- Email: [email protected]
Stability of monoclonal antibody-defined epitopes
Journal of Immunological Methods, 139 (1991) 55-64
55
© 1991 Elsevier Science Publishers B.V. 0022-1759/91/$03.50 ADONIS 002217599100160A
JIM 05912
Stability of monoclonal antibody-defined epitopes * Byron E. Wilson, Susan Sun, M e h m e t O z t u r k a n d Jack R. W a n d s Molecular Hepatology Laboratory, Massachusetts General Hospital Cancer Center, Charlestown, MA 02129, U.S.A., and Department of Medicine, Harvard Medical School, Boston, MA 02114, U.S.A. (Received 28 December 1990, accepted 15 January 1991)
Epitope instability can limit the applications of monoclonal antibody (mAb) technology in laboratory and clinical research. We exposed a group of representative antigens on human hepatocellular carcinoma (HCC) cells to physiochemical insults to study epitope stability as measured by mAb immunoreactivity. Each epitope was found to have a unique pattern of instability which serves to biophysically characterize the antigen and de'fines the conditions to which the antigen can be exposed during laboratory and clinical investigations. Individual antigens were found to be unstable within a surprisingly well defined window of solvent polarities while being stable on either side of that window. Several antigens were observed to be unstable when exposed to transient changes in pH. When a critical temperature between 42 ° C and 65 °C was achieved, epitopes which were thermosensitive underwent a sudden loss in immunoreactivity. This critical temperature was found to be pH dependent. The effects of polarity, pH, and temperature on epitope stability are consistent with changes in protein structural conformation. In addition, we found that certain fixatives cause a time and concentration dependent loss of epitope immunoreactivity. This study provides a rapid and easy determination of monoclonal antibody-defined epitope stability; the results of which serve to guide further studies on the antigen and to characterize the antigen on the basis of its unique physiochemical stability. Key words." Cell surface antigen; Epitope; Monoclonal antibody
Introduction
The development of monoclonal antibody (mAb) technology has led to a new generation of
Correspondence to: J.R. Wands, Molecular Hepatology Laboratory, M G H Cancer Center, 7th Floor, M G H East, 149 13th Street, Charlestown, MA 02129, U.S.A. * This work was supported in part by Grants CA-3571, AA02666, AA-01862 and HD-20469 from the National Institutes of Health and a grant from Association pour la Recherche sur le Cancer, Villejuif, France. J.R.W. is the recipient of a Research Scientist Award AA-00048.
research and clinical reagents. Morloclonal antibodies are now instrumental in antigen identification, characterization, and purification; in cloning of translated genetic products; and in immunohistology for investigative and diagnostic purposes (Gatter et al., 1984; Tainer et al., 1984; DeLellis et al., 1987; Takahashi et al., 1988, 1989a, b; Ozturk et al., 1989). In addition, monoclonal antibodies are becoming increasing useful as potential in vivo immunotargeting and therapeutic reagents. (Vitetta et al., 1987; Takahashi et al., 1988, 1989a, b). cad The advantage of monoclonal over polyclonal antibodies is derived from the exquisite specificity
56 afforded by single epitope recognition. However, the potential applications of a given mAb can be limited if the epitope recognized by that antibody is unstable. For example, immunohistological techniques often require extensive processing of tissue specimens for preservation of morphology; Western immunoblotting requires antigen solubilization and often denaturation; protein purification may involve acidic and alkaline conditions, use of polar and nonpolar solvents, or thermal precipitation. Under such non-physiological conditions, a given epitope's immunoreactivity may be destroyed or substantially altered. Polyclonal antibodies often recognize a spectrum of many different epitopes such that physiochemical modifications destroy only a subset of epitopes, resulting in a lowered, but generally not completely lost, total immunoreactivity (Vita et al., 1985; DeLellis et al., 1987). In contrast, as a single epitope system, the immunoreactivity of an antigen to a monoclonal antibody may be completely lost following a physiochemical insult (Hancock et al., 1982; Garter et al., 1984; Oda et al., 1984; Ciocca et al., 1986; DeLellis et al., 1987; Pollard et al., 1987). Determining the stability of a given epitope provides a partial characterization of the antigen. For example, changes in immunoreactivity following physiochemical insult have been used to differentiate between (and define the structural basis for) rheumatoid factor crossidiotypes (Agnello et al., 1986); to detect subtle differences in type IV collagen between tissues (Linsenmayer et al., 1984); to study the conformational substructure of thermolysin (Vita et al., 1985); and to study the relationship between structure and function for IgE domains containing the mast cell receptor binding site (Dorrington et al., 1973; Hastings et al., 1988). On a more practical level, determining the stability of a given epitope delineates the conditions necessary for use of the monoclonal antibody as an investigative or clinical tool (Hancock et al., 1982; Oda et al., 1984; Pollard et al., 1987). We have developed methods by which one can readily determine the stability of epitopes to various physiochemical conditions often encountered in laboratory and clinical research, and have applied these methods to a panel of well defined tumor-associated antigens.
Materials and methods
Cell culture The hepatocellular carcinoma cell line FOCUS developed by us (Lun et al., 1984) was maintained in Eagle's minimum essential medium (M.A. Bioproducts, Walkersville, MD) supplemented with 10 /~M of nonessential amino acids and 10% fetal bovine serum inactivated at 56°C. Immunizing cells were grown from the second passage of the FOCUS cell line and were harvested from monolayer culture with EDTA/Versene in the absence of trypsin, then washed with PBS. Monoclonal antibody production 6-week-old B A L B / c female mice were first immunized by intraperitoneal injection of 4.0 x 10 6 FOCUS cells emulsified in 50% (v/v) complete Freund's adjuvant. Secondary immunizations were performed 6-10 weeks later by intravenous injection of 4.0 × 106 intact cells in 0.2 ml of PBS. 3 days later, mouse splenocytes were fused with either SP20, NS1, or X63 mouse myeloma cell lines with 30% (w/v) polyethylene glycol. Resulting hybridomas were maintained and selected as described (Laemmli et al., 1970). Screening for mAbs was performed by the indirect binding assay as described (Wilson et al., 1988). Purification and iodination of mAbs were performed as described (Takahashi et al., 1989; Ey et al., 1990). Properties of human tumor associated antigens We have selected six anti-FOCUS monoclonal antibodies, named AF5, AF20, FB50, SF25, SF90 and XF8, for use in this study. All these mAbs recognize antigens which share the common feature of having a high level of expression in hepatocellular carcinoma (HCC) with little or no expression in adjacent uninvolved liver (Wilson et al., 1988; unpublished results). Three antibodies (AF5, FB50, and SF90) recognize proteins identifiable by Western immunoblotting. Furthermore these mAbs are known to recognize epitopes contained within the primary structure of the protein. The antigen recognized by AF5 is a 36 kDa phosphoprotein, which in addition to HCC tissue, has been found on all transformed cell lines tested (Frohlich et al., 1990). The
57 mAb FB50 recognizes two families of 50 kDa and 100 kDa proteins (data not shown). SF90 recognizes a 50 kDa protein distinct from that recognized by FB50 (Ozturk et al., 1989). In addition to high HCC expression, the FB50 and SF90 antigens are highly expressed in normal adrenal glands but are expressed at considerably lower levels in other normal human tissues. The SF90 antigen has been present on all malignant human cells thus far tested. AF20, SF25, and XF8 recognize antigens highly expressed in adenocarcinomas of the colon, lung and liver and most other malignant tumors, with little if any expression on normal human tissues (Takahashi et al., 1988, 1989a). None of these three antigens are demonstrable by Western immunoblotting. All three antigens have been successful in imaging tumors in nude mouse models (Takahashi et al., 1988, 1989a, b). Although none of these antigens are demonstrable by Western immunoblotting, all have been fully characterized by metabolic labeling and immunoprecipitation (unpublished results). All three antibodies recognize epitopes found on the protein cores of cell surface glycoptoteins. XF8 and AF20 recognize dissimilar epitopes on a 180 kDA homodimer which is also detectable in normal adrenal glands. SF25 recognizes a 130 kDa heterodimer also expressed by renal proximal tubule cells.
