0022-1554/91/$3.

30
The Journal of Histochemistry and Cytochemistry Vol. 39, No. 2, pp. 225-229, 1991
Copyright © 1991 by The Histochemical Society, Inc. Printed in USA.

Brief Report

Effects of Formaldehyde Fixation on Protein Secondary

Structure: A Calorimetric and Infrared Spectroscopic
Investigation’

JEffREY T MASON and TIMOTHY J. O’LEARY2

Department of Cellular Pathology, Armed Forces Institute ofPathology, Washington, DC 20306-6000.

Received for publication February 21, 1990 and in revised form September 28, 1990; accepted September 29, 1990 (0B1911).

We investigated the effects offormaldehyde fixation on the and hemoglobin. Infrared spectra obtained on the unfixed
secondary structure ofisolated proteins (bovine serum albu- and fixed proteins were essentially identical. This demon-
mm, ribonudease A, and hemoglobin) using high-sensitivity strates that the “fixed” proteins retain the secondary struc-
differential scanning calorimetry and Fourier transform in- flare present before fixation. We therefore condude that the
frared spectroscopy. Whereas thermograms obtained by scan- cross-linking of proteins that occurs in the process of form-
ning calorimetry on unfixed purified proteins demonstrated aidehyde fixation “locks in” the secondary structure of these
denaturation transitions in the 70-9(YC temperature range, protein molecules. (J Histochem Cytochem 39:225-229,
the thermograms showed no denaturation transitions in this 1991)
temperature range when the proteins had been placed in KEY WORDS: Formaldehyde; Fixation; Calorimetry; Infrared; Spectros-
formaldehyde solutions. Thus, fixation destroyed the denatu- copy; Proteins.
ration transition ofbovine serum albumin, ribonudease A,

Introduction methylene diamides. Formaldehyde can also react with alcoholic

hydroxyl groups to form acetals, and with suffhydro groups to form
Formaldehyde is the most widely used fixative in pathology. When
sulfhydral acetal analogues. Formaldehyde can also react with aro-
dissolved in water it is rapidly converted to methylene glycol; vir-
matic rings, forming hydroxymethyl groups which, as mentioned
tually no aldehyde is present in aqueous solutions of “formalde-
above, can be involved in ring formation and cross-linking reac-
hyde” (1). Details of the fixation process remain obscure, however.
tions. Because such a wide variety of reactions are possible, corn-
Although it is widely appreciated that formaldehyde cross-links tis-
plete characterization of the reactions that occur in the process of
sue constituents, particularly proteins, the nature ofthis cross-linking
fixation has not been accomplished, to our knowledge, in even
has not been completely established. Protein structures can be
a single protein.
characterized by at least four levels ofstructural organization. “Pri-
There is a great deal of evidence that the secondary structure
mary structure” refers to amino acid sequence. It seems highly un-
ofa protein is largely, ifnot entirely, determined by its amino acid
likely that this is altered during fixation. “Secondary structure:’
sequence. For this reason, it is possible to make reasonable predic-
“tertiary structure:’ and “quaternary structure” refer to the regular
tions about the secondary structure of a protein, given its amino
arrangements of the polypeptide backbone, the three-dimensional
acid sequence alone (3). Although the effects of chemically alter-
structure of the globular protein, and the structures of aggregates
ing the amino acid substitutents ofa protein on the secondary struc-
of globular proteins, respectively. Fixation could, in principle, re-
ture of that protein have not been extensively investigated, there
suit in structural alterations at any or all of these levels.
is reason to believe that it might have an effect. For example, the
The effects offormaldehyde fixation on the chemical structures
polyamino acids poly-L-lysine and poly-L-glutamic acid may exist
of the amino acid backbone have been well characterized, and are
in a-helix, 3-sheet, or random coil forms, depending on the solu-
summarized by French and Edsall(2). Formaldehyde can react with
tion pH (and hence the ionization state of the amino acid side
primary amines to form Schiff bases, and with amide groups to
chains) (3). It is not unreasonable to suppose that chemical modifi-
form hydroxymethyl compounds. The hydroxymethyl compounds
cation of the side chains by formaldehyde might also exert an ef-
can further condense with another amide moiety to form stable
fect on protein secondary structure.
It has not previously been determined whether the secondary
i Supported in part by a grant from the American Registry of Pa- structure ofproteins is preserved during fixation, although the ques-
thology. tion ofsecondary structure preservation has both academic and prac-
2 To whom correspondence should be addressed. tical significance. Recent advances in infrared spectroscopy make

