptide minimum
were identified from the formalin-fixed tissue, this protocol was
more effective. But it is far from ideal compared with the
number of proteins identified from the frozen fresh tissue.
To further improve the efficiency of protein extraction from
formalin-fixed tissue and considering the cross-linking of
proteins in these tissues, the violent conditions, such as heating,
should be tried to increase the recovery of proteins. Thus, the
heating treatment was applied to the protocol with 6 M
guanidine-HCl. After the formalin-fixed mouse liver tissue was
homogenized in lysis buffer (40 mM Tris, 6 M guanidine-HCl,
and 65 mM DTT), the resulting mixture was incubated at 100
C for 30 min. The protein concentration was determined to
be 10.9 mg/mL, which was comparable with that from frozen
fresh tissue. After LC-MS/MS analysis of the resulting tryptic
digest, 3005 unique peptides and 827 unique proteins were
identified. The number of proteins matched with at least two
peptides was 470, which was slightly lower than that for fresh-
frozen tissue. These results indicated that the protocol using 6
M guanidine-HCl with heating was very efficient for protein
extraction from formalin-fixed tissue. The results obtained from
formalin-fixed tissue were comparable to those obtained from
the frozen fresh tissue.
Furthermore, due to the cross linking of proteins by formalin
leading to the insolubility of some proteins, there may be some
cross-linked proteins that still remained in the pellet. To probe
the tissue proteome in depth, the pellet was also processed for
LC-MS/MS analysis. Besides the highly cross-linked protein
complex, the hydrophobic membrane proteins were also
presented in the pellet. It was reported that the membrane
proteins could be cleaved off by cyanogens bromide (CNBr),
and the identification of membrane proteins could be achieved
by LC-MS/MS analysis.24 Similarly, the unlinked portions of
the insoluble highly cross-linked complex may also be cleaved
off by CNBr for shotgun proteome analysis. Therefore, the
pellets were treated with CNBr as the procedure described in
the Experimental Section. The resulting fragments from these
insoluble proteins were further digested by trypsin for LC-MS/
MS analysis. After database searching, 1129 unique peptides
were matched to result in the identification of 526 unique
proteins. Among those proteins, 202 proteins were identified
by two peptides minimum (Table 2).
Previous reported studies of proteins extraction from for-
malin-fixed tissues for shotgun proteomics was mainly based
research articles
on utilizing lysis buffer containing 2% SDS with heating.15,18,30
The protocol was also applied to prepare protein extract for
formalin-fixed mouse liver tissue in this study. The tissue was
boiled in a solution of Tris-HCl containing 2% SDS for 20 min
followed by incubation at 60 C for 2 h. The protein concentra-
tion of the resulting extract was determined to be 6.4 mg/mL
(Table 1), which was lower than that of the protocol utilizing 6
M guanidine-HCl with heating. After LC-MS/MS analysis of
the resulting tryptic digest, the identification of 2540 unique
peptides corresponding to 820 unique proteins was achieved
as shown in Table 2. Among those identified proteins, 395
proteins were identified by at least two peptides.
As described above, five different sample preparation pro-
tocols were investigated for extraction of proteins or peptides
from formalin-fixed tissues. As can be seen from Tables 1 and
2, the protocol using 6 M guanidine-HCl with heating yields
the best results, and the protocol using 2% SDS with heating
yields the second-best results. But the protocols without
heating, that is, the protocol using 6 M guanidine-HCl at low
temperature and the protocol with direct digestion of the tissue
homogenates by trypsin, yielded relatively poor results. Heating
treatment of formalin-fixed tissue was typically applied to
retrieve antigen in IHC because of the hydrolysis of cross-links
between formalin and protein at high temperature.31 The
breakdown of the formalin-induced cross-links at high tem-
perature also explains the reason the protocols with heating
are more efficient for the extraction of proteins from formalin-
fixed tissues. However, efficient protein extraction from for-
malin-fixed tissue cannot be achieved only by the heating
treatment. The buffer containing only 40 mM Tris (pH 8.2) was
also utilized to extract protein from such tissues with the same
condition as the protocol using 6 M guanidine-HCl with heating
or using 2% SDS with heating, and very few proteins were
extracted. Therefore, proper lysis buffer and heating conditions
are both vitally important for protein extraction from formalin-
fixed tissue.