Cell homogenates Harvested FOCUS cells were centrifuged at 1000 × g for 10 min. The pellet was homogenized for 20 s on ice with a Polytron (Brinkman) in 50 vols. of PBS (pH 7.2) containing 0.1% NaN 3, to yield a final protein concentration of approximately 0.5 mg/ml. Filter binding assay Assays were performed in 96-well plates containing glass filters (V & P Scientific, San Diego, CA) as previously described (Wilson et al., 1988). All incubations were performed at room temperature unless otherwise specified. Briefly, FOCUS cell homogenate (50 or 100 lug of protein) was applied to each filter with gentle aspiration to trap proteins. Filters were then incubated with 250 lul of 100% calf serum (CS) for 15 min to block nonspecific protein binding sites. After filtering
off the calf serum, wells were preincubated for 15 min with 50% C S / P B S or with 100 lul of a 1/100 dilution of unlabeled blocking mAb in 50% CS/PBS. To each well, 105 cpm of 125I-labeled mAb diluted in 50% C S / P B S was added and incubated for 2 h. Filters were washed five times (1.5 ml/well) with 5% C S / P B S and counted for gamma radioactivity. The data obtained from several different experiments were normalized with respect to specific binding. Specific binding was determined by subtracting the binding of radiolabeled mAb in the presence of unlabeled homologous mAb (nonspecific binding) from the unblocked binding of radiolabeled mAb (total binding).
Antigen stability assays The effect on epitope immunoreactivity of exposing the antigens to physiochemical modifications was assayed using the above filter binding assay with the following experimental manipulations. The effect of formalin, glutaraldehyde and alcohols on epitope stability was studied by incubating the filter-bound FOCUS homogenate in the various reagents for one hour. The filters were then aspirated dry and washed five times with PBS pH 7.2 (1.5 ml/well) prior to addition of blocking calf serum. To study the effect of ethanol, acetone, and xylene, filters containing bound homogenate were gently removed from 96-well plates and incubated in the organic solvent for 1 h. The solvent was evaporated off, and the filters were replaced in the wells. The filter binding assay was then performed as above. Thus the effects of ethanol were studied by two methods, which gave nearly identical results, indicating that the observed immunoreactive modifying effects of ethanol were not due to washing away of antigen bound to the filter plates. The effect of pH on antigen stability was determined by incubating filters containing bound FOCUS homogenate with various pH solutions of PBS (200 lul). The pH solutions were then aspirated and filters were washed five times (1.5 ml/well) with 5% C S / P B S pH 7.2 prior to addition of blocking calf serum. An alternative method, which gave identical results, was to preincubate the FOCUS homogenate with various pH solutions for 30 min followed by neutralization of the solu-
58
tions to pH 7 prior to binding the homogenate to the filter. Thermoinactivation studies were conducted by heating FOCUS homogenates in glass tubes at the desired temperature for 1 h, or at 1 0 0 ° C for 20 min. After heating, the homogenates were allowed to gradually cool to room temperature to allow for any reversible processes to occur. Any precipitates were gently resuspended. The cell homogenates were then applied to filters and assayed as described above.
Kinetic studies T h e kinetics of t h e r m a l d e n a t u r a t i o n were studied by a d d i n g F O C U S h o m o g e n a t e to glass tubes p r e h e a t e d in a 75 ° C r o t a t i n g water bath. A t t i m ed intervals, the h o m o g e n a t e s were r e m o v e d f r o m the w at er b a t h an d further t h e r m o i n a c t i v a tion was q u e n c h e d with 1 ml of ice-cold PBS buffer. H o m o g e n a t e s were then l o a d e d o n t o filters for assay. T o study the effect of f o r m a l i n fixation with time, cell h o m o g e n a t e s were first b o u n d to filters
XF8
5°°°IAF5
t00C 80C 60C
L
20O0
40C
20C
4000 0c
C
SF25
AF20
IOOC
3000
80C 2000
Jr~ ii~ii
60C 40C
~000 20C C
SFgO
FB50
3000
5000 4000
2000
3000 2000
4000
4000
-~
o
z o
.~ ~
N
g~
o o
*~ ° o
=
~
~
~
z
o o
e
~
N
~
~
~ e 8
~
~
o
~
-
8_
Fig. l. The effect on epitope immunoreactivity of transient exposure to solvents and fixatives. For four epitopes (recognized by AF5, AF20, XF8 and SF25) there is a range of intermediate solvent polarities destructive to immunoreactivity. The positive control was normal saline.
59
then incubated with 200 ffl of 10% formalin in PBS for various time intervals. Filters were washed 5 times with PBS and assayed as described above.
~
-8
F20
• Melhonol i c Elhonor 1 IsoproponoI
Lb
t
20
-
15
Results
Six anti-FOCUS monoclonal antibodies (named AF5, AF20, FB50, SF25, SF90, and XF8) were selected for this study because the antigens they recognize have been previously well characterized with respect to tissue distribution, biochemical properties and some of their molecular characteristics (Takahashi et al.,, 1988, 1989a,b; Wilson et al., 1988; Ozturk et al., 1989; unpublished results). All are known to be at least partially protein in nature. Fig. 1 is a summary of three experiments on the effects of transient exposure to solvents and tissue fixatives on the immunoreactivity of these antigens. The solvents were selected to represent a broad spectrum of polarity. All epitopes were found to be stable in saline solutions. Four antigens (recognized by mAbs AF5, AF20, SF25, and XF8) were highly sensitive to treatment with 80% ethanol, a solvent of moderate polarity. Exposure to 100% ethanol, which is less polar, destroyed the XF8 and SF25 epitopes while only mildly affecting the immunoreactivities of the AF5 and AF20 epitopes. Treatment with acetone, a weakly polar solvent, reduced the immunoreactivities of only the XF8 and SF25 epitopes. Xylene, a non-polar solvent, did not have an effect on the immunoreactivities of any of the epitopes studied. Thus it appears that for four of the six antigens there is a range of solvent polarities that reduces immunoreactivity, whereas the other two antigens (recognized by FB50 and SF90) appear stable following exposure to all solvent polarities tested. To more closely examine the effect of polarity on epitope reactivity, ethanol solutions of varying concentrations were studied. As shown in Fig. 2, the immunoreactivity of the XF8 epitope drops quickly as ethanol concentration is raised (or as solvent polarity is lowered) and remains depressed at 100% ethanol. The AF5 and AF20 epitopes are also unstable in ethanol solutions. However in contrast to XFS, these two epitopes are more stable at the extremes of polarity. They are not
ic, I£
40
l ~
:5
40
80
0(
80
40
80
PERCENT ALCOHOL
Fig. 2. The effect on epitope immunoreactivity of transient exposure to alcohols diluted with water to achieve gradations of solvent polarities. The alcohols in order of decreasing polarity are methanol, ethanol, and isopropanol. The immunoreactivity curves for the antigens recognized by AF5 and AF20 shift according to alcohol polarity, indicating that a set range of solvent polarities are destructive to epitope immunoreactivity.
inactivated until solvent polarities lower than those which destroy the XF8 epitope are reached and they are only partially destroyed at the even lower polarity of 100% ethanol, a polarity at which XF8 is still unstable. The hypothesis that there is a window of intermediate solvent polarity within which given antigens are unstable but outside of which they are stable is further supported with the finding that an alcohol of lower polarity than ethanol (isopropanol) gave a curve shifted to the left of that for ethanol whereas an alcohol of
~l
5oo
800f XF8
E
40C
600
500
0
L 4
2
4OO
200
± 5
0
I / t I I I 60 t20 ~80
MINUTES
Fig. 3. The kinetics of epitope inactivation on heating to 75 ° C (A) and fixing with 3.7% formaldehyde (B). Thermodestruction of the epitope recognized by XF8 is very rapid and exhibits first order kinetics. The kinetics of inactivation by formaldehyde crosslinking is of high order and is concentration dependent (data not shown).
60
higher polarity (methanol) gave a curve shifted to the right of that for ethanol. It appears from the above studies, that the six epitopes can be divided into three groups based upon their stability in solvents of varying polarity: the FB50 and SF90 defined epitopes show stability in all polarities tested; the AF20 and AF5 epitopes show instability in a narrow range of polarities corresponding to solutions of 40-90% ethanol; and the XF8 and SF25 epitopes show a broad range of instability starting at high polarity (30% ethanol) and continuing past the low polarity of 100% acetone. Fig. 1 shows the effects of cross-linking fixatives on immunoreactivity. The epitopes recognized by five of the six antibodies studied (AF5, AF20, XF8, SF25, and FB50) were sensitive to 2% glutaraldehyde such that virtually all of the binding activity was destroyed. In contrast, formaldehyde substantially altered the immunoreactivity of only one epitope (XF8), and mildly lowered the reactivity of two others (AF20 and SF25). Thus we found glutaraldehyde to be the more immunodestructive of the two crosslinking agents; affecting even the remarkably stable epitope recognized by FB50. The fixative-induced decrease in immunoreactivity varied with concentration (data not shown) and duration of fixation. Fig. 3B demonstrates the time course of inactivation for the epitope recognized by XF8 using 3.7% formaldehyde. The AF20 and SF25 defined epitopes demonstrated similar kinetics, but with slightly longer half lives (data not shown). Fig. 4 illustrates results of four experiments on the effects of temporarily heating the antigens. The epitopes recognized by AF5 and SF25 are abruptly thermoinactivated between 4 2 ° C and 55°C whereas the XF8 and AF20 epitopes are thermoinactivated between 55 ° C and 65°C. The FB50 and SF90 epitopes were quite stable, even if boiled for 20 min. The destruction of thermal-sensitive epitopes occurs very rapidly. Fig. 3A illustrates two experiments on the kinetics of thermoinactivation for the XF8 epitope. Similar results were obtained for the antigens recognized by AF5 and AF20 (data not shown). The thermoinactivation of our epitopes occurred with half-lives of less than 30 s at 7 5 ° C (Fig. 3A); findings consistent with non-co-
i ,~F5
XF8
....
oo
4 0 0 0 I'
t5i~
2Z!I'
L
,oooi 50
19fI
.
