Anatomic Pathology / A MOLECULAR MODEL OF FORMALIN FIXATION
A Molecular Mechanism of Formalin Fixation
and Antigen Retrieval
Seshi R. Sompuram, PhD,1,2 Kodela Vani, MS,2 Elizabeth Messana,2 and Steven A. Bogen, MD, PhD1,2
Key Words: Formalin fixation; Antigen retrieval; Amino acid; Phage display; Immunohistochemistry; Peptide
DOI: 10.1309/BRN7CTX1E84NWWPL
Abstract
Despite the popularity of antigen-retrieval
techniques, the precise molecular mechanism
underlying the process remains enigmatic. We examined
the molecular features underlying the loss of
immunoreactivity following formalin fixation, with
subsequent recovery by antigen retrieval. To do this, we
first created a molecular model using short peptides
that mimic the antibody-binding site of common clinical
protein targets. The advantage of this model is that we
know the amino acid sequence in and around the
antibody-binding site. We observed that some, not all,
of the peptides exhibited the formalin-fixation and
antigen-retrieval phenomenon. Other peptides did not
lose their ability to be recognized by antibody, even
after prolonged incubation in formalin. A third,
intermediate group exhibited the formalin-fixation and
antigen-retrieval phenomenon only if another irrelevant
protein was mixed with the peptide before fixation.
Amino acid sequence analysis indicates that fixation
and antigen retrieval are associated with a tyrosine in
or near the antibody-binding site and with an arginine
elsewhere, implicating the Mannich reaction as
important in fixation and antigen retrieval.
190
190
Am J Clin Pathol 2004;121:190-199
DOI: 10.1309/BRN7CTX1E84NWWPL
In 1991, Shi and coworkers1 described their finding that
boiling tissue sections in heavy metal solutions reversed the
formalin-fixation effect. Namely, the reactivity of many antibodies for tissue epitopes can be restored by boiling tissue
sections, a process often referred to as antigen retrieval. This
finding and subsequent refinement of the technique helped
facilitate the dramatic growth in the use of immunohistochemical analysis for surgical pathology. During the ensuing
decade, numerous procedural modifications were described.
These modifications include the composition of the antigenretrieval buffer, temperature (eg, using a pressure cooker or
not), and the use (or not) of microwave irradiation.2 Despite
these methodological improvements, previous studies of the
technique have largely been empirical in nature. Namely,
certain procedural modifications were correlated with better
or worse immunostaining, without a mechanistic understanding of the underlying process. To date, it has not been
possible to delineate the precise molecular and structural
features that are responsible for the formalin-fixation and
antigen-retrieval phenomenon.
Our understanding of the effects of formaldehyde on
proteins traces back to the original work of Fraenkel-Conrat
and colleagues, published during the 1940s.3-5 Formaldehyde
is capable of a variety of cross-linking reactions, recently
summarized by Shi et al. 6 In solution, formaldehyde is
capable of binding to the following amino acids: lysine (K),
arginine (R), tyrosine (Y), asparagine (N), histidine (H), glutamine (Q), and serine (S).6 It is not clear, however, which (if
any) of these reactions might be relevant in the context of
antigen retrieval. For example, it is not even clearly established whether antigen retrieval actually breaks formaldehyde
cross-links.7 Other proposed hypotheses include extraction of
© American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
diffusible blocking proteins, precipitation of proteins, and
rehydration of the tissue section, thereby allowing better
penetration of antibody 8 ; removal of cage-like calcium
complexes 9 ; and heat mobilization of trace remaining
amounts of paraffin.7 Thus, formaldehyde has a variety of
effects on tissue, only some of which are likely to be associated with the antigen-retrieval phenomenon. It was our goal
in the present study to identify the formaldehyde-induced
effects associated with the loss of immunoreactivity and
subsequent recovery associated with antigen retrieval.
To do this, we created a model immunostaining system
using synthetic peptides.10,11 Each peptide mimics the antibody-combining site of a particular antigen, such as
estrogen receptor (ER), progesterone receptor (PR), Ki-67,
p53, or HER-2. We identified the peptides using phage
display, a combinatorial display technique in which an antibody (or other binder) selects from billions of possible
peptide combinations. The selected peptides represent
essentially 1 epitope of a larger protein. The peptides are
relatively short (approximately 20 mer) and conformationally constrained in a cyclic orientation. For this reason, we
initially thought it unlikely that such peptides could undergo
formalin fixation, leading to loss of antibody reactivity. It
was, therefore, with surprise that we discovered that some
of the peptides undergo fixation and antigen retrieval, just as
the native proteins do. Other peptides do not. Since we
know the precise amino acid composition of the peptides,
we were able to precisely correlate the presence of specific
amino acids with the formalin-fixation and antigen-retrieval
phenomenon. We describe the model system for fixation
and antigen retrieval, the correlation of the peptides’ fixation characteristics with their amino acid sequence, and 2
potential applications of these findings.
Materials and Methods
Peptides, Proteins, and Antibody Reagents
ER, PR, and p53 protein–specific monoclonal antibody
clones 1D5, PR 636, and DO-7, respectively, were purchased
from DakoCytomation, Carpinteria, CA. HER-2/neu monoclonal antibody clone 9C2 was obtained from the American
Type Culture Collection and grown in our laboratory or by
Bioexpress Cell Culture Services (West Lebanon, NH).