Characterization of Identified Peptides from Frozen Tissue
and Formalin-Fixed Tissue. The process of tissue fixation by
formalin, a formaldehyde solution, has been divided into two
stages:22,31 the primary reaction is an addition reaction between
a primary amine group on the protein molecule and an
aldehyde group on formaldehyde, R-NH2 + HCHO f R-NH-
CH2OH, followed by a secondary condensation reaction, R-NH-
CH2OH + H2N-CO-R f R-NH-CH2-NH-CO-R + H2O.
Therefore, proteins in the fixed tissue are primary cross-linked
via the side group of lysine residues. Since proteins undergo
both intra- and inter-protein covalent cross-linking, it is
reasonable that the extraction of these proteins at low tem-
perature is not efficient. As we demonstrated, the proteins in
the formalin-fixed tissue could be efficiently extracted with
heating due to the breakdown of some cross-links. Trypsin
digestion of the resulting proteins will produce three kinds of
peptides: peptides without modifications, peptides with some
amino acid residues modified, and peptides that are cross-
linked to other peptides or form internal cross-links. The cross-
linked peptides cannot be identified by the Sequest database
search algorithm. The identification of unmodified peptides can
be easily achieved by Sequest database searching with standard
parameters, while the identification of the modified peptides
could only be achieved if the chemistry of the modification is
clear. In accordance with the above-mentioned chemical
reactions, the primary amine of lysine is modified by CH2OH;
therefore, a variable modification with 30 Da was set for the
Journal of Proteome Research Vol. 6, No. 3, 2007 1041
research articles
lysine for Sequest search of the MS/MS data obtained from
formalin-fixed tissue. Because trypsin cannot cleave peptides
with the modified lysine residue, the percentage of tryptic
peptides with lysyl termini must be dramatically decreased for
formalin-fixed tissue if a majority of lysine residues were
modified. The search results showed that only about 5.0% of
the identified peptides containing modified lysyl residues and
a majority of lysine residues were still unmodified for the
proteins obtained from formalin-fixed tissue. Although other
modifications may also occur to the lysine residue and other
amino acid residues, because the process of formalin fixation
and the influence of heating treatment are not well-understood,
and the chemistry of modified peptides was not clear, no
modification was set for the database search for the charac-
terization of identified peptides/proteins from frozen tissue and
formalin-fixed tissue in this study.
An important issue when using shotgun proteomics for
analysis of formalin-fixed tissue sample is whether the covalent
modifications resulting from formalin fixation affect the pro-
teome analysis results. Therefore, the results obtained from the
analysis of formalin-fixed tissue and frozen fresh tissue should
be compared. In this study, the proteins in frozen fresh tissue
were extracted using buffer containing 6 M guanidine-HCl
without heating. As we described above, the extraction of
proteins from formalin-fixed tissue was also efficient when the
buffer containing 6 M guanidine-HCl was used in combination
with heating treatment. Table 2 shows the numbers of identi-
fied peptides and proteins were comparable for these two cases.
Because the same buffer was utilized and a similar number of
proteins and peptides identified, the comparison for the
proteome analysis results on the peptide and protein levels was
based on the two datasets mentioned above (Sample A vs
Sample E in Table 2; for a complete lists, see Supporting
Information data 1 and 2). And 55% of the identified proteins
(281 proteins) were observed in both samples (for complete
list, see Supporting Information data 3).
Besides the modifications to lysine residues, the modification
of other residues may also happen during the formalin fixation.
Furthermore, the heating treatment for extraction of proteins
may lead to the occurrence of additional modifications. There-
fore, it is necessary to evaluate the influence of these modifica-
tions on the shotgun proteome analysis of formalin-fixed tissue.
If the modification of a certain residue did happen in the
formalin-fixed tissue sample, the number of unmodified pep-
tides containing the residue must decrease, and thus, the
percentage of the residue among all the identified peptides
should also decrease. The total numbers of amino acid residues
in those identified unique peptides from the frozen tissue
sample and the formalin-fixed tissue sample were 24 310 and
23 738, respectively. The percentages of the number of indi-
vidual amino acid residues among the total number of amino
acid residues in the identified unmodified peptides from frozen
tissue and fixed tissue were shown in Figure 2. The percentages
of lysine residues are 6.33% and 5.73% for frozen sample and
formalin-fixed sample, respectively. The percentage of lysine
residues for the formalin-fixed sample is about 10% lower than
that for the frozen sample, which indicated that a portion of
lysine residues was modified during formalin fixation. This is
consistent with the results derived from the decreased percent-
age of peptides with lysyl termini. Besides that of lysine residue,
the percentages of phenylalanine (F), asparagine (N), tryp-
tophan (W), and tyrosine (Y) residues in the peptides identified
from fixed tissue sample were obviously lower than those from
1042 Journal of Proteome Research Vol. 6, No. 3, 2007
Jiang et al.