-
: .r2o
! s~25 ~oooi ]
3500, -
25OO
800
_~
2O00 }
600
~:~
45001 1000
400 : 2O
53O 0.
.
.
.
.
~
F850
i
SF90
400O~
l <8o
'009 50C
!
:: 2::, 4( 6c a(, 4,-~, :, ~c 41o ~'o ~'; ~,:,_, TEMPERATURE
(°C)
]Pig. 4. Thermoinactivation of epitopes followed by slow cooling to allow for renaturation. Thermosensitive epitopes (recognized by AF5, AF20, XF8 and SF-25) underwent a sudden loss in immunoreactivity when a characteristic critical temperature was achieved.
valent thermoinactivation (Ahern et al., 1985, 1988). Fig. 5 shows four experiments depicting the effect on immunoreactivity of transiently changing [ xJF5 4~CL
XF8
ic~O
....} ~F20
6r25 eco
60O:) 4">0O
rex: i
?OCO
,
SF90
FB50
4000'
i q
z
4
6
8
40 12 ~4
9
4
6
~
,8
~? ,4
pH
Fig. 5. The effect on epitope immunoreactivityof transiently changing the pH. Susceptible epitopes were destroyed at the extremes of pH, but the decline in immunoreactivity as a function of pH differed betweenepitopes.
61
•
AF20 1oooo
8000
2000 O"
0
t0
20
30 40 50 60 70
TEMPERA TURE ( °C ) Fig. 6. The relation between p H and thermoinactivation of epitopes. Acidic conditions lower the temperature necessary for epitope inactivation.
the pH. The XF8 epitope was completely inactivated with a small change in pH above 10.0 or below 5.5. The epitopes recognized by AF5, AF20, and SF25 showed incremental destruction as the pH varied from an optimum of around 6. The FB50 and SF90 epitopes remained stable at all pHs. A relationship was observed between the temperature stability and pH stability of the epitopes studied. As shown in Fig. 6, we observed that as the pH is lowered, there is a decrease in the temperature necessary for reduction of immunoreactivity.
Discussion
The antigens studied here are of interest to us because the monoclonal antibodies directed against them are promising tools for studying the cellular biology of malignant transformation, and are potential reagents for immunodiagnosis and therapy (Takahashi et al., 1988, 1989a, b; Wilson et al., 1988; Ozturk et al., 1989; unpublished results). For this study, unpurified cellular antigens were used so as to best represent the conditions most often encountered when employing monoclonal antibodies for laboratory and clinical research purposes. Based on a simple filter binding assay, we have been able to rapidly produce individual profiles of epitope stability for each of the antigens studied.
The results indicate that epitope stability, as determined by immunoreactivity, can be affected by changes in solvent polarity, pH, and temperature and by presumed covalent alteration. This reflects the nature of the antigen-antibody interaction, as complete binding of an antibody to an epitope can only occur if the epitope region can assume a well defined three-dimensional configuration (Tainer et al., 1984; Westhof et al., 1984; Vita et al., 1985; Hastings et al., 1988). As confirmed by the present investigation, when the ability of the epitope to assume that configuration is hampered by denaturation, crosslinking, or degradation then immunoreactivity will be lost (Dorrington et al., 1973; Hancock et al., 1982; Linsenmayer et al., 1984; Oda et al., 1984; Vita et al., 1985; Agnello et al., 1986; Pollard et al., 1987). For example, degradation by at least one or more protease abolished the immunoreactivity of every epitope presented here (data not shown). Furthermore, when treated with crosslinking agents such as glutaraldehyde or formaldehyde, the immunoreactivities of several of the epitopes were destroyed (Fig. 1). Others have reported similar results (Hancock et al., 1982; Ciocca et al., 1986; DeLellis et al., 1987; Pollard et al., 1987). Perhaps the crosslinking bonds chemically or configurationally alter the epitope or they shield the epitope from antibody binding; further studies are required. Nonetheless, given the kinetics of epitope inactivation, it is apparent that controlling duration and concentration of fixation is critically i m p o r t a n t when performing routine immunohistology. It has been proposed that the stability of a given protein (or epitope) conformation is determined by the free energy of the system composed of the solvent, the protein in the given conformation, and other interacting molecules (lipids, ions, solutes, and other proteins) (Tanford, 1968; Baldwin et al., 1987; Creighton et al., 1987; DeLellis et al., 1987). A protein in the native folded state achieves a low free energy by forming the most optimal bonds available and by maximizing the freedom of solvent movement. However this lowering of free energy is offset by a resultant loss in freedom of protein movement compared to that found in the randomly unfolded state (Westhof et al., 1984; Baldwin, 1986). A transition to or
62 from the protein's native immunoreactive state will occur if it results in a lowering of the overall free energy. However, even if the folded state is energetically more favorable than the unfolded state, refolding may not occur: kinetic barriers, energetically unfavorable intermediate states, and antigen aggregation may prevent refolding of the denatured protein (Teipel et al., 1971; Dorrington et al., 1973; Baldwin et al., 1987; Klibanov et al., 1987; Ahern et al., 1988). As temperature rises the kinetic energy and thus the random motion of molecules also increase. Consequently states with more randomness become favorable and the protein usually unfolds (Baldwin, 1986; Kilbanov et al., 1987). As the protein is cooled back to room temperature it will assume the state with lowest free energy unless inhibited from doing so (Teipel et al., 1971; Baldwin et al., 1987; Ahern et al., 1988). Indeed, two epitopes (SF90 and FB50) maintained their immunoreactivity even after boiling, suggesting that for these epitopes the native state is energetically favorable (Fig. 4). In contrast, the other four antigens studied here presumably unfolded at sharply defined temperatures within a range of 42 ° C-65 ° C and their epitopes remained non-reactive upon cooling. Others have reported similar results (Dorrington et al., 1973; Hancock et al., 1982; Linsenmayer et al., 1984; Oda et al., 1984; Agnello et al., 1986; Pollard et al., 1987). For each of our antigens we found a highly reproducible and unique response to transient changes in pH (Fig. 5). Four epitopes (recognized by AF5, AF20, SF25, and XF8) lose their immunoreactivities when exposed to acidic or alkaline conditions. In contrast, the FB50 and SF90 epitopes appear stable throughout the pH range tested. The temperature necessary for AF20 epitope thermoinactivation decreased as the pH of the solution was decreased (Fig. 6). It appears that the native folded state becomes more susceptible to thermodenaturation when destabilized by changes in the p H of the solvent (Tanford, 1968; Fujita et al., 1979, 1982; Cupane et al., 1982; Zaks et al., 1984, 1988; Baldwin et al., 1987; Klibanov et al., 1987; Ahem et al., 1988). Figs. 1 and 2 show that epitopes recognized by AF20 and AF5 are unstable in a range of polarities corresponding to solutions of 40-90% ethanol,
and the XF8 and SF25 epitopes are unstable in a wider range of polarity: from 30% ethanol down to 100% acetone. Outside of these polarity ranges the antigens are stable. The FB50 and SF90 epitopes are able to resume immunoreactive configurations on restoration to a normal saline solvent. The non-polar moieties of proteins in polar solvents are internalized as a result of solvent shell formation, while in low polarity solvents the bonds between polar moieties become more favorable (Fujita et al., 1979, 1982; Cupane et al., 1982; Zaks et al., 1984, 1988; Baldwin, 1986, 1987; Creighton et al., 1987; Ahem et al., 1988). These solvent effects stabilize the folded protein. Consequentially, one would expect the native states of proteins to be stable in high polarity aqueous solvents as well as in low polarity solvents, but to be unstable in intermediate polarity solvents where neither of these two solvent effects are sufficient to stabilize the native state. Our experimental results fit this model well. On a practical level our results indicate that for certain antigens there is a window of solvent polarity which is destructive to immunoreactivity and must be avoided. Unfortunately many experimental techniques use solvents of these intermediate polarities. For example, graded solutions of alcohol are often used to dehydrate tissues or to destroy endogenous peroxidases for immunohistology (Ciocca et al., 1986; Pollard et al., 1987), and the transfer buffer for Western immunoblotting often contains an alcohol (Laemmli, 1970). Our results suggest other buffers and other immunohistological methods must be employed when studying susceptible antigens. We have demonstrated that epitope stability is likely to be analogous to protein stability (Tanford, 1968). Alterations in protein stability explain the effects of temperature, pH, and solvent polarity on epitope stability, showing that each environmental manipulation effects epitope stability. We found that the native folded state of the epitopes recognized by FB50 and SF90 are the most stable. Consequently, these epitopes are not irreversibly destroyed by changes in pH, temperature, and solvent polarity. These proteins should therefore be amenable to the wide ranges of environmental manipulation involved in characterization, purification, cloning, and cytology.