Peptides that mimic the monoclonal antibody binding site on
ER, PR, p53, and HER-2/neu were previously identified and
characterized.10,11 The peptides were synthesized by SynPep,
Dublin, CA. Recombinant ER and PR were purchased from
Panvera, Madison, WI, and Cascade Biosciences,
Winchester, MA, respectively. Immunohistochemical detection reagents were purchased from DakoCytomation.
Protein or Peptide Coupling
Synthetic cyclic peptides and recombinant proteins were
coupled covalently to the isocyanate-derivatized glass
surface of microscope slides.11,12 For coupling to isocyanatecoated microscope slides, appropriate working dilutions of
peptides and proteins were made in potassium phosphate
(0.5-mol/L concentration, pH 8.9) buffer. Briefly, 1 µL of
various peptide or protein concentrations were spotted onto
activated, isocyanate-derivatized microscope glass slides.
The peptides or proteins were permitted to covalently couple
to the glass for 15 minutes at 42°C to 45°C. The slides then
were rinsed, and the remaining reactive isocyanate groups
were quenched with a protein cocktail containing bovine
gamma globulins (0.025%, Sigma Chemical, St Louis, MO).
Immunohistochemical Staining
Peptides and recombinant proteins that were coupled to
isocyanate-activated slides were subjected to a standard
immunohistochemical staining procedure, as described
earlier.10,11 Briefly, microscope slides first were immersed in
deparaffinization reagents; the slides were serially immersed
in 2 xylene baths, 3 minutes each, and then in decreasing
concentrations of ethyl alcohol, ultimately ending in distilled
water. Antigen retrieval then was performed by incubating
the slides in a pressure cooker (Nordic Ware, Minneapolis,
MN) for 30 minutes in a 0.01-mol/L concentration of citrate
buffer (pH 6.0). Slides then were immunostained with a
labeled streptavidin-biotin detection system, as previously
described.10,11
For ER staining, we used the 1D5 monoclonal
antibody,13 incubated at approximately 3 µg/mL at room
temperature for 40 minutes. Other primary antibodies were
incubated at 37°C for 20 minutes. After the primary antibody, the slides were rinsed and then incubated with biotinconjugated horse antimouse IgG (heavy and light
chain–specific) secondary antibody for 20 minutes at room
temperature. The slides then were rinsed again and incubated
with horseradish peroxidase–conjugated streptavidin for 20
minutes at room temperature. The color then was produced
with liquid-stable diaminobenzidine (3,3'-diaminobenzidine
tetrahydrochloride)/hydrogen peroxide for 10 minutes and
enhanced with 5% (wt/vol) copper enhancer (cupric sulfate
pentahydrate) for 10 minutes.
Peptide Fixation
Peptides (coupled to slides) were fixed by immersing the
slides in 10% neutral buffered formalin for various lengths of
time: 20 minutes or 1, 6, or 16 to 20 hours (overnight). As indicated in the experiments, some peptide-conjugated slides were
not fixed and were used as unfixed controls. Of the formalinfixed slides, 1 set was not subjected to antigen retrieval
(non–antigen retrieval control slides), and the other set was
Am J Clin Pathol 2004;121:190-199
© American Society for Clinical Pathology
191
DOI: 10.1309/BRN7CTX1E84NWWPL
191
191
Sompuram et al / A MOLECULAR MODEL OF FORMALIN FIXATION
Peptide
Group 1
Not
Fixed
Formalin
Fixed
Effect of Formalin Fixation and Antigen Retrieval
p53
HER-2
Group 2
ER
ER + Protein Mix
Group 3
PR
PR + Protein Mix
No Antigen
Retrieval
Antigen
Retrieved
❚Image 1❚ Montage of images, after immunostaining of
peptides that mimic the analytes indicated in the far left
column. The antibodies for these analytes are DO-7 (p53),
9C2 (HER-2), 1D5 (estrogen receptor [ER]), and 636
(progesterone receptor [PR]). The peptides were spotted in
duplicate, adjacent to each other. The spots in the left
column are those that were not fixed, as indicated above the
images. The right column of peptide spots were formalin
fixed and antigen retrieved. For an explanation of the groups,
see the “Effect of Formalin Fixation and Antigen Retrieval”
section, and for product information, see the “Peptides,
Proteins, and Antibody Reagents” section.
subjected to antigen retrieval for various lengths of time (5,
15, or 30 minutes or 1 hour). ER-positive tissue (formalinfixed and paraffin-embedded) section slides were used as
positive control slides through these various treatments.
After antigen retrieval, the slides were subjected to
immunohistochemical staining as described in the
preceding section.
Peptide Spot Intensity Measurement
The colorimetric intensity of the immunohistochemically stained peptide and protein spots was obtained by
scanning the slides with a flatbed scanner (Perfection
1200U, Epson America, Torrance, CA). The image was
stored in Adobe Photoshop, Adobe Systems, San Jose, CA.