Figure 2. Distribution of amino acid residues in the peptides
identified from frozen mouse liver tissue and formalin-fixed
mouse liver tissue.
frozen tissue sample, which indicated that the number of
unmodified peptides containing these residues decreased in
the fixed tissue sample. The decrease of lysine residue percent-
age is largely because due to formalin modifying this residue.
Since the other residues do not likely react with formaldehyde,
the decreasing of their percentages may be caused by the
heating treatment, which results in the degradation of these
residues. Figure 2 shows that all the amino acid residues were
observed in the formalin-fixed tissue sample and the percent-
ages of these residues were quite similar to those of frozen
tissue. Except the rare amino acid residue, tryptophan, which
account for only about 1% in protein sequence, the decrease
of the percentages of other residues were not significant. This
means that a majority of peptides obtained from formalin-fixed
tissue were unmodified. This is the reason that shotgun
proteome analysis approach could be successfully applied to
analysis of formalin-fixed tissues.
Characterization of Identified Proteins from Frozen Tissue
and Formalin-Fixed Tissue. Both the formalin fixation process
and the heating treatment for sample preparation will lead to
the modifications of side chains of some proteins, and might
even induce the degradation of some proteins. An important
issue for proteome analysis of formalin-fixed tissue is whether
the integrity of the proteome is compromised by these modi-
fications. To answer this question, the identified proteins from
frozen fresh liver tissue and those from formalin-fixed tissue
should be comprehensively compared.
As referred above, the amino acid composition of peptides
identified from formalin-fixed tissue was slightly different from
that of peptides identified from frozen tissue because of amino
acid modifications induced by formalin reaction and heating
treatment. However, at the protein level, do such modifications
affect proteins identified from formalin-fixed tissue in contrast
with frozen tissue? To answer this question, the amino acid
composition of proteins identified from frozen tissue and
formalin-fixed tissue was investigated in detail. We found that
the distributions of the average amino acid composition of
proteins identified from frozen fresh tissue and formalin-fixed
tissue, as shown in Figure 3, were very similar. Slight discrep-
ancies were observed for some amino acids. For example,
proteins with higher percentage of A, K, and R residues were
observed for the formalin-fixed sample, while proteins with
higher percentage of D, F, and Y residues were observed for
the frozen tissue. Interestingly, the discrepancies in the amino
Shotgun Proteome Analysis of Formalin-Fixed Tissue
Figure 3. Distribution of average amino acid composition for the
proteins identified from frozen mouse liver tissue and formalin-
fixed mouse liver tissue.
acid composition were different between the protein level and
peptide level. As we described above, less peptides containing
lysine residue (K) were identified in formalin-fixed tissue
because of the modifications induced by the formalin fixation.
But at the protein level, proteins with a higher percentage of
lysine residue were identified in formalin-fixed tissue. This is
surprising, considering the proteins with higher percentage of
K residues are more heavily cross-linked and. therefore. these
proteins should not be easily extracted nor accessible for
trypsin. But the opposite results were observed in this study.
This probably can be explained by the fact that the most of
the cross-links were opened by the heating treatment during
the protein extraction step. Overall, the profiles for the distribu-
tions of proteins with different amino acid composition were
almost identical, which indicates the modifications of some
amino acid residues induced by the formalin fixation hardly
impact the integrity of the proteome for the analysis of fixed
tissue.
As addressed above, proteins with relative higher percentage
of K and R residues were identified from formalin-fixed tissue.
Since the K and R residues are very hydrophilic, the proteins
identified from formalin-fixed tissue probably are more hy-
drophilic than those identified from frozen tissue. To compare
the hydrophobicity of the identified proteins, the grand average
hydrophobicity (GRAVY) values for all proteins identified from
the fixed tissue and frozen tissue were determined according
to Kyte and Doolittle.28 The higher the GRAVY value is, the more
hydrophobic the protein is. The results are shown in Figure 4.
More proteins were identified from fixed tissue in the range of
GRAVY value < -0.5, which indicated that the more hydrophilic
proteins were identified from formalin-fixed tissue. The average
GRAVY values for the proteins identified from the fixed tissue
and frozen tissue were -0.333 and -0.287, respectively. These
results indicate that shotgun proteome analysis of formalin-
fixed tissue is slightly bias toward hydrophilic proteins.