63
In contrast, the other four antigens (recognized by mAbs AF5, AF20, SF25, and XF8) were found to be unstable to many physiochemical insults, and were not able to resume their native folded states with removal of these insults. Therefore, many standard protocols have to be modified to obtain reliable results with these antibodies. For example, of these four antigens, only the one recognized by AF5 is detectable by Western immunoblotting. Loss of immunoreactivity of the other antigens is possibly due to methanol in the transfer buffer, alkylation or crosslinking by bromphenol blue, or heating of the samples either before loading or during electrophoresis. Furthermore, these antigens have proven difficult to study by immunohistochemistry due to the physiochemical manipulations required for preservation of tissue morphology (Hancock et al., 1982; Gatter et al., 1984; Ciocca et al., 1986; DeLellis et al., 1987; Pollard et al., 1987). These limitations have led us to use alternative protocols such as immunoprecipitation and fresh frozen tissue immunohistology to study these potentially important tumor associated-antigens. As monoclonal antibodies continue to be employed in the research and clinical laboratories, determination of epitope stability by rapid and simple techniques will become increasingly valuable (Hancock et al., 1982). This study was able to define some of the environments in which epitopes on six well defined tumor-associated antigens will remain stable. Indeed, before obtaining the results of the simple 'experiments described here, we had experienced failed attempts at routine histology, cloning, and purification of these fragile tumor-associated antigens, and some assay systems gave misleading false negative results by failing to detect the antigens in tissues and cells. Using several assay techniques and proper controis will minimize such misleading results, but determining the stability of the epitopes early on in the course of investigation greatly aids in interpreting results and planning further experimental strategies. In addition to the practical nature of these studies, the unique susceptibility of each antigen to environmental modifications in and of itself serves to biophysically characterize these tumor-associated antigens.
Acknowledgements The authors thank Drs. M. Nishiyama, P. Mottt, and M. Frohlich for information on the distribution and molecular weights of the AF5 and FB50 antigens; R. Carlson for superb technical assistance; and Dr. Hubert E. Blum for his careful reading of the manuscript and thoughtful suggestions. The authors also thank Ms. Kristin Cambria-Shaw for her assistance in preparing the manuscript for publication.
References Agnello, V. and Barnes J.L. (1986) Human rheumatoid factor crossidiotypes WA and BLA are heat-labile conformational antigens requiring both heavy and light chains. J. Exp. Med. 164, 1809. Ahem, T.J. and Klibanov, A.M. (1985) The mechanism of irreversible enzyme inactivation at 100°C. Science 228, 1280. Ahem, T.J. and Klibanov, A.M. (1988) Analysis of processes causing thermal inactivation of enzymes. Methods Biochem. Anal. 33, 91. Baldwin, R.L. (1986) Temperature dependence of the hydrophobic interaction in protein folding. Proc. Natl. Acad. Sci. U.S.A. 83, 8069. Baldwin, R.L. and Eisenberg, D. (1987) Protein stability. In: D.L. Oxender and C.F. Fox (Eds.), Protein Engineering. Alan R. Liss, New York, p. 127. Ciocca, D.R. and Bjercke, R.J. (1986) Immunohistochemical techniques using monoclonal antibodies. Methods Enzymol. 121, 562. Creighton, T.E. (1987) Protein folding. Protein stability. In: D.L. Oxender and C.F. Fox (Eds.), Protein Engineering. Alan R. Liss, New York, p. 83. Cupane, A., Giacomazza, D. and Cordone, L. (1982) Kinetics of thermal denaturation of methemoglobin in perterbed solvent: relevance of bulk-electrostatic and hydrophobic interactions. Biopolymers 21,1081. DeLellis, R.A. and Dayal, Y. (1987) The role of immunochemistry in the diagnosis of poorly diffferentiated malignant neoplasms. Semin. Oncol. 14, 173. Dorrington K.J. and Bennich, H. (1973) Thermally induced structural changes in immunoglobulin Eur. J. Biol. Chem. 248, 8378. Ey, P.L., Prowse, S.J. and Jenkins, C.R. (1978) Isolation of pure IgG1, IgG2a, and IgG2b immunoglobulins from mouse serum using protein A-sepharose. Immunochemistry 15, 429. Frohlich. M., Mottt, P., Galvin, K., Takahashi, H. and Wands, J.R. (1990) Identification of a hepatoma-associated antigen as the p36 substrate for growth factor receptor and retroviral protein kinases. Mol. Cell Biol. 10, 3216.
64 Fujita, Y., Miyanaga, A. and Noda Y. (1979) Effect of alcohols on the thermal denaturation of lysozyme as measured by differential scanning calorimetry. Bull. Chem. Soc. Jpn. 52, 3659. Fujita, Y., lsawa, Y. and Noda, Y. (1982) The effect of polyhydric alcohols on the thermal denaturation of lysozyme as measured by differential scanning calorimetry. Bull. Chem. Soc. Jpn. 55, 1896. Gatter, K.C., Alcock, C., Heryet, A., Pulford, K.A., Hyderman, E., Taylor-Papadimitrou, J., Stein, H. and Mason, D.Y. (1984) The differential diagnosis of routinely processed anaplastic tumors using monoclonal antibodies. Am. J. Clin. Pathol. 82, 33. Hancock, W.W., Becker, G.J. and Atkins, R.C. (1982) A comparison of fixatives and immunohistochemical technics for use with monoclonal antibodies to cell surface antigens. Am. J Clin. Pathol. 78, 825. Hastings, G.Z., Burt, D.S. and Stanworth, D.R. (1988) A monoclonal antibody raised against a synthetic peptide representative of part of the amino acid sequence of rat immunoglobulin E detects thermally induced changes in that region of the IgE molecule. Immunology. 65, 149. Klibanov, A.M. and Ahern, T.J. (1987) Thermal Stability of Proteins. In: D.L. Oxender and C.F. Fox (Eds.), Protein Engineering. Alan R. Liss, New York, p. 213. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 277, 680. Linsenmayer, T.F., Gibney, E., Fitch, J.M., Gross, J. and Mayne, R. (1984) Thermal stability of the helical structure to type IV collagen within basement membranes in situ: determination with a conformation-dependent monoclonal antibody. J. Cell Biol. 99, 405. Lun, H., Isselbacher, K.J., Wands, J.R., Goodman, H.M., Shih, C. and Quaroni, A. (1984). Establishment and characterization of a new human hepatocellular carcinoma cell line. In Vitro 20, 493. Oda, Y. and Yasai, K. (1984). Monoclonal antibodies against fl-galactoside-binding lectin of chick embryo recognizes conformational change of the antigen. Biochem. Biophys. Res. Commun. 23, 1179. Ozturk, M., Motte, P., Takahashi, H., Frohlich, M. Wilson, B., Cummings, W., Bressac, B. and Wands J.R. (1989) Identification and characterization of a M r 50,000 in human hepatocellular carcinoma. Cancer Res. 49, 6764-6773. Pollard, K., Lunny, D., Holgate, C.S., Jackson, P.K. and Bird, C.C. (1987) Fixation, processing, and immunochemical reagent effects on preservation of T-lymphocyte surface membrane antigens in paraffin-embedded tissue. J. Histochem. Cytochem. 35, 1329.