The color intensity of the spots then was quantified by using
an image program (Scion, Frederick, MD). Peptide spot
color intensity was expressed as mean pixel optical density,
on a scale of 1 to 256.
192
192
Results
Am J Clin Pathol 2004;121:190-199
DOI: 10.1309/BRN7CTX1E84NWWPL
Previously, we demonstrated that certain small peptides
could bind to monoclonal antibodies specific for ER, PR,
p53, HER-2, and Ki-67.10,11 The peptides mimic the antibody-combining site of the native antigen, as documented
through competition experiments between the peptide and
native antigen.10 Thus, the “ER peptide” simulates the site on
the ER that is recognized by the 1D5 monoclonal antibody.
Similarly, the “PR peptide” mimics the site on the PR that is
recognized by the 636 monoclonal antibody. The same also
applies for the “p53 peptide” (DO-7 antibody), the “HER-2
peptide” (9C2 antibody), and the “Ki-67 peptide” (MIB-1
antibody). We attached these peptides to glass microscope
slides through a protected isocyanate coupling chemistry.11,12
These peptides, once attached to glass slides, are detected by
conventional immunohistochemical procedures.
In the present study, we investigated whether formalin
fixation might alter any of the peptides so as to cause a loss
of immunoreactivity. To do this, peptide-coated slides were
immersed in neutral buffered formalin overnight at room
temperature. A montage of the results is shown in ❚Image 1❚.
Each peptide was placed on the glass slide as a 3-mm-diameter spot, which (like a tissue section) turned brown at the
end of our immunohistochemical staining protocol. Peptides
were spotted onto glass slides in duplicate. The far left
column shows a scanned image of the peptide spots (each in
duplicate) before formalin fixation. Each of the spots (p53,
HER-2, ER, PR) was recognized by its specific antibody,
resulting in a colored spot. These spots (left column) represent a positive control, in that the ability of the peptides to be
recognized by their antibodies was described previously.11
The middle column represents images of the replicate
spots that were fixed in formalin. For the peptides recognized by the p53 (DO-7) and HER-2 (9C2) antibodies,
formalin fixation caused a loss of immunoreactivity. For
purposes of discussion and explanation, we term peptides
with this property as group 1. Among the group 1 peptides,
formalin fixation consistently abrogated the ability of the
peptide to be bound by antibody. As shown in the right
column, antigen retrieval reversed the formalin effect for
group 1 peptides, causing a recovery of immunoreactivity.
These data are represented quantitatively in ❚Figure 1❚ .
Figure 1 also includes an additional control to the far right,
not shown in Image 1. Namely, it shows the same 3 groups
(no fixation, fixation, fixation and antigen retrieval) plus an
additional control of antigen retrieval in the absence of fixation. Although tissue sections would be lost irretrievably by
such a harsh treatment as antigen retrieval without previous
fixation, the peptides are highly stable. As shown in Figure
© American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
Kinetics of Fixation and Antigen Retrieval
We then studied the time course of fixation and antigen
retrieval of the peptides on glass slides. We were interested
to learn whether it would approximate that for tissue fixation. ❚Figure 4❚ shows the time course of formalin fixation
for group 1 (p53) and group 2 (ER with proteins) peptides.
60
Spot Intensity
50
40
30
20
10
0
Fixation: –
+
+
–
Antigen retrieved: –
–
+
+
Slide Treatment
❚Figure 1❚ Group 1 peptides (p53 [black bars] and HER-2
[white bars]) are susceptible to formalin fixation, which is
reversible by antigen retrieval. The error bars represent the
mean ± SE of the color intensity after immunostaining with
p53 DO-7 or HER-2 9C2 monoclonal antibodies. Control
treatments are shown at the far left (no treatment) and far
right (unfixed and antigen retrieved). For product information,
see the “Peptides, Proteins, and Antibody Reagents” section.
80
70
60
Spot Intensity
1, antigen retrieval did not affect the immunoreactivity of the
peptides in the absence of fixation.
Formalin fixation did not cause a loss of immunoreactivity for the group 2 peptide (ER, recognized by the 1D5
antibody). As shown in Image 1, middle column, the ER
peptide was recognized by antibody, resulting in a colored
spot, even after formalin fixation. However, this pattern of
reactivity can be changed. If certain other proteins were
coimmobilized to the glass microscope slide along with the
ER peptide (ER + protein mix), the ER peptide was susceptible to formalin. Formalin fixation of the combination
caused changes in the ER peptide that abrogated the recognition of the 1D5 ER antibody. Antigen retrieval reversed the
effect (far right column, Image 1). For purposes of discussion, we refer to peptides with this property as group 2.
Namely, they are unaffected by formalin fixation unless
certain other proteins are coimmobilized alongside. The data
are expressed quantitatively in ❚Figure 2❚.
The last group (group 3) is composed of peptides that
were not affected by formalin fixation, regardless of the presence of other neighboring proteins. The PR and Ki-67
peptides fall into this category. Representative data for PR
are shown in Image 1, in which the PR 636 antibody recognized the PR peptide under all conditions. These data are
expressed quantitatively in ❚Figure 3❚.