The distributions of isoelectric point (pI) and molecular
weight (MW), which were also important physicochemical
properties of proteins, were examined. Proteins identified from
frozen tissue and formalin-fixed tissue all spread in a broad
range of pI and MW as shown in Figure 5. However, there was
some difference in specific scope of pI and MW. The percentage
of proteins identified from formalin-fixed tissue is higher than
that from frozen tissue in the scope of 4-6 and >10 of pI values.
And in the range of 6-10 of pI values, the percentage of
research articles
Figure 4. Distribution of Grand average of hydropathy (GRAVY)
of proteins identified from frozen mouse liver tissue and formalin-
fixed liver tissue.
proteins identified from formalin-fixed tissue is lower than that
from frozen tissue. The average pI values for the proteins
identified from formalin-fixed tissue and frozen tissue were 7.44
and 7.26, respectively. More basic proteins were identified from
formalin-fixed tissue. This is consistent with the distribution
of identified protein with different amino acid composition as
shown in Figure 3, where more proteins with high percentage
of basic amino acid residues, K and R, were identified from
the formalin-fixed tissue. For the difference in the distribution
of MW as shown in Figure 5, more proteins were identified
from formalin-fixed tissue in the range of 15-30 kDa. The
average MW of the proteins identified from formalin-fixed and
frozen tissue was 55.8 and 58.9 kDa, respectively. The proteins
identified from the formalin-fixed tissue are slightly biased
toward low MW.
The proteins identified in the analysis of fresh frozen tissue
and formalin-fixed tissue were subjected to classification using
Gene Ontology. The resulting cellular localization of proteins
was shown in Figure 6. Similar to those identified from fresh
frozen tissue, the proteins identified from the formalin-fixed
tissue also came from different cell compartments. Slightly less
membrane proteins were identified from formalin-fixed tissue.
The proteins identified from the formalin-fixed tissue also have
a broad range of molecular functions as shown in Figure 6.
Compared with frozen tissue, more proteins with molecular
structural function were identified from formalin-fixed tissue.
Although a slight bias was observed, broad protein coverage
could be obtained by analysis of formalin-fixed tissue. Pro-
teome samples are typically very complex, and mass spectrom-
eter cannot collect tandem mass spectra from all eluting
peptides because of limited scan rate. The proteins identified
by the shotgun proteomics approach with different LC-MS/
MS runs are not exactly the same even for the same sample.
Liu et al.32 compared the proteins identified from 9 replicate
SCX-RPLC-MS/MS runs for a same yeast protein digest sample
and found that about 35.4% were identified in every run.
However, in this study two different proteome samples, one
from formalin-fixed tissue and another from frozen tissue, were
analyzed, and it was found that more than 55% of the identified
proteins (based on two peptides or more for each protein) were
the same. Therefore, the proteome analysis results obtained
from formalin-fixed tissue were quite comparable to those
obtained from frozen tissue. The fact that proteome coverage
obtained from formalin-fixed tissue sample by shotgun pro-
Journal of Proteome Research Vol. 6, No. 3, 2007 1043
research articles
Figure 5. Distribution of molecular weight (MW) and pI of proteins identified from frozen mouse liver tissue and formalin-fixed mouse
liver tissue. (A) MW distribution with 5 kDa increments; (B) pI distribution of the proteins with 0.5 pH unit increments.
teome analysis was not compromised because of formalin
fixation indicated that the archived fixed tissues might be good
samples for biomarker discovery instead of frozen-fresh tissue.
Discussion
Fresh frozen tissue is difficult to obtain in large numbers,
whereas extensive archives of animal and human well-defined
formalin-fixed tissue which contains associated experimental
and clinical information representing a highly valuable reservoir
of potential biomarkers are readily available. Unfortunately,
effective and routine analysis of proteins in formalin-fixed
tissue has been limited to IHC. This limitation is due to the
chemical cross-linking properties of the common fixative
reagent of formalin. The proteins extracted from the formalin-
fixed tissue have different physicochemical properties from
those of fresh proteins. For example, the pI values of proteins
will change because of the decrease of primary amines resulting
from the reaction of lysine side chains with formaldehyde; the
MW of the extracted proteins may also increase because of the
possible cross-link with other proteins and biomolecules.
Moreover, the reactions between protein and formaldehyde are
1044 Journal of Proteome Research Vol. 6, No. 3, 2007
Jiang et al.
not likely gone completely, that means a series of products may
result for a single protein. Because of above reasons, the
conventional protein analysis approaches are not suitable for
the analysis of proteins extracted from formalin-fixed tissue.