225
226 MASON, O’LEARY

it possible to determine the secondary structure of abnormal pro- We have used differential scanning calorimetry to characterize
teins in formalin-fixed, paraffin-embedded tissue (4,5); knowing the effects of formaldehyde fixation on the denaturation transi-
whether the proteins in such fixed tissue retain their native 5cc- tions ofhemoglobin, bovine serum albumin, and nibonuclease A.
ondary structure is essential to interpreting the significance of such We have further characterized the effects offixation on protein 5cc-
data. In addition, immunohistochemical assays in formalin-fixed, ondary structure using infrared spectroscopy. This report presents
paraffin-embedded tissue sometimes, but not always, require tis- the results of these studies, which demonstrate that formaldehyde
sue digestion by proteolytic enzymes to obtain a satisfactory result; fixation does not, in fact, alter the secondary structure of purified
it has not been certain whether the efficacy oftissue digestion results globular proteins.
from enhanced penetration of antibodies into the tissue or from
reversal of protein conformational changes induced by fixation.
Although it would be ideal to examine the secondary structure Materials and Methods
effects offormaldehyde fixation on proteins in their native tissues, Desiccated proteins were obtained from Sigma (St Louis, MO) and were
in practice this is not possible. Since all tissues contain a number used without further purification. Formaldehyde was obtained from Baker
of different protein constitutents, spectroscopic measurements on (Phillipsburg, NJ) as a 37% solution. This was diluted to 3.7% with dis-
intact tissue can give data only on the “average” protein present. tilled water and phosphate-buffered to a pH of 7.4. No attempt was made

Different proteins could respond to fixation in different ways, while to exclude small amounts of methanol or formic acid from the solution,
since these contaminants are commonly found in the fixatives used in pathol-
yielding an unchanged “average.” For this reason, it is more practi-
ogy laboratories.
cal to observe the effects offixation on purified proteins, for which
Infrared spectra were obtained on a Bomem DA3.02 Fourier transform
spectroscopically observed structure changes can be interpreted un-
infrared spectrometer in a vacuum-tight internal reflection cell (Micro-Circle;
ambiguously. Spectra-Tech, Stamford, CT). Aqueous protein solutions for infrared spec-
Differential scanning calorimetry is a powerful tool by which troscopy were made up at a concentration of3O% (weight/volume) by dilut-
to examine the effects of formaldehyde fixation on isolated ing in distilled water. Formaldehyde-fixed proteins were prepared by fur-
biomolecules. Calorimetry allows precise measurements of phase then diluting these solutions in equal volumes of3.7% buffered formaldehyde
transitions, including the protein denatunation transition, which solutions, resulting in a final protein concentration of 15% and a final form-
might be affected by the process of fixation. Major disruption of aldehyde concentration of 1.85%. Unfixed protein control solutions were
protein structure, such as denaturation, as a result offixation would prepared by diluting the original 30% protein solution in distilled water,
resulting in a final protein concentration identical to that of the form-
eliminate the phase transition seen calorimetrically. Similarly, stabili-
aldehyde-fixed proteins.
zation ofstructure by the process offixation might also destroy the
The infrared spectrometer was evacuated during spectrum acquisition.
phase transition. In contrast, rapidly reversible stabilization of a
Spectra were acquired using 1000 co-additions at a resolution of 4 cm
protein structure might increase the temperature ofa phase transi-
and apodized using a Blackman-Harris function. This resolution has proven
tion but would not totally eliminate it. sufficient to determine protein secondary structure (reported as percen-
Infrared spectroscopy complements scanning calorimetry in that tages of a-helix, 3-sheet, and disordered structures) by both infrared and
it allows precise determination of secondary structure changes in Raman spectroscopy to within about 10% of the values obtained by X-ray
biological assemblies. The infrared spectra of proteins composed diffraction methods (8). Spectral contributions due to water were subtracted
purely of 13-sheet, a-helix, or disordered structures are readily as- using the 2130-cm’ water association band as a normalizing feature (6,7).