We next extended these initial observations with experiments aimed at elucidating why group 3 peptides were not
susceptible to formalin fixation. We considered whether the
protein backbone itself, away from the antibody-combining
site, also might have a role in determining susceptibility to
formalin fixation. For example, formaldehyde cross-linking
might cause the larger protein to fold in a way that obscures
the antibody-combining site. Neighboring proteins might be
too far away to have an effect on group 3 peptides. We
considered whether the peptide’s sensitivity to formalin might
change if the PR peptide were conjugated directly onto a
larger protein. Therefore, we conjugated the PR peptide onto
2 different proteins—bovine serum albumin and bovine
gamma globulin. After attaching the conjugates to glass
microscope slides, we tested their susceptibility to formalin
fixation. As shown in ❚Image 2❚, the conjugates were not
susceptible to formalin fixation. No significant decrement in
staining intensity was seen after formalin fixation (Image 2,
right column) compared with peptide or peptide conjugate
that was not immersed in formalin (Image 2, left column).
50
40
30
20
10
0
Fixation: –
+
+
–
Antigen retrieved: –
–
+
+
Slide Treatment
❚Figure 2❚ Group 2 peptide (estrogen receptor [ER], black
bars) is susceptible to formalin fixation only if other proteins
are mixed with the ER peptide (white bars). The error bars
represent the mean ± SE of the color intensity after
immunostaining with the ER 1D5 monoclonal antibody.
Control treatments are shown at the far left (no treatment)
and far right (unfixed and antigen retrieved). For product
information, see the “Peptides, Proteins, and Antibody
Reagents” section.
Am J Clin Pathol 2004;121:190-199
© American Society for Clinical Pathology
193
DOI: 10.1309/BRN7CTX1E84NWWPL
193
193
Sompuram et al / A MOLECULAR MODEL OF FORMALIN FIXATION
80
Unfixed
70
Fixed
PR
60
Spot Intensity
PR-BGG
50
PR-BSA
40
30
20
10
0
Fixation: –
+
+
–
Antigen retrieved: –
–
+
+
Slide Treatment
❚Figure 3❚ Group 3 peptide (progesterone receptor [PR], black
bars) is not susceptible to formalin fixation, regardless of
whether other proteins are mixed in (white bars). The error
bars represent the mean ± SE of the color intensity after
immunostaining with the PR 636 monoclonal antibody.
Control treatments are shown at the far left (no treatment)
and far right (unfixed and antigen retrieved). For product
information, see the “Peptides, Proteins, and Antibody
Reagents” section.
Depending on the group, the peptides required between 6
and 16 hours for complete fixation and loss of immunoreactivity. The fixative did not penetrate to any depth, since the
peptides were on a molecular layer on the glass microscope
slide. Consequently, the time was solely a reflection of the
kinetics of a chemical reaction at room temperature. We also
studied whether the presence of adjacent proteins (such as
for group 2 peptides) affected the kinetics of fixation, by
adding proteins adjacent to the p53 peptide. The p53 peptide,
as part of group 1, did not require adjacent proteins to be
susceptible to formalin fixation. The presence of adjacent
proteins did not alter the kinetics of fixation (data not
shown). ❚Figure 5❚ shows the time course for antigen
retrieval of p53 peptide–spotted slides, fixed for 6 or 16
hours. The peptides were retrieved optimally by boiling in a
pressure cooker (approximately 120°C) for 15 to 30 minutes.
These kinetics seem to approximate those for cell lines (data
not shown) and tissue biopsy specimens.14-16
Amino Acid Sequence Analysis
To gain insight into the mechanism of fixation and
antigen retrieval, we compiled the amino acid sequences of
the peptides ❚Figure 6❚ . In addition to the previously
194
194
Am J Clin Pathol 2004;121:190-199
DOI: 10.1309/BRN7CTX1E84NWWPL
❚Image 2❚ Montage of images, after immunostaining of the
progesterone receptor (PR) peptide, PR peptide conjugated to
bovine gamma globulin (PR-BGG), and PR peptide conjugated
to bovine serum albumin (PR-BSA). Fixation (right column)
does not abrogate the ability of the PR 636 antibody to
recognize any of these targets. For product information, see
the “Peptides, Proteins, and Antibody Reagents” section.
described peptides, we also included an additional ER
peptide (ER 1D5 #6) and the Ki-67 peptide, recognized by
the MIB-1 antibody. The ER 1D5 #6 peptide demonstrated
comparable susceptibility to formalin as the previously
described ER peptide (ER 1D5 #3; data not shown). The Ki67 peptide behaved in a manner similar to that of the PR
peptide (data not shown), in that it was not affected by
formalin.
It is important to note that the sequences do not
include any lysines. We purposely engineered a single
lysine in each peptide, appended to the right side of each
sequence, beyond the positions shown. The epsilon amino
group of that lysine was the target for immobilization of the
peptide through the isocyanate coupling chemistry. By
placing a lysine there, the peptide was oriented on the glass
slide so that the antibody-binding site was relatively distant
from the glass surface and freely available to bind to antibody. This was done purposely, to minimize steric interference between the antibody and the glass surface. In addition,
the amino terminus of each peptide was acetylated, eliminating the possibility that the amino terminus (far left on
each peptide) would react with formaldehyde. The carboxyl
terminus of each peptide (far right) was amidated, also
blocking the reactivity of that group. As a result, the only
possible reactions are with the side chains of the amino
acids shown in Figure 6.