Because of the poor results obtained from 2D-PAGE resolution
of proteins extracted from formalin-fixed tissue, Ahram et al.33
concluded that the utility of formalin-fixed tissue for protein
separation studies was limited.
Because of the development of the shotgun proteomics
strategy in recent years, it has become possible to conduct
discovery-driven investigation in formalin-fixed tissue.10 The
separation takes place at the protein level in conventional
protein analysis approaches, while the separation takes place
at the peptide level in shotgun proteomics. The proteins are
digested into peptides, and then these peptides are analyzed
by LC-MS/MS in shotgun proteomics. Although some side
chains of protein are modified by formaldehyde in the formalin-
fixed tissue, many unmodified peptides are generated after
trypsin digestion. The identification of proteins can be easily
achieved by LC-MS/MS analysis of the unmodified peptides.
In this study, all 20 amino acid residues were found in the
Shotgun Proteome Analysis of Formalin-Fixed Tissue
Figure 6. Gene ontology cellular location and molecular function
of the identified proteins from frozen mouse liver tissue and
formalin-fixed mouse liver tissue. (A) Cellular location; (B)
molecular function.
peptides identified from formalin-fixed tissue, and the distribu-
tion of these amino acid residues was very similar to that of
frozen tissue. These results indicated that a majority of peptides
from formalin-fixed tissue were unmodified. This is the reason
shotgun proteomics approach could be successfully applied to
the analysis of formalin-fixed tissue.
Recently, increased attention was paid to shotgun proteome
analysis of formalin-fixed tissue.10,15-18,30 Most of these studies
use SDS lysis buffer to extract proteins from formalin-fixed
tissue. We found in this study that proteins can be more
efficiently extracted from the fixed tissue using 6 M guanidine-
HCl with heating. Considering highly cross-linked protein
complex and hydrophobic membrane proteins still presented
in the pellet, the remaining pellet was further treated with
CNBr. Although the protocol with CNBr treatment resulted in
a relatively low number of protein identifications, it generated
about 17% of membrane proteins which was a much higher
number than that obtained using the 6 M guanidine-HCl
protocol. To comprehensively analyze the proteome of forma-
lin-fixed tissue, the use of multiple sample preparation proto-
cols together is preferable. Combining the proteins identified
through the three protocols, that is, 6 M guanidine-HCl with
heating, 2% SDS with heating, and CNBr treatment, resulted
in the identification of 772 unique proteins with two peptides
minimum for each protein.
The proteins identified in the formalin-fixed tissue/cell and
frozen-fresh tissue/cell were also classified by Gene Ontology
in previous reports.15-17 Similarly to this study, a slight bias was
observed, but the overall profiles of the protein distributions
in term of molecular function and cellular localization were
research articles
quite similar. The difference in the amino acid composition of
proteins identified from the two samples was first investigated
in this study, and we found the distributions were also quite
similar. Because relatively more proteins with high percentage
of K and R residues were observed in formalin-fixed tissue, the
proteins identified from formalin-fixed tissue were slightly more
hydrophilic and basic. A slight bias toward low MW proteins
for shotgun proteome analysis of fixed tissue was also observed.
But for all the properties of protein investigated, including
amino acid composition, hydrophobicity, pI value, and MW,
the overall distribution was all similar, which indicated that
the formalin fixation dose not adversely affect the diversity of
proteins identified from tissue sample. The analysis of protein
extract prepared by the protocol of 6 M guanidine-HCl with
heating for formalin-fixed tissue resulted in the identification
of 470 unique proteins based on two or more peptides for each
protein. The number of identified proteins was comparable to
that identified from frozen-fresh tissue, and 55% of identified
proteins were observed in both samples. All of these results
indicated that similar proteome coverage could be obtained
by shotgun proteome analysis of formalin-fixed tissue.
Moreover, notably, the incomplete overlap of the proteins
identified from both samples was observed. As previous reports
described,11,34,35 in any given large-scale proteomic experiment,
only a subset of the entire proteome was identified due to
technical limitations of current proteomic technologies, and
shotgun proteome approaches might expect to identify different
peptide subsets from identical biological samples, even across
replicate analyses. This was a main reason that peptides/
proteins identified from frozen tissue and formalin-fixed tissue
were not exactly the same. To comprehensively analyze the
proteome of formalin-fixed tissue, more effective LC-MS
technologies and the variety of fractionation approaches should
be applied in routine implementation of shotgun proteomic
profiling platform. Furthermore, because of the importance of
post-translational modifications, such as phosphorylation and
glycosylation, further studies should be performed to investi-
gate if the actual form of the protein post-modifications were
preserved in formalin-fixed tissue.