Tainer, J.A., Getzoff, E.D., Alexander, H., Houghten, R.A., Olsen, A.J., Lerner, R.A. and Henderickson, W.A. (1984) The reactivity of anti-peptide antibodies is a function of the atomic mobility of sites in a protein. Nature 312, 127. Takahashi, H., Ozturk, M., Wlson, B., Maki, A., Ozawa, K., Koizumi, M., Endo, K., Strauss, W., Shouval, D. and Wands J.R. (1988) In vivo expression of two novel tumor associated antigens and their use in immunolocalization of human hepatocellular carcinoma. Hepatology 9, 625. Takahashi, H., Ozturk, M., Wlson, B., Maki, A., Ozawa, K., Koizumi, M., Endo, K., Strauss, W., Shouval, D. and Wands J.R. (1989a) Radioimmunolocation of hepatic and pulmonary metastasis of human colon adenocarcinoma. Gastroenterology 96,1317. Takahashi, H., Wilson, B., Ozturk, M., Motte, P., Strauss, W., Isselbacher, K.J. and Wands, J.R. (1989b) In vivo localization of colon adenocarcinoma by monoclonal antibody binding to a highly expressed cell surface antigen. Cancer Res. 48, 6573. Tanford, C. (1968) Protein denaturation. Adv. Prot. Chem. 23,121. Teipel J.W., Koshland, Jr., D.E. (197I) Kinetic aspects of conformational changes in proteins. If. Structural changes in renaturation of denatured proteins. Biochemistry 10, 798. Vita, C., Fontana, A. and Chaiken, I.M. (1985) Folding of thermolysin fragments. Eur. J. Biochem. 151,191. Vitetta, E.S., Fulton, R.J., May, R.D., Till, M. and Uhr, J.W. (1987) Redesigning nature's poisons to create anti-tumor reagents. Science 238, 1098. Wands, J.R. and Zurawski, Jr., V.R. (1981) High affinity monoclonal antibodies to hepatitis B surface antigen (HBsAg) produced by somatic cell hybrids. Gastroenterology 80, 225. Westhof, E., Altschuh, D., Moras, D., Bloomer, A.C., Mondragon, A., Klug, A. and Van Regenmortel, M.H.V. (1984) Correlation between segmental mobility and the location of antigenic determinants in proteins. Nature 311. 123. Wilson, B., Ozturk, M., Takahashi, H., Motte, P., Kew, M.C., Isselbacher, K.J. and Wands, J.R. (1988) Cell surface changes associated with transformation of human hepatocytes to the malignant phenotype. Proc. Natl. Acad. Sci. U.S.A. 85, 3140. Zaks, A. and Klibanov, A.M. (1984) Enzymatic catalysis in organic media at 100 o C. Science 224, 1249. Zaks, A. and Klibanov, A.M. (1988) Enzymatic catalysis in nonaqueous solvents. J. Biol. Chem. 263, 3194.
55
© 1991 Elsevier Science Publishers B.V. 0022-1759/91/$03.50 ADONIS 002217599100160A
JIM 05912
Stability of monoclonal antibody-defined epitopes * Byron E. Wilson, Susan Sun, M e h m e t O z t u r k a n d Jack R. W a n d s Molecular Hepatology Laboratory, Massachusetts General Hospital Cancer Center, Charlestown, MA 02129, U.S.A., and Department of Medicine, Harvard Medical School, Boston, MA 02114, U.S.A. (Received 28 December 1990, accepted 15 January 1991)
Epitope instability can limit the applications of monoclonal antibody (mAb) technology in laboratory and clinical research. We exposed a group of representative antigens on human hepatocellular carcinoma (HCC) cells to physiochemical insults to study epitope stability as measured by mAb immunoreactivity. Each epitope was found to have a unique pattern of instability which serves to biophysically characterize the antigen and de'fines the conditions to which the antigen can be exposed during laboratory and clinical investigations. Individual antigens were found to be unstable within a surprisingly well defined window of solvent polarities while being stable on either side of that window. Several antigens were observed to be unstable when exposed to transient changes in pH. When a critical temperature between 42 ° C and 65 °C was achieved, epitopes which were thermosensitive underwent a sudden loss in immunoreactivity. This critical temperature was found to be pH dependent. The effects of polarity, pH, and temperature on epitope stability are consistent with changes in protein structural conformation. In addition, we found that certain fixatives cause a time and concentration dependent loss of epitope immunoreactivity. This study provides a rapid and easy determination of monoclonal antibody-defined epitope stability; the results of which serve to guide further studies on the antigen and to characterize the antigen on the basis of its unique physiochemical stability. Key words." Cell surface antigen; Epitope; Monoclonal antibody
Introduction
The development of monoclonal antibody (mAb) technology has led to a new generation of
Correspondence to: J.R. Wands, Molecular Hepatology Laboratory, M G H Cancer Center, 7th Floor, M G H East, 149 13th Street, Charlestown, MA 02129, U.S.A. * This work was supported in part by Grants CA-3571, AA02666, AA-01862 and HD-20469 from the National Institutes of Health and a grant from Association pour la Recherche sur le Cancer, Villejuif, France. J.R.W. is the recipient of a Research Scientist Award AA-00048.
research and clinical reagents. Morloclonal antibodies are now instrumental in antigen identification, characterization, and purification; in cloning of translated genetic products; and in immunohistology for investigative and diagnostic purposes (Gatter et al., 1984; Tainer et al., 1984; DeLellis et al., 1987; Takahashi et al., 1988, 1989a, b; Ozturk et al., 1989). In addition, monoclonal antibodies are becoming increasing useful as potential in vivo immunotargeting and therapeutic reagents. (Vitetta et al., 1987; Takahashi et al., 1988, 1989a, b). cad The advantage of monoclonal over polyclonal antibodies is derived from the exquisite specificity
56 afforded by single epitope recognition. However, the potential applications of a given mAb can be limited if the epitope recognized by that antibody is unstable. For example, immunohistological techniques often require extensive processing of tissue specimens for preservation of morphology; Western immunoblotting requires antigen solubilization and often denaturation; protein purification may involve acidic and alkaline conditions, use of polar and nonpolar solvents, or thermal precipitation. Under such non-physiological conditions, a given epitope's immunoreactivity may be destroyed or substantially altered. Polyclonal antibodies often recognize a spectrum of many different epitopes such that physiochemical modifications destroy only a subset of epitopes, resulting in a lowered, but generally not completely lost, total immunoreactivity (Vita et al., 1985; DeLellis et al., 1987). In contrast, as a single epitope system, the immunoreactivity of an antigen to a monoclonal antibody may be completely lost following a physiochemical insult (Hancock et al., 1982; Garter et al., 1984; Oda et al., 1984; Ciocca et al., 1986; DeLellis et al., 1987; Pollard et al., 1987). Determining the stability of a given epitope provides a partial characterization of the antigen. For example, changes in immunoreactivity following physiochemical insult have been used to differentiate between (and define the structural basis for) rheumatoid factor crossidiotypes (Agnello et al., 1986); to detect subtle differences in type IV collagen between tissues (Linsenmayer et al., 1984); to study the conformational substructure of thermolysin (Vita et al., 1985); and to study the relationship between structure and function for IgE domains containing the mast cell receptor binding site (Dorrington et al., 1973; Hastings et al., 1988). On a more practical level, determining the stability of a given epitope delineates the conditions necessary for use of the monoclonal antibody as an investigative or clinical tool (Hancock et al., 1982; Oda et al., 1984; Pollard et al., 1987). We have developed methods by which one can readily determine the stability of epitopes to various physiochemical conditions often encountered in laboratory and clinical research, and have applied these methods to a panel of well defined tumor-associated antigens.