signed to the proper proteins by examination ofcharacteristic amide Data points were recorded at 0.5-cm’ intervals. Peak frequencies are ac-
curate to within ± 0.5 cm; frequencies of the shoulders are reported to
vibrational features present in the 1500-1600 cm (Amide II) and
within approximately 1 cm’.
1600-1700 cm (Amide I) regions, as seen in l#{224}ble1 (6-10). Thus,
Thermograms were obtained in a Hart high-sensitivity differential scan-
denaturation or other protein conformational changes caused
ning calorimeter, at a scan rate of 20’C/hr. Data points were acquired at
by fixation can rapidly be identified by examination of a protein’s
10-sec intervals. Protein solutions for calorimetry were obtained either by
infrared spectrum. dissolving the anhydrous protein in the formaldehyde solution or by dis-
solving it in water and then mixing the resulting solution with formalde-
hyde. Approximately 30 mg ofpnotein dissolved in 400 .tl ofwaten(or 1.85%
Table 1. Amide I and Amide II frequencies asso ciated with formaldehyde solution) were utilized for protein thermograms.
various polypeptide secondaty structures”

Frequency (cm’)

Conformation Amide I Amide II

Results and Discussion
Figures 1, 2, and 3 summarize the calorimetric results for bovine
Unordered 1658 1520
Antiparallel chain serum albumin, hemoglobin, and ribonuclase A, respectively. The
3-sheet 1685 (weak) 1530 (strong) upper scan in each panel represents protein dissolved in water; the
1632 (strong) 15 10 (weak) lower scan represents protein dissolved in formaldehyde. The pro-
1650 (very weak) 1550 (weak) tein denaturation transition has been totally eliminated by form-
Parallel chain aldehyde fixation in each case.
3-sheet 1632 (strong) 1550 (weak) Infrared spectra of the commercial formaldehyde solutions (data
a-helix 1650 (strong) 1516 (weak)
not shown) demonstrated no evidence offormic acid or offree form-
1646 (weak) 1546 (strong)
aldehyde, both of which would have been characterized by strong
a Reproduced by permission from (5). isolated carbonyl C = 0 stretching bands. This confirms previously
FORMALDEHYDE FIXATION 227

Bovine Serum Albumin 17 Ribonuclease A

0
E 0
E
C’
.

C’

a U
C.) a
0
C.)
I 0
0
I
0 0
C
K C
U’ U
CC
U’

55 60 65
50 60 70
Temperature (#{176}C)
Temperature (#{176}C)
Figure 1 . Thermogram of bovine serum albumin obtained by differential scan-
ning calorimetry in the presence (lower trace) and absence (upper trace) of form- Figure a Thermogram of ribonuclease A obtained by differential scanning
aldehyde. Formaldehyde abolishes the thermal phasetransition ofthis protein. calorimetry in the presence (lower trace) and absence (upper trace) of formal-
dehyde. Formaldehyde abolishes the thermal phase transition of this protein.

reported nuclear magnetic resonance spectroscopy data suggesting

that most ofthe formaldehyde present in fixative solutions has been
converted to methylene glycol (1). 0.86
Figures 4, 5, and 6 show the infrared spectra for albumin, he- A
moglobin, and ribonuclease A in the presence of water (upper
panels) and in the presence of formaldehyde (lower panels). The
spectra of the fixed and unfixed proteins are virtually identical.
Since the infrared spectra of proteins are very sensitive to second-
ary structure alterations, the similarity in spectra obtained before
and after formaldehyde fixation indicates that formaldehyde fixa-
tion results in little on no protein secondary structure change. Pro-
tein denatunation, which transforms helical and n-sheet secondary
structure into disordered structures, would be expected to cause
0
a 0.00
C
0 1740
22 Hemoglobin
0
0
.0
0.39
B
0
E
C
U
.

U
C
a
C
C.)

I
C

0
U
K
U’

1490 1615 1865

45 55 65 75
Frequency (cm’)
Temperature (#{176}C)
Figure 4. Amide I and II region infrared spectra of bovine serum albumin in
Figure 2. Thermogram of human hemoglobin obtained by differential scanning the absence (A) and presence (B) of formaldehyde. Although baseline differ-
calorimetry in the presence (lower trace) and absence (upper trace) of formal- ences are seen, the spectra are otherwise nearly identical, indicating that simi-
dehyde. Formaldehyde abolishes the thermal phase transition of this protein. lar secondary structure is present in the two specimens.
228 MASON, OLEARY

0.12 0.33 A
A

0
a
0 C
a 0 0.00
C .0
0
0
0
0 .0
C
.0
0.30 B
B

0.00
1 1486 1615 1740 1865 1360 1485 1610

Frequency (cm1) Frequency (cm’)