As valuable proteome information can be retrieved from
formalin-fixed tissue samples, the shotgun proteome analysis
approach provides the ability to unlock the proteome of the
worlds vast reservoir of archived tissue for large-scale discovery
and validation of biomarkers to improve disease diagnosis and
therapy. As we know, the archived formalin-fixed tissue were
typically collected at different place and time. Before the
formalin-fixed tissue could be used as an alternative to frozen-
fresh tissue for biomarker discovery, studies examining forma-
lin-fixed tissues that have been stored for varying lengths of
time and collected from different labs should be conducted to
determine if these conditions affect proteome analysis results.
Conclusion
The development of methods to use state-of-the-art pro-
teomic discovery tools to analyze formalin-fixed tissue provides
an exciting new opportunity to identify disease-specific biom-
arkers in pathologically defined samples. In this study, proto-
cols for the efficient extraction of proteins from formalin-fixed
tissue were developed, which enabled the use of high-
throughput shotgun proteome analysis approach for large-scale
and in-depth analysis of proteins presented in archived tissue
Journal of Proteome Research Vol. 6, No. 3, 2007 1045
research articles
samples. The peptides and proteins identified from formalin-
fixed tissue were comprehensively characterized. It was found
that the formalin fixation does not compromise the proteome
coverage of the analysis.
Acknowledgment. Financial supports from the National
Natural Sciences Foundation of China (No. 20327002), the
China State Key Basic Research Program Grant (2005CB522701),
and the Knowledge Innovation program of DICP to H.Z. are
gratefully acknowledged.
Supporting Information Available: Tables listing the
complete datasets for the comparison of the proteome analysis
results on the peptide and protein levels, and the dataset of
the identified proteins observed in both samples. This material
is available free of charge via the Internet at http://pubs.acs.org.
References
(1) Edgar, P. F.; Schonberger, S. J.; Dean, B.; Faull, R. L.; Kydd, R.;
Cooper, G. J. A comparative proteome analysis of hippocampal
tissue from schizophrenic and Alzheimers disease individuals.
Mol. Psychiatry 1999, 4, 173-178.
(2) Kim, J.; Kim, S. H.; Lee, S. U.; Ha, G. H.; Kang, D. G.; Ha, N. Y.;
Ahn, J. S.; Cho, H. Y.; Kang, S. J.; Lee, Y. J.; Hong, S. C.; Ha, W. S.;
Bae, J. M.; Lee, C. W.; Kim, J. W. Proteome analysis of human
liver tumor tissue by two-dimensional gel electrophoresis and
matrix assisted laser desorption/ionization-mass spectrometry for
identification of disease-related proteins. Electrophoresis 2002,
23, 4142-4156.
(3) Chen, J.; Kahne, T.; Rocken, C.; Gotze, T.; Yu, J.; Sung, J. J.; Chen,
M.; Hu, P.; Malfertheiner, P.; Ebert, M. P. Proteome analysis of
gastric cancer metastasis by two-dimensional gel electrophoresis
and matrix assisted laser desorption/ionization-mass spectrom-
etry for identification of metastasis-related proteins. J. Proteome
Res. 2004, 3, 1009-1016.
(4) Friedman, D. B.; Hill, S.; Keller, J. W.; Merchant, N. B.; Levy, S.
E.; Coffey, R. J.; Caprioli, R. M. Proteome analysis of human colon
cancer by two-dimensional difference gel electrophoresis and
mass spectrometry. Proteomics 2004, 4, 793-811.
(5) Lee, I. N.; Chen, C. H.; Sheu, J. C.; Lee, H. S.; Huang, G. T.; Yu, C.
Y.; Lu, F. J.; Chow, L. P. Identification of human hepatocellular
carcinoma-related biomarkers by two-dimensional difference gel
electrophoresis and mass spectrometry. J. Proteome Res. 2005,
4, 2062-2069.
(6) Davidsson, P.; Paulson, L.; Hesse, C.; Blennow, K.; Nilsson, C. L.
Proteome studies of human cerebrospinal fluid and brain tissue
using a preparative two-dimensional electrophoresis approach
prior to mass spectrometry. Proteomics 2001, 1, 444-
452.