Among these 6 peptides, a pattern emerged. Namely,
group 1 peptides have a tyrosine at the antibody-binding site
and an arginine elsewhere. Group 2 peptides have a tyrosine
but no arginine. Group 3 peptides do not have a tyrosine.
These findings suggest that the loss of immunoreactivity
associated with formalin fixation is due to a cross-link at the
antibody-binding site. Moreover, the pattern suggests that the
Mannich reaction is responsible, at least in this model
system, for the susceptibility to formalin fixation.
© American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
50
50
40
Spot Intensity
Spot Intensity
40
30
20
30
20
10
10
0
0
0
0
1
6
❚Figure 4❚ Time course for formalin fixation of estrogen
receptor (ER, squares) and p53 peptides (diamonds). The ER
peptide is coimmobilized to the glass slide with other
proteins, as required for inducing susceptibility to formalin
fixation for group 2 peptides. For product information, see
the “Peptides, Proteins, and Antibody Reagents” section.
Effect of Temperature
Our model of fixation provides a highly quantitative
representation of the formalin fixation process. Therefore, we
used the model to address the role of fixation temperature.
Specifically, we wondered to what extent we could speed up
the formalin reaction by mildly raising the temperature, to
42°C. Our model system is capable of quantifying the difference in kinetics, unfettered by any secondary complication
associated with tissue autolysis. ❚Figure 7❚ shows the kinetics
of formalin fixation for the p53 peptide at room temperature
and at 42°C. The data suggest that raising the formalin fixation temperature to 42°C significantly reduced the required
incubation time. Namely, at 1 hour, the p53 peptide at 42°C
was almost completely fixed. In contrast, the peptide at room
temperature was barely affected by formalin.
We also studied antigen retrieval of the p53 peptide after
varying intervals of fixation. The data shown on the right
side of Figure 7 demonstrate that antigen retrieval is, for the
most part, capable of reversing the formalin-fixation effect.
However, the p53 peptide fixed at 42°C for 20 hours did not
develop as strong a signal as other groups. This finding
suggests that overnight (20-hour) fixation at elevated temperatures might be excessive.
Application to Immunohistochemical Quality Control
A potential implication of these findings is in developing
improved quality control techniques for immunohistochemical
15
30
60
Antigen Retrieval Time (min)
16
Formalin Fixation Time (h)
5
❚Figure 5❚ Time course for antigen retrieval of formalin-fixed
p53 peptide. The kinetics are similar regardless of whether
the peptides were fixed for 6 (squares) or 16 (diamonds)
hours. For product information, see the “Peptides, Proteins,
and Antibody Reagents” section.
Group 1
p53 DO7:
Her 9C2:
S S 3 4 5 6 Y 8 9 10 Y R 13 14 15 16 17 18 19 20
1 S N 4 5 6 R S Y 10 11 12 13 14 15 16 17 18 19 20 21 22
Group 2
ER 1D5 #3:
ER 1D5 #6
1 2 Q 4 5 Y 7 8 9 10 11 N 13 14 15 16 17 18 19
H S H 4 Q 6 7 Y 9 10 11 12 13 14 15 16 17 18 19 20 21
Group 3
PR 636:
Ki-67 MIB-1:
S R 3 4 5 6 7 8 9 10 H 12 13 14 S 16 17 18 19 20 21
S 2 N 4 5 6 7 8 9 10 11 S 13 14 N H N 18 19 20 21 22
❚Figure 6❚ Amino acid sequence of peptides showing only
the amino acid residues that can react with formaldehyde.
The peptides are aligned, starting with the first amino acid on
the amino terminus. Amino acids that do not react with
formalin are not indicated, except for a number denoting the
position in the sequence. Underlined residues are phagescreening consensus sequences, which we believe are
antibody contact (binding) sites. The single letter amino acid
codes are as follows: K, lysine; R, arginine; Y, tyrosine;
N, asparagine; H, histidine; Q, glutamine; and S, serine.
For product information, see the “Peptides, Proteins, and
Antibody Reagents” section.
analysis. Our findings provide a basis for developing inexpensive controls that can distinguish staining variability associated with antigen retrieval (preanalytic variability) from that
associated with alterations in staining reagents or procedure
(analytic variability). An example for illustrative purposes is
shown in ❚Image 3❚. Three types of control material are illustrated: peptides on glass slides (analyte controls), cell line
Am J Clin Pathol 2004;121:190-199
© American Society for Clinical Pathology
195
DOI: 10.1309/BRN7CTX1E84NWWPL
195
195
Sompuram et al / A MOLECULAR MODEL OF FORMALIN FIXATION
50
40
30
20
10
0
Fixation time (h): 1
6
Antigen retrieved: –
–
20
1
6
20
–
+
+
+
Cell Line Controls
Spot Intensity
Antigen Retrieved
Analyte Controls
No Antigen Retrieval
❚Figure 7❚ Formalin fixation proceeds faster at 42°C
compared with room temperature. The p53 peptide–coupled
slides were fixed for 1, 6, and 16 hours (overnight), at room
temperature (black bars) or at 42°C (white bars). The slides
then were immunostained for p53 (DO-7 antibody). The color
intensity of the peptide spot was quantified in duplicate.