Materials and methods
Cell culture The hepatocellular carcinoma cell line FOCUS developed by us (Lun et al., 1984) was maintained in Eagle's minimum essential medium (M.A. Bioproducts, Walkersville, MD) supplemented with 10 /~M of nonessential amino acids and 10% fetal bovine serum inactivated at 56°C. Immunizing cells were grown from the second passage of the FOCUS cell line and were harvested from monolayer culture with EDTA/Versene in the absence of trypsin, then washed with PBS. Monoclonal antibody production 6-week-old B A L B / c female mice were first immunized by intraperitoneal injection of 4.0 x 10 6 FOCUS cells emulsified in 50% (v/v) complete Freund's adjuvant. Secondary immunizations were performed 6-10 weeks later by intravenous injection of 4.0 × 106 intact cells in 0.2 ml of PBS. 3 days later, mouse splenocytes were fused with either SP20, NS1, or X63 mouse myeloma cell lines with 30% (w/v) polyethylene glycol. Resulting hybridomas were maintained and selected as described (Laemmli et al., 1970). Screening for mAbs was performed by the indirect binding assay as described (Wilson et al., 1988). Purification and iodination of mAbs were performed as described (Takahashi et al., 1989; Ey et al., 1990). Properties of human tumor associated antigens We have selected six anti-FOCUS monoclonal antibodies, named AF5, AF20, FB50, SF25, SF90 and XF8, for use in this study. All these mAbs recognize antigens which share the common feature of having a high level of expression in hepatocellular carcinoma (HCC) with little or no expression in adjacent uninvolved liver (Wilson et al., 1988; unpublished results). Three antibodies (AF5, FB50, and SF90) recognize proteins identifiable by Western immunoblotting. Furthermore these mAbs are known to recognize epitopes contained within the primary structure of the protein. The antigen recognized by AF5 is a 36 kDa phosphoprotein, which in addition to HCC tissue, has been found on all transformed cell lines tested (Frohlich et al., 1990). The
57 mAb FB50 recognizes two families of 50 kDa and 100 kDa proteins (data not shown). SF90 recognizes a 50 kDa protein distinct from that recognized by FB50 (Ozturk et al., 1989). In addition to high HCC expression, the FB50 and SF90 antigens are highly expressed in normal adrenal glands but are expressed at considerably lower levels in other normal human tissues. The SF90 antigen has been present on all malignant human cells thus far tested. AF20, SF25, and XF8 recognize antigens highly expressed in adenocarcinomas of the colon, lung and liver and most other malignant tumors, with little if any expression on normal human tissues (Takahashi et al., 1988, 1989a). None of these three antigens are demonstrable by Western immunoblotting. All three antigens have been successful in imaging tumors in nude mouse models (Takahashi et al., 1988, 1989a, b). Although none of these antigens are demonstrable by Western immunoblotting, all have been fully characterized by metabolic labeling and immunoprecipitation (unpublished results). All three antibodies recognize epitopes found on the protein cores of cell surface glycoptoteins. XF8 and AF20 recognize dissimilar epitopes on a 180 kDA homodimer which is also detectable in normal adrenal glands. SF25 recognizes a 130 kDa heterodimer also expressed by renal proximal tubule cells.
Cell homogenates Harvested FOCUS cells were centrifuged at 1000 × g for 10 min. The pellet was homogenized for 20 s on ice with a Polytron (Brinkman) in 50 vols. of PBS (pH 7.2) containing 0.1% NaN 3, to yield a final protein concentration of approximately 0.5 mg/ml. Filter binding assay Assays were performed in 96-well plates containing glass filters (V & P Scientific, San Diego, CA) as previously described (Wilson et al., 1988). All incubations were performed at room temperature unless otherwise specified. Briefly, FOCUS cell homogenate (50 or 100 lug of protein) was applied to each filter with gentle aspiration to trap proteins. Filters were then incubated with 250 lul of 100% calf serum (CS) for 15 min to block nonspecific protein binding sites. After filtering
off the calf serum, wells were preincubated for 15 min with 50% C S / P B S or with 100 lul of a 1/100 dilution of unlabeled blocking mAb in 50% CS/PBS. To each well, 105 cpm of 125I-labeled mAb diluted in 50% C S / P B S was added and incubated for 2 h. Filters were washed five times (1.5 ml/well) with 5% C S / P B S and counted for gamma radioactivity. The data obtained from several different experiments were normalized with respect to specific binding. Specific binding was determined by subtracting the binding of radiolabeled mAb in the presence of unlabeled homologous mAb (nonspecific binding) from the unblocked binding of radiolabeled mAb (total binding).
Antigen stability assays The effect on epitope immunoreactivity of exposing the antigens to physiochemical modifications was assayed using the above filter binding assay with the following experimental manipulations. The effect of formalin, glutaraldehyde and alcohols on epitope stability was studied by incubating the filter-bound FOCUS homogenate in the various reagents for one hour. The filters were then aspirated dry and washed five times with PBS pH 7.2 (1.5 ml/well) prior to addition of blocking calf serum. To study the effect of ethanol, acetone, and xylene, filters containing bound homogenate were gently removed from 96-well plates and incubated in the organic solvent for 1 h. The solvent was evaporated off, and the filters were replaced in the wells. The filter binding assay was then performed as above. Thus the effects of ethanol were studied by two methods, which gave nearly identical results, indicating that the observed immunoreactive modifying effects of ethanol were not due to washing away of antigen bound to the filter plates. The effect of pH on antigen stability was determined by incubating filters containing bound FOCUS homogenate with various pH solutions of PBS (200 lul). The pH solutions were then aspirated and filters were washed five times (1.5 ml/well) with 5% C S / P B S pH 7.2 prior to addition of blocking calf serum. An alternative method, which gave identical results, was to preincubate the FOCUS homogenate with various pH solutions for 30 min followed by neutralization of the solu-
58
tions to pH 7 prior to binding the homogenate to the filter. Thermoinactivation studies were conducted by heating FOCUS homogenates in glass tubes at the desired temperature for 1 h, or at 1 0 0 ° C for 20 min. After heating, the homogenates were allowed to gradually cool to room temperature to allow for any reversible processes to occur. Any precipitates were gently resuspended. The cell homogenates were then applied to filters and assayed as described above.
Kinetic studies T h e kinetics of t h e r m a l d e n a t u r a t i o n were studied by a d d i n g F O C U S h o m o g e n a t e to glass tubes p r e h e a t e d in a 75 ° C r o t a t i n g water bath. A t t i m ed intervals, the h o m o g e n a t e s were r e m o v e d f r o m the w at er b a t h an d further t h e r m o i n a c t i v a tion was q u e n c h e d with 1 ml of ice-cold PBS buffer. H o m o g e n a t e s were then l o a d e d o n t o filters for assay. T o study the effect of f o r m a l i n fixation with time, cell h o m o g e n a t e s were first b o u n d to filters
XF8
5°°°IAF5
t00C 80C 60C
L
20O0
40C
20C
4000 0c
C
SF25
AF20
IOOC
3000
80C 2000
Jr~ ii~ii
60C 40C
~000 20C C
SFgO
FB50
3000
5000 4000
2000
3000 2000
4000
4000
-~
o
z o
.~ ~
N
g~
o o
*~ ° o
=
~
~
~
z
o o
e
~
N
~
~
~ e 8
~
~
o
~
-
8_
Fig. l. The effect on epitope immunoreactivity of transient exposure to solvents and fixatives. For four epitopes (recognized by AF5, AF20, XF8 and SF25) there is a range of intermediate solvent polarities destructive to immunoreactivity. The positive control was normal saline.
59
then incubated with 200 ffl of 10% formalin in PBS for various time intervals. Filters were washed 5 times with PBS and assayed as described above.
~
-8
F20
• Melhonol i c Elhonor 1 IsoproponoI
Lb
t
20
-
15
Results
Six anti-FOCUS monoclonal antibodies (named AF5, AF20, FB50, SF25, SF90, and XF8) were selected for this study because the antigens they recognize have been previously well characterized with respect to tissue distribution, biochemical properties and some of their molecular characteristics (Takahashi et al.,, 1988, 1989a,b; Wilson et al., 1988; Ozturk et al., 1989; unpublished results). All are known to be at least partially protein in nature. Fig. 1 is a summary of three experiments on the effects of transient exposure to solvents and tissue fixatives on the immunoreactivity of these antigens. The solvents were selected to represent a broad spectrum of polarity. All epitopes were found to be stable in saline solutions. Four antigens (recognized by mAbs AF5, AF20, SF25, and XF8) were highly sensitive to treatment with 80% ethanol, a solvent of moderate polarity. Exposure to 100% ethanol, which is less polar, destroyed the XF8 and SF25 epitopes while only mildly affecting the immunoreactivities of the AF5 and AF20 epitopes. Treatment with acetone, a weakly polar solvent, reduced the immunoreactivities of only the XF8 and SF25 epitopes. Xylene, a non-polar solvent, did not have an effect on the immunoreactivities of any of the epitopes studied. Thus it appears that for four of the six antigens there is a range of solvent polarities that reduces immunoreactivity, whereas the other two antigens (recognized by FB50 and SF90) appear stable following exposure to all solvent polarities tested. To more closely examine the effect of polarity on epitope reactivity, ethanol solutions of varying concentrations were studied. As shown in Fig. 2, the immunoreactivity of the XF8 epitope drops quickly as ethanol concentration is raised (or as solvent polarity is lowered) and remains depressed at 100% ethanol. The AF5 and AF20 epitopes are also unstable in ethanol solutions. However in contrast to XFS, these two epitopes are more stable at the extremes of polarity. They are not
ic, I£
40
l ~
:5
40
80
0(
80
40
80
PERCENT ALCOHOL
Fig. 2. The effect on epitope immunoreactivity of transient exposure to alcohols diluted with water to achieve gradations of solvent polarities. The alcohols in order of decreasing polarity are methanol, ethanol, and isopropanol. The immunoreactivity curves for the antigens recognized by AF5 and AF20 shift according to alcohol polarity, indicating that a set range of solvent polarities are destructive to epitope immunoreactivity.
inactivated until solvent polarities lower than those which destroy the XF8 epitope are reached and they are only partially destroyed at the even lower polarity of 100% ethanol, a polarity at which XF8 is still unstable. The hypothesis that there is a window of intermediate solvent polarity within which given antigens are unstable but outside of which they are stable is further supported with the finding that an alcohol of lower polarity than ethanol (isopropanol) gave a curve shifted to the left of that for ethanol whereas an alcohol of
~l
5oo
800f XF8
E
40C
600
500
0
L 4
2
4OO
200
± 5
0
I / t I I I 60 t20 ~80
MINUTES
Fig. 3. The kinetics of epitope inactivation on heating to 75 ° C (A) and fixing with 3.7% formaldehyde (B). Thermodestruction of the epitope recognized by XF8 is very rapid and exhibits first order kinetics. The kinetics of inactivation by formaldehyde crosslinking is of high order and is concentration dependent (data not shown).