(7) Ha, G. H.; Lee, S. U.; Kang, D. G.; Ha, N. Y.; Kim, S. H.; Kim, J.;
Bae, J. M.; Kim, J. W.; Lee, C. W. Proteome analysis of human
stomach tissue: separation of soluble proteins by two-dimen-
sional polyacrylamide gel electrophoresis and identification by
mass spectrometry. Electrophoresis 2002, 23, 2513-
2524.
(8) Wang, H.; Kachman, M. T.; Schwartz, D. R.; Cho, K. R.; Lubman,
D. M. Comprehensive proteome analysis of ovarian cancers using
liquid phase separation, mass mapping and tandem mass
spectrometry: a strategy for identification of candidate cancer
biomarkers. Proteomics 2004, 4, 2476-2495.
(9) Liang, C. R.; Leow, C. K.; Neo, J. C.; Tan, G. S.; Lo, S. L.; Lim, J.
W.; Seow, T. K.; Lai, P. B.; Chung, M. C. Proteome analysis of
human hepatocellular carcinoma tissues by two-dimensional
difference gel electrophoresis and mass spectrometry. Proteomics
2005, 5, 2258-2271.
(10) Hood, B. L.; Conrads, T. P.; Veenstra, T. D. Mass spectrometric
analysis of formalin-fixed paraffin-embedded tissue: unlocking
the proteome within. Proteomics 2006, 6, 4106-4114.
(11) Nesvizhskii, A. I.; Aebersold, R. Interpretation of shotgun pro-
teomics data: the protein inference problem. Mol. Cell. Proteom-
ics 2005, 4, 1419-1440.
(12) Feng, S.; Ye, M.; Jiang, X.; Jin, W.; Zou, H. Coupling the
immobilized trypsin microreactor of monolithic capillary with
RPLC-MS/MS for shotgun proteome analysis. J. Proteome Res.
2006, 5, 422-428.
1046 Journal of Proteome Research Vol. 6, No. 3, 2007
Jiang et al.
(13) Xie, C.; Ye, M.; Jiang, X.; Jin, W.; Zou, H. Octadecylated silica
monolith capillary column with integrated nanoelectrospray
ionization emitter for highly efficient proteome analysis. Mol. Cell.
Proteomics 2006, 5, 454-461.
(14) Prieto, D. A.; Hood, B. L.; Darfler, M. M.; Guiel, T. G.; Lucas, D.
A.; Conrads, T. P.; Veenstra, T. D.; Krizman, D. B. Liquid Tissue:
proteomic profiling of formalin-fixed tissues. BioTechniques 2005,
Suppl., 32-35.
(15) Hood, B. L.; Darfler, M. M.; Guiel, T. G.; Furusato, B.; Lucas, D.
A.; Ringeisen, B. R.; Sesterhenn, I. A.; Conrads, T. P.; Veenstra, T.
D.; Krizman, D. B. Proteomic analysis of formalin-fixed prostate
cancer tissue. Mol. Cell. Proteomics 2005, 4, 1741-
1753.
(16) Crockett, D. K.; Lin, Z.; Vaughn, C. P.; Lim, M. S.; Elenitoba-
Johnson, K. S. Identification of proteins from formalin-fixed
paraffin-embedded cells by LC-MS/MS. Lab. Invest. 2005, 85,
1405-1415.
(17) Shi, S. R.; Liu, C.; Balgley, B. M.; Lee, C.; Taylor, C. R. Protein
extraction from formalin-fixed, paraffin-embedded tissue sec-
tions: quality evaluation by mass spectrometry. J. Histochem.
Cytochem. 2006, 54, 739-743.
(18) Hwang, S. I.; Thumar, J.; Lundgren, D. H.; Rezaul, K.; Mayya, V.;
Wu, L.; Eng, J.; Wright, M. E.; Han, D. K. Direct cancer tissue
proteomics: a method to identify candidate cancer biomarkers
from formalin-fixed paraffin-embedded archival tissues. Oncogene
2007, 26, 65-76.
(19) Peng, J. M.; Elias, J. E.; Thoreen, C. C.; Licklider, L. J.; Gygi, S. P.
Evaluation of multidimensional chromatography coupled with
tandem mass spectrometry (LC/LC-MS/MS) for large-scale
protein analysis: the yeast proteome. J. Proteome Res. 2003, 2,
43-50.
(20) Zheng, X.; Baker, H.; Hancock, W. S. Analysis of the low molecular
weight serum peptidome using ultrafiltration and a hybrid ion
trap-Fourier transform mass spectrometer. J. Chromatogr., A.