Error bars represent the mean ± SE. The 3 groups on the
right side of the figure demonstrate recovery of
immunostaining after antigen retrieval. For product
information, see the “Peptides, Proteins, and Antibody
Reagents” section.
controls (MCF-7, for ER, middle row), and tissue controls
(bottom row). The left column was not antigen retrieved.
The right column was optimally retrieved, as described in
the “Materials and Methods” section. As Image 3 illustrates,
the type of signal seems different, but the information each
conveys is similar; each control developed a comparable
color after antigen retrieval and immunostaining. The
analyte controls have the additional advantage of potentially
distinguishing preanalytic variability (ie, antigen retrieval)
from analytic variability, depending on whether the peptides
are fixed.
Discussion
We present a new in vitro model of formalin fixation.
Studying formalin fixation on native proteins is complicated
by the fact that the majority of epitopes are conformational,
not linear. Antibodies generally bind to amino acids that are
not adjacent to each other in the linear sequence. Rather, the
196
196
Am J Clin Pathol 2004;121:190-199
DOI: 10.1309/BRN7CTX1E84NWWPL
Tissue Controls
Slide Treatment
❚Image 3❚ Comparison of fixed controls for immunohistochemical analysis. The lack of antigen retrieval causes no color
to appear in the peptide spots, cell line controls, and tissue
controls. When the specimens are subjected to antigen
retrieval, each of the control specimens can be immunostained,
resulting in the colored signal that is shown. The inset is a
higher magnification of the cluster of cells (circled).
amino acids comprising the antibody-binding site are
brought into juxtaposition through the normal folding of a
native protein. Therefore, for most proteins, it is not possible
to clearly identify the antibody-binding site. Our model
addresses this problem. Amino acids in the native protein
that might not be adjacent in the linear sequence are brought
together in a small peptide. By identifying consensus
sequences during the phage-screening process, it is possible
to determine the antibody-binding region.10
The use of phage display to identify antibody-binding
sites is not new. What is new, however, is the finding that
such short peptides can be susceptible to formalin fixation.
We found that the peptides segregated into 3 groups based on
their sensitivity to formalin fixation. Group 1 was sensitive
to fixation, in that it was no longer recognized by antibody.
Antigen retrieval reversed the effect. Group 3 peptides were
not susceptible to formalin fixation. Group 2 peptides, both
of them specific for the ER 1D5 monoclonal antibody,
became susceptible to formalin fixation only when neighboring proteins were coimmobilized adjacent to the peptide.
© American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
We initially speculated that the neighboring proteins in
the group 2 scenario might be sterically blocking access to
the peptide. However, once we analyzed the peptide
sequences, it became apparent that the adjacent protein
might be doing something quite distinct: we now believe it
most likely that the adjacent proteins contributed an arginine
residue for the Mannich reaction. Such a contribution was
not necessary for group 1 peptides, since group 1 peptides
have both tyrosine and arginine. The Mannich reaction,
along with other potential formaldehyde reactions with
proteins, is summarized in a recent review by Shi et al.6
Briefly, formaldehyde reacts with an amine (such as on an
arginine), creating a reactive iminium ion intermediate. The
iminium ion then reacts with the phenol group of tyrosine,
creating a covalent bond. We hypothesize that the formalininduced reaction of adjacent peptides (group 1) or peptideproteins (group 2) accounts for the loss of immunoreactivity. According to this hypothesis, the absence of tyrosine
in group 3 peptides accounts for their insensitivity to
formalin fixation.
The preponderance of tyrosines in antibody-binding
sites on our peptides seems to somewhat defy random
chance. It is possible that the reason tyrosine is so prevalent
is that it is an immunogenic residue, capable of stimulating a
strong immune response. According to this line of reasoning,
the prevalence of tyrosines in antibody-binding sites is
selected by the ability of the mouse’s immune system to
recognize that epitope when the monoclonal antibodies were
initially generated.
Our data raise an alternative possibility. Each of the
antibodies was selected for its ability to react with tissue
after formalin fixation and antigen retrieval. That is, in
fact, why the antibody clones are in widespread clinical
use. We wondered whether the presence of a tyrosine in
the antibody-binding site might be a determining feature,
explaining why some antibodies are able to bind to tissue
sections after antigen retrieval. This hypothesis, however,
has one potential flaw. If (according to this hypothesis)
antibodies suitable for antigen retrieval require a tyrosine
at the antibody-binding site, then why do the group 3
peptides not have a tyrosine? Those antibodies (Ki-67
[MIB-1] clone and PR 636 clone) also work well with
antigen retrieval.
To address this question, we reexamined the amino acid
sequence of the 2 group 3 peptides. We found that each (PR
and Ki-67) has aromatic amino acid residues in the antibodybinding site. In one instance, there is a phenylalanine and in
the other, a tryptophan. The phage-display technique we used
to identify the peptides could easily result in a conserved
substitution (tyrosine to a phenylalanine or tryptophan) of this
nature. Thus, it is possible that the native protein has a tyrosine in the antibody-binding site but that it is represented as a
phenylalanine or tryptophan in our selected random peptide.