60
higher polarity (methanol) gave a curve shifted to the right of that for ethanol. It appears from the above studies, that the six epitopes can be divided into three groups based upon their stability in solvents of varying polarity: the FB50 and SF90 defined epitopes show stability in all polarities tested; the AF20 and AF5 epitopes show instability in a narrow range of polarities corresponding to solutions of 40-90% ethanol; and the XF8 and SF25 epitopes show a broad range of instability starting at high polarity (30% ethanol) and continuing past the low polarity of 100% acetone. Fig. 1 shows the effects of cross-linking fixatives on immunoreactivity. The epitopes recognized by five of the six antibodies studied (AF5, AF20, XF8, SF25, and FB50) were sensitive to 2% glutaraldehyde such that virtually all of the binding activity was destroyed. In contrast, formaldehyde substantially altered the immunoreactivity of only one epitope (XF8), and mildly lowered the reactivity of two others (AF20 and SF25). Thus we found glutaraldehyde to be the more immunodestructive of the two crosslinking agents; affecting even the remarkably stable epitope recognized by FB50. The fixative-induced decrease in immunoreactivity varied with concentration (data not shown) and duration of fixation. Fig. 3B demonstrates the time course of inactivation for the epitope recognized by XF8 using 3.7% formaldehyde. The AF20 and SF25 defined epitopes demonstrated similar kinetics, but with slightly longer half lives (data not shown). Fig. 4 illustrates results of four experiments on the effects of temporarily heating the antigens. The epitopes recognized by AF5 and SF25 are abruptly thermoinactivated between 4 2 ° C and 55°C whereas the XF8 and AF20 epitopes are thermoinactivated between 55 ° C and 65°C. The FB50 and SF90 epitopes were quite stable, even if boiled for 20 min. The destruction of thermal-sensitive epitopes occurs very rapidly. Fig. 3A illustrates two experiments on the kinetics of thermoinactivation for the XF8 epitope. Similar results were obtained for the antigens recognized by AF5 and AF20 (data not shown). The thermoinactivation of our epitopes occurred with half-lives of less than 30 s at 7 5 ° C (Fig. 3A); findings consistent with non-co-
i ,~F5
XF8
....
oo
4 0 0 0 I'
t5i~
2Z!I'
L
,oooi 50
19fI
.
-
: .r2o
! s~25 ~oooi ]
3500, -
25OO
800
_~
2O00 }
600
~:~
45001 1000
400 : 2O
53O 0.
.
.
.
.
~
F850
i
SF90
400O~
l <8o
'009 50C
!
:: 2::, 4( 6c a(, 4,-~, :, ~c 41o ~'o ~'; ~,:,_, TEMPERATURE
(°C)
]Pig. 4. Thermoinactivation of epitopes followed by slow cooling to allow for renaturation. Thermosensitive epitopes (recognized by AF5, AF20, XF8 and SF-25) underwent a sudden loss in immunoreactivity when a characteristic critical temperature was achieved.
valent thermoinactivation (Ahern et al., 1985, 1988). Fig. 5 shows four experiments depicting the effect on immunoreactivity of transiently changing [ xJF5 4~CL
XF8
ic~O
....} ~F20
6r25 eco
60O:) 4">0O
rex: i
?OCO
,
SF90
FB50
4000'
i q
z
4
6
8
40 12 ~4
9
4
6
~
,8
~? ,4
pH
Fig. 5. The effect on epitope immunoreactivityof transiently changing the pH. Susceptible epitopes were destroyed at the extremes of pH, but the decline in immunoreactivity as a function of pH differed betweenepitopes.
61
•
AF20 1oooo
8000
2000 O"
0
t0
20
30 40 50 60 70
TEMPERA TURE ( °C ) Fig. 6. The relation between p H and thermoinactivation of epitopes. Acidic conditions lower the temperature necessary for epitope inactivation.
the pH. The XF8 epitope was completely inactivated with a small change in pH above 10.0 or below 5.5. The epitopes recognized by AF5, AF20, and SF25 showed incremental destruction as the pH varied from an optimum of around 6. The FB50 and SF90 epitopes remained stable at all pHs. A relationship was observed between the temperature stability and pH stability of the epitopes studied. As shown in Fig. 6, we observed that as the pH is lowered, there is a decrease in the temperature necessary for reduction of immunoreactivity.
Discussion
The antigens studied here are of interest to us because the monoclonal antibodies directed against them are promising tools for studying the cellular biology of malignant transformation, and are potential reagents for immunodiagnosis and therapy (Takahashi et al., 1988, 1989a, b; Wilson et al., 1988; Ozturk et al., 1989; unpublished results). For this study, unpurified cellular antigens were used so as to best represent the conditions most often encountered when employing monoclonal antibodies for laboratory and clinical research purposes. Based on a simple filter binding assay, we have been able to rapidly produce individual profiles of epitope stability for each of the antigens studied.