Figure 5. Amide I and II region infrared spectra of human hemoglobin in the Figure 6. Amide I and II region infrared spectra of ribonuclease A in the ab-
absence(A) and presence (B)offormaldehyde. Although baseline differences sence (A) and presence (B) of formaldehyde. Although baseline differences
are seen, the spectra are otherwise nearly identical, indicating that similar sec- are seen, the spectra are otherwise nearly identical, indicating that similar sec-
ondary structure is present in the two specimens. ondary structure is present in the two specimens.

significant changes in the infrared spectra of these proteins. There- digestion in “unmasking” antigens which have been “masked” by
fore, the abolition ofthe denaturation transition indicates that the formaldehyde fixation results from a reversal of formaldehyde’s ef-
original secondary structure has been effectively “locked in” by the fect on protein secondary structure. Rather, this unmasking proba-
process of fixation, at least to the highest temperatures recorded bly results from increased penetration of antibodies to the anti-
in the thermograms. gens, which results solely from removal of large aggregates of
Although the observation that formaldehyde fixation causes no cross-linked protein.
secondary structure changes in hemoglobin may at first seem to In summary, thermodynamic and spectroscopic results indicate
conflict with the fact that hemoglobin’s oxygen affinity is increased that formaldehyde cross-links globular proteins, and that the pro-
by reaction with formaldehyde (1 1), there is really no inconsistency. cess ofcross-linking does not result in discernible alteration of pro-
Changes in oxygen affinity of hemoglobin resulting from oxygen tein secondary structure.
binding and protonation result from changes in quaternary
structure - that is, the interactions between the four hemoglobin Disclaimer
subunits (12,13). These changes in quaternary structure occur with-
The opinions expressedin this paper are the personal views ofthe authors
out discernible change in secondary structure. It is therefore not
and are not to be construed as official or as representing the views of the
surprising to find that formaldehyde fixation similarly affects the Department ofthe Army or the Department of Defense.
function of binding sites in other proteins consisting of multiple
subunits, also without alteration of secondary structure.
It is highly unlikely that any process subsequent to fixation in
the processing, embedding, and staining oftissue (other than pro- Literature Cited
teolytic digestion) can alter the secondary structure of tissue speci- 1. Fox CH, Johnson FB, Whiting), Roller PP. Formaldehyde fixation. J
mens. Similarly, it seems unlikely that the efficacy of proteolytic Histochem Cytochem 1985;33:845
FORMALDEHYDE FIXATION 229

2. French D, EdsallJT The reactions of formaldehyde with amino acids 8. VincentJS, Steer C), Levin 1W. Infrared spectroscopic study ofthe pH-
and proteins. Adv Protein Chem 1945;2:278 dependent secondary structure of brain clathrin. Biochemistry 1984;

3. Fasman GD. The development ofthe prediction ofprotein structure. 18:291

In Fasman GD, ed. Prediction ofprotein structure and the principles 9. O’Leary TJ. Ross PD, Lieber MR, Levin 1W. Effect ofcyclosporin A on
of protein conformation, New York: Plenum Press, 1990 biomembranes; vibrational spectroscopic, calorimetric and hemolysis
4. O’Leary ‘I), Levin 1W. Secondary structure of endocrine amyloid: in- stUdies. Biophrs J 1986;49:795
frared spectroscopy of medullary carcinoma of the thyroid. Lab Invest 10. Susi H, Byler DM. Fourier deconvolution of the Amide I Raman band
1985;53:240 of proteins as related to conformation. AppI Spectrosc 1988;42:819.

5. O’Leary TJ, Engler WF, Ventre KM. Infrared microspectroscopy ofhu- 11. Guthe KR The effect of formaldehyde on the oxygen equilibrium of
man tissue. Appl Spectrosc 1989;43:1095 hemoglobin. J Gen Physiol 1954:37:775

6. Susi H. Infrared spectra of biological macromolecules and related sys- 12. Smith FR, Ackers GK. Experimental resolution ofcooperative free ener-
tems. In Timashjef SM, Fasman G, eds. Structure and stability of bio- gies for the ten ligation states of human hemoglobin. Proc NatI Acad
logical macromolecules. Part II. New York: Marcel Dekker, 1969:576 Sci USA 1985;82:5347

7 . Thomas GH, Kyogoku Y. Biological science. In Brame EG, Grasselli 13. Johnson ML, Turner BW, Ackers GK. A quantitative model for the
JG, eds. Infrared and Raman spectroscopy. Part C. New York: Marcel cooperative mechanism ofhuman hemoglobin. Proc Narl Acad Sci USA
Dekker, 1977:717 1984;81:1093