In aggregate, these findings are highly consistent with the
conclusion that the Mannich reaction is an important chemical reaction in the context of antibodies suitable for antigen
retrieval.
It is possible that our formalin-fixation model did not
accurately represent the chemistry occurring in the native
protein. To further evaluate that possibility, we examined the
kinetics of fixation and antigen retrieval. We found that the
time course was similar to that commonly observed for
tissue sections. Despite the fact that the formalin did not
need to penetrate even 1 µm deep into the glass surface, it
took at least 6 hours to begin observing the abrogation of
antibody binding. This is consistent with the amount of time
required for tissue biopsy specimens. Antigen retrieval times
also were of a time scale comparable with that required for
tissue sections. The comparability of kinetics does not, on its
own, prove the model to be valid, but it does lend further
support to it.
As a model of formalin fixation, it can be useful for
predicting the behavior of tissue biopsy specimens if we
change the conditions of formalin fixation. Therefore, we
applied the model to a problem of everyday importance to
many surgical pathology laboratories. In practice, surgical
specimens often arrive in the surgical pathology accession
area toward the end of the day, making it difficult to
ensure that the biopsy specimens are adequately fixed and
still have the paraffin blocks ready to cut the next
morning. Consequently, there is interest in speeding up
the process. We wondered whether a mild elevation in
temperature, close to body temperature, could meaningfully enhance the formalin-fixation reaction. Our data
suggest that it could. Of course, the model does not
substitute for an actual test with tissue samples. Such
enhancement of formalin fixation on tissue samples by
warming has been described previously.17-19 Our model,
using peptides conjugated to glass slides, facilitates a
highly quantitative comparison and can be instrumental in
understanding the phenomenon.
For the future, the peptides provide a potentially valuable tool for developing standardized quality control test
substrates for immunohistochemical analysis. Current practice requires biologic material for quality control, in the form
of tissue biopsy specimens or cell lines. Peptides, as
described in the present report and elsewhere,10,11 have the
unique characteristic that they can be provided in both
unfixed and fixed forms. The unfixed version will not be
affected by antigen retrieval. Thus, any staining variation
using the unfixed form of a peptide will be due solely to
changes in immunohistochemical reagents or procedure. The
fixed form of the peptide, on the other hand, must be antigen
retrieved adequately to be recognized by antibody (Image 3).
Am J Clin Pathol 2004;121:190-199
© American Society for Clinical Pathology
197
DOI: 10.1309/BRN7CTX1E84NWWPL
197
197
Sompuram et al / A MOLECULAR MODEL OF FORMALIN FIXATION
Therefore, it will be sensitive to variability in the analytic
components of an immunostain (reagents, procedure) and
antigen retrieval. This capability is unique to analyte controls
using peptides.
An additional potential advantage of using analyte
controls as an immunohistochemical quality control is that
the controls can be manufactured in a highly reproducible
form, for mass production. By contrast, standardized manufacture of controls using cell lines is not a trivial endeavor,
because the level of expression for any particular protein
varies from cell to cell (in a population) and over time,
during passage of a cell line. Moreover, the use of biologic
materials such as cell lines requires paraffin sectioning, a
process that is labor intensive and approximately 5 to 10
times more expensive than simply placing a peptide spot on
a protected isocyanate slide. For these reasons, we believe
that this technology ultimately might be useful in fostering
interlaboratory and intralaboratory standardization for quantitative immunohistochemical stains.
These findings suggest that monoclonal antibodies
selected for their clinical usefulness with formalin-fixed
biopsy specimens recognize epitopes containing a tyrosine.
Our data support the possibility that during formalin fixation, a tyrosine is bound covalently to a nearby arginine, as
described for the Mannich reaction.6 These findings do not
rule out other possible reactions for formalin. Notably, our
model system will not detect the effects of formalin on
lysine residues, since we did not select peptides with
lysines. The epsilon amino group of lysine, in the presence
of formaldehyde, can react with numerous other amino
acids, including other lysines, serine, and molecules
containing carbonyl groups. The importance of lysine
residues in formalin fixation is supported further by Hua et
al.20 A distinction between their method and ours is that
theirs did not use antibodies capable of working in the
context of antigen retrieval. Our method did. Consequently,
the types of epitopes that we identified might be biased by
the nature of the antibodies in identifying sites where
formalin cross-linking can be reversed. Our findings on
reversible cross-linking also agree with those of FraenkelConrat and colleagues.3,4 Namely, they found that of all the
protein cross-linking reactions that occur as a consequence
of formalin fixation, the Mannich reaction is different, in
that the cross-links can be hydrolyzed with heat or alkaline
treatment.3,4 Because our molecular dissection focusing on
reversible formalin cross-links also implicates residues that
would participate in a Mannich reaction, the 2 sets of findings agree.
In future work, the availability of defined peptide antibody targets will permit us to further dissect the important
chemical interactions occurring during formalin fixation.