2006, 1120, 173-184.
(21) Costa, P. P.; Jacobsson, B.; Collins, V. P.; Biberfeld, P. Unmasking
antigen determinants in amyloid. J. Histochem. Cytochem. 1986,
34, 1683-1685.
(22) Kiernan, J. Formaldehyde, formalin, paraformaldehyde and glu-
taraldehyde: What they are and what they do. Microsc. Today
2000, 00-1, 8-12.
(23) Pileri, S. A.; Roncador, G.; Ceccarelli, C.; Piccioli, M.; Briskomatis,
A.; Sabattini, E.; Ascani, S.; Santini, D.; Piccaluga, P. P.; Leone,
O.; Damiani, S.; Ercolessi, C.; Sandri, F.; Pieri, F.; Leoncini, L.;
Falini, B. Antigen retrieval techniques in immunohistochemis-
try: comparison of different methods. J. Pathol. 1997, 183, 116-
123.
(24) Washburn, M. P.; Wolters, D.; Yates, J. R., III. Large-scale analysis
of the yeast proteome by multidimensional protein identification
technology. Nat. Biotechnol. 2001, 19, 242-247.
(25) Yi, E. C.; Lee, H.; Aebersold, R.; Goodlett, D. R. A microcapillary
trap cartridge-microcapillary high-performance liquid chroma-
tography electrospray ionization emitter device capable of pep-
tide tandem mass spectrometry at the attomole level on an ion
trap mass spectrometer with automated routine operation.
Rapid Commun. Mass Spectrom. 2003, 17, 2093-
2098.
(26) Xie, H.; Bandhakavi, S.; Griffin, T. J. Evaluating preparative
isoelectric focusing of complex peptide mixtures for tandem mass
spectrometry-based proteomics: a case study in profiling chro-
matin-enriched subcellular fractions in Saccharomyces cerevisiae.
Anal. Chem. 2005, 77, 3198-3207.
(27) Zeeberg, B. R.; Feng, W.; Wang, G.; Wang, M. D.; Fojo, A. T.;
Sunshine, M.; Narasimhan, S.; Kane, D. W.; Reinhold, W. C.;
Lababidi, S.; Bussey, K. J.; Riss, J.; Barrett, J. C.; Weinstein, J. N.
GoMiner: a resource for biological interpretation of genomic and
proteomic data. GenomeBiology 2003, 4, R28.
(28) Kyte, J.; Doolittle, R. F. A simple method for displaying the
hydropathic charater of a protein. J. Mol. Biol. 1982, 157, 105-
132.
(29) Saravanan, R. S.; Rose, J. K. A critical evaluation of sample
extraction techniques for enhanced proteomic analysis of recal-
citrant plant tissues. Proteomics 2004, 4, 2522-2532.
(30) Palmer-Toy, D. E.; Krastins, B.; Sarracino, D. A.; Nadol, J. B., Jr.;
Merchant, S. N. Efficient method for the proteomic analysis of
fixed and embedded tissues. J. Proteome Res. 2005, 4, 2404-2411.
(31) Shi, S. R.; Cote, R. J.; Taylor, C. R. Antigen retrieval immunohis-
tochemistry: past, present, and future. J. Histochem. Cytochem.
1997, 45, 327-343.
Shotgun Proteome Analysis of Formalin-Fixed Tissue
(32) Liu, H.; Sadygov, R. G.; Yates, J. R., III. A model for random
sampling and estimation of relative protein abundance in
shotgun proteomics. Anal. Chem. 2004, 76, 4193-4201.
(33) Ahram, M.; Flaig, M. J.; Gillespie, J. W.; Duray, P. H.; Linehan, W.
M.; Ornstein, D. K.; Niu, S.; Zhao, Y.; Petricoin, E. F., III; Emmert-
Buck, M. R. Evaluation of ethanol-fixed, paraffin-embedded
tissues for proteomic applications. Proteomics 2003, 3, 413-
421.
research articles
(34) Chen, R.; Pan, S.; Brentnall, T. A.; Aebersold, R. Proteomic
profiling of pancreatic cancer for biomarker discovery. Mol. Cell.
Proteomics 2005, 4, 523-533.
(35) Elias, J. E.; Haas, W.; Faherty, B. K.; Gygi, S. P. Comparative
evaluation of mass spectrometry platforms used in large-scale
proteomics investigations. Nat. Methods 2005, 2, 667-675.
PR0605318
Journal of Proteome Research Vol. 6, No. 3, 2007 1047