The results indicate that epitope stability, as determined by immunoreactivity, can be affected by changes in solvent polarity, pH, and temperature and by presumed covalent alteration. This reflects the nature of the antigen-antibody interaction, as complete binding of an antibody to an epitope can only occur if the epitope region can assume a well defined three-dimensional configuration (Tainer et al., 1984; Westhof et al., 1984; Vita et al., 1985; Hastings et al., 1988). As confirmed by the present investigation, when the ability of the epitope to assume that configuration is hampered by denaturation, crosslinking, or degradation then immunoreactivity will be lost (Dorrington et al., 1973; Hancock et al., 1982; Linsenmayer et al., 1984; Oda et al., 1984; Vita et al., 1985; Agnello et al., 1986; Pollard et al., 1987). For example, degradation by at least one or more protease abolished the immunoreactivity of every epitope presented here (data not shown). Furthermore, when treated with crosslinking agents such as glutaraldehyde or formaldehyde, the immunoreactivities of several of the epitopes were destroyed (Fig. 1). Others have reported similar results (Hancock et al., 1982; Ciocca et al., 1986; DeLellis et al., 1987; Pollard et al., 1987). Perhaps the crosslinking bonds chemically or configurationally alter the epitope or they shield the epitope from antibody binding; further studies are required. Nonetheless, given the kinetics of epitope inactivation, it is apparent that controlling duration and concentration of fixation is critically i m p o r t a n t when performing routine immunohistology. It has been proposed that the stability of a given protein (or epitope) conformation is determined by the free energy of the system composed of the solvent, the protein in the given conformation, and other interacting molecules (lipids, ions, solutes, and other proteins) (Tanford, 1968; Baldwin et al., 1987; Creighton et al., 1987; DeLellis et al., 1987). A protein in the native folded state achieves a low free energy by forming the most optimal bonds available and by maximizing the freedom of solvent movement. However this lowering of free energy is offset by a resultant loss in freedom of protein movement compared to that found in the randomly unfolded state (Westhof et al., 1984; Baldwin, 1986). A transition to or
62 from the protein's native immunoreactive state will occur if it results in a lowering of the overall free energy. However, even if the folded state is energetically more favorable than the unfolded state, refolding may not occur: kinetic barriers, energetically unfavorable intermediate states, and antigen aggregation may prevent refolding of the denatured protein (Teipel et al., 1971; Dorrington et al., 1973; Baldwin et al., 1987; Klibanov et al., 1987; Ahern et al., 1988). As temperature rises the kinetic energy and thus the random motion of molecules also increase. Consequently states with more randomness become favorable and the protein usually unfolds (Baldwin, 1986; Kilbanov et al., 1987). As the protein is cooled back to room temperature it will assume the state with lowest free energy unless inhibited from doing so (Teipel et al., 1971; Baldwin et al., 1987; Ahern et al., 1988). Indeed, two epitopes (SF90 and FB50) maintained their immunoreactivity even after boiling, suggesting that for these epitopes the native state is energetically favorable (Fig. 4). In contrast, the other four antigens studied here presumably unfolded at sharply defined temperatures within a range of 42 ° C-65 ° C and their epitopes remained non-reactive upon cooling. Others have reported similar results (Dorrington et al., 1973; Hancock et al., 1982; Linsenmayer et al., 1984; Oda et al., 1984; Agnello et al., 1986; Pollard et al., 1987). For each of our antigens we found a highly reproducible and unique response to transient changes in pH (Fig. 5). Four epitopes (recognized by AF5, AF20, SF25, and XF8) lose their immunoreactivities when exposed to acidic or alkaline conditions. In contrast, the FB50 and SF90 epitopes appear stable throughout the pH range tested. The temperature necessary for AF20 epitope thermoinactivation decreased as the pH of the solution was decreased (Fig. 6). It appears that the native folded state becomes more susceptible to thermodenaturation when destabilized by changes in the p H of the solvent (Tanford, 1968; Fujita et al., 1979, 1982; Cupane et al., 1982; Zaks et al., 1984, 1988; Baldwin et al., 1987; Klibanov et al., 1987; Ahem et al., 1988). Figs. 1 and 2 show that epitopes recognized by AF20 and AF5 are unstable in a range of polarities corresponding to solutions of 40-90% ethanol,
and the XF8 and SF25 epitopes are unstable in a wider range of polarity: from 30% ethanol down to 100% acetone. Outside of these polarity ranges the antigens are stable. The FB50 and SF90 epitopes are able to resume immunoreactive configurations on restoration to a normal saline solvent. The non-polar moieties of proteins in polar solvents are internalized as a result of solvent shell formation, while in low polarity solvents the bonds between polar moieties become more favorable (Fujita et al., 1979, 1982; Cupane et al., 1982; Zaks et al., 1984, 1988; Baldwin, 1986, 1987; Creighton et al., 1987; Ahem et al., 1988). These solvent effects stabilize the folded protein. Consequentially, one would expect the native states of proteins to be stable in high polarity aqueous solvents as well as in low polarity solvents, but to be unstable in intermediate polarity solvents where neither of these two solvent effects are sufficient to stabilize the native state. Our experimental results fit this model well. On a practical level our results indicate that for certain antigens there is a window of solvent polarity which is destructive to immunoreactivity and must be avoided. Unfortunately many experimental techniques use solvents of these intermediate polarities. For example, graded solutions of alcohol are often used to dehydrate tissues or to destroy endogenous peroxidases for immunohistology (Ciocca et al., 1986; Pollard et al., 1987), and the transfer buffer for Western immunoblotting often contains an alcohol (Laemmli, 1970). Our results suggest other buffers and other immunohistological methods must be employed when studying susceptible antigens. We have demonstrated that epitope stability is likely to be analogous to protein stability (Tanford, 1968). Alterations in protein stability explain the effects of temperature, pH, and solvent polarity on epitope stability, showing that each environmental manipulation effects epitope stability. We found that the native folded state of the epitopes recognized by FB50 and SF90 are the most stable. Consequently, these epitopes are not irreversibly destroyed by changes in pH, temperature, and solvent polarity. These proteins should therefore be amenable to the wide ranges of environmental manipulation involved in characterization, purification, cloning, and cytology.
63
In contrast, the other four antigens (recognized by mAbs AF5, AF20, SF25, and XF8) were found to be unstable to many physiochemical insults, and were not able to resume their native folded states with removal of these insults. Therefore, many standard protocols have to be modified to obtain reliable results with these antibodies. For example, of these four antigens, only the one recognized by AF5 is detectable by Western immunoblotting. Loss of immunoreactivity of the other antigens is possibly due to methanol in the transfer buffer, alkylation or crosslinking by bromphenol blue, or heating of the samples either before loading or during electrophoresis. Furthermore, these antigens have proven difficult to study by immunohistochemistry due to the physiochemical manipulations required for preservation of tissue morphology (Hancock et al., 1982; Gatter et al., 1984; Ciocca et al., 1986; DeLellis et al., 1987; Pollard et al., 1987). These limitations have led us to use alternative protocols such as immunoprecipitation and fresh frozen tissue immunohistology to study these potentially important tumor associated-antigens. As monoclonal antibodies continue to be employed in the research and clinical laboratories, determination of epitope stability by rapid and simple techniques will become increasingly valuable (Hancock et al., 1982). This study was able to define some of the environments in which epitopes on six well defined tumor-associated antigens will remain stable. Indeed, before obtaining the results of the simple 'experiments described here, we had experienced failed attempts at routine histology, cloning, and purification of these fragile tumor-associated antigens, and some assay systems gave misleading false negative results by failing to detect the antigens in tissues and cells. Using several assay techniques and proper controis will minimize such misleading results, but determining the stability of the epitopes early on in the course of investigation greatly aids in interpreting results and planning further experimental strategies. In addition to the practical nature of these studies, the unique susceptibility of each antigen to environmental modifications in and of itself serves to biophysically characterize these tumor-associated antigens.
Acknowledgements The authors thank Drs. M. Nishiyama, P. Mottt, and M. Frohlich for information on the distribution and molecular weights of the AF5 and FB50 antigens; R. Carlson for superb technical assistance; and Dr. Hubert E. Blum for his careful reading of the manuscript and thoughtful suggestions. The authors also thank Ms. Kristin Cambria-Shaw for her assistance in preparing the manuscript for publication.
References Agnello, V. and Barnes J.L. (1986) Human rheumatoid factor crossidiotypes WA and BLA are heat-labile conformational antigens requiring both heavy and light chains. J. Exp. Med. 164, 1809. Ahem, T.J. and Klibanov, A.M. (1985) The mechanism of irreversible enzyme inactivation at 100°C. Science 228, 1280. Ahem, T.J. and Klibanov, A.M. (1988) Analysis of processes causing thermal inactivation of enzymes. Methods Biochem. Anal. 33, 91. Baldwin, R.L. (1986) Temperature dependence of the hydrophobic interaction in protein folding. Proc. Natl. Acad. Sci. U.S.A. 83, 8069. Baldwin, R.L. and Eisenberg, D. (1987) Protein stability. In: D.L. Oxender and C.F. Fox (Eds.), Protein Engineering. Alan R. Liss, New York, p. 127. Ciocca, D.R. and Bjercke, R.J. (1986) Immunohistochemical techniques using monoclonal antibodies. Methods Enzymol. 121, 562. Creighton, T.E. (1987) Protein folding. Protein stability. In: D.L. Oxender and C.F. Fox (Eds.), Protein Engineering. Alan R. Liss, New York, p. 83. Cupane, A., Giacomazza, D. and Cordone, L. (1982) Kinetics of thermal denaturation of methemoglobin in perterbed solvent: relevance of bulk-electrostatic and hydrophobic interactions. Biopolymers 21,1081. DeLellis, R.A. and Dayal, Y. (1987) The role of immunochemistry in the diagnosis of poorly diffferentiated malignant neoplasms. Semin. Oncol. 14, 173. Dorrington K.J. and Bennich, H. (1973) Thermally induced structural changes in immunoglobulin Eur. J. Biol. Chem. 248, 8378. Ey, P.L., Prowse, S.J. and Jenkins, C.R. (1978) Isolation of pure IgG1, IgG2a, and IgG2b immunoglobulins from mouse serum using protein A-sepharose. Immunochemistry 15, 429. Frohlich. M., Mottt, P., Galvin, K., Takahashi, H. and Wands, J.R. (1990) Identification of a hepatoma-associated antigen as the p36 substrate for growth factor receptor and retroviral protein kinases. Mol. Cell Biol. 10, 3216.
64 Fujita, Y., Miyanaga, A. and Noda Y. (1979) Effect of alcohols on the thermal denaturation of lysozyme as measured by differential scanning calorimetry. Bull. Chem. Soc. Jpn. 52, 3659. Fujita, Y., lsawa, Y. and Noda, Y. (1982) The effect of polyhydric alcohols on the thermal denaturation of lysozyme as measured by differential scanning calorimetry. Bull. Chem. Soc. Jpn. 55, 1896. Gatter, K.C., Alcock, C., Heryet, A., Pulford, K.A., Hyderman