This can be important, since an increasing number of clinical
198
198
Am J Clin Pathol 2004;121:190-199
DOI: 10.1309/BRN7CTX1E84NWWPL
immunohistochemical tests require accurate and reproducible quantitative data. To the extent that formalin fixation
complicates and confounds assay reproducibility and standardization, a better understanding of formalin fixation
might be helpful.
From 1the Department of Pathology and Laboratory Medicine,
Boston University School of Medicine, Boston, MA; and 2Medical
Discovery Partners LLC, Boston.
Supported by grants 1R43CA/AI94557 and 1R44CA81950
(Dr Bogen) from the National Cancer Institute, Bethesda. MD.
Address reprint requests to Dr Sompuram: Medical
Discovery Partners, LLC, 715 Albany St, Boston, MA 02118.
References
1. 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;39:741748.
2. Shi S-R, Gu J, Kalra K, et al. Antigen retrieval technique: a
novel approach to immunohistochemistry on routinely
processed tissue sections. In: Gu J, ed. Analytical Morphology.
Natick, MA: Eaton Publishing; 1997:1-40.
3. Fraenkel-Conrat H, Brandon B, Olcott H. The reaction of
formaldehyde with proteins, IV: participation of indole
groups: gramicidin. J Biol Chem. 1947;168:99-118.
4. Fraenkel-Conrat H, Olcott H. Reaction of formaldehyde with
proteins, VI: crosslinking of amino groups with phenol,
imidazole, or indole groups. J Biol Chem. 1948;174:827-843.
5. Fraenkel-Conrat H, Olcott H. The reaction of formaldehyde
with proteins, V: crosslinking between amino and primary
amide or guanidyl groups. J Am Chem Soc. 1948;70:26732684.
6. Shi S-R, Gu J, Turrens J, et al. Development of the antigen
retrieval technique: philosophical and theoretical bases. In:
Shi S-R, Gu J, Taylor C, eds. Antigen Retrieval Techniques:
Immunohistochemistry & Molecular Morphology. Natick, MA:
Eaton Publishing; 2000:17-40.
7. Gown A, Wever ND, Battifora H. Microwave-based antigenic
unmasking: a revolutionary new technique for routine
immunohistochemistry. Appl Immunohistochem. 1993;1:256266.
8. Suurmeijer A, Boon M. Notes on the application of
microwaves for antigen retrieval in paraffin and plastic tissue
sections. Eur J Morphol. 1993;31:144-150.
9. Morgan J, Navabi H, Schimid K, et al. Possible role of tissuebound calcium ions in citrate-mediated high-temperature
antigen retrieval. J Pathol. 1994;174:301-307.
10. Sompuram SR, Kodela V, Ramanathan H, et al. Synthetic
peptides identified from phage-displayed combinatorial
libraries as immunodiagnostic assay surrogate quality control
targets. Clin Chem. 2002;48:410-420.
11. Sompuram SR, Vani K, Zhang K, et al. A novel quality
control slide for quantitative immunohistochemistry testing.
J Histochem Cytochem. 2002;50:1425-1434.
12. Sompuram S, McMahon D, Vani K, et al. A novel microscope
slide adhesive for poorly adherent tissue sections.
J Histotechnol. 2003;26:117-123.
© American Society for Clinical Pathology
Anatomic Pathology / ORIGINAL ARTICLE
13. Al Saati T, Clamens S, Cohen-Knafo E, et al. Production of
monoclonal antibodies to human estrogen-receptor protein
(ER) using recombinant ER. Int J Cancer. 1993;55:651-654.
14. Rhodes A, Jasani B, Balaton A, et al. Immunohistochemical
demonstration of oestrogen and progesterone receptors:
correlation of standards achieved on in house tumours with
that achieved on external quality assessment material in over
150 laboratories from 26 countries. J Clin Pathol. 2000;53:292301.
15. von Wasielewski R, Mengel M, Wiese B, et al. Tissue array
technology for testing interlaboratory and interobserver
reproducibility of immunohistochemical estrogen receptor
analysis in a large multicenter trial. Am J Clin Pathol.
2002;118:675-682.
16. Lee H, Douglas-Jones A, Morgan J, et al. The effect of
fixation and processing on the sensitivity of oestrogen
receptor assay by immunohistochemistry in breast carcinoma.
J Clin Pathol. 2002;55:236-238.
17. Ni C, Chang T, Searl S, et al. Rapid paraffin fixation for use
in histologic examinations. Ophthalmology. 1981;88:13721376.
18. Durgun-Yucel B, Dere F, Yucel A, et al. Rapid fixation of
whole organ specimens and attendant problems. Acta Med
Okayama. 1992;46:75-81.
19. Helander K. Kinetic studies of formaldehyde binding in tissue.
Biotech Histochem. 1994;69:177-179.
20. Hua C, Langlet C, Buferne M, et al. Selective destruction by
formaldehyde fixation of an H-2Kb serological determinant
involving lysine 89 without loss of T-cell reactivity.
Immunogenetics. 1985;21:227-234.
Am J Clin Pathol 2004;121:190-199
© American Society for Clinical Pathology
199
DOI: 10.1309/BRN7CTX1E84NWWPL
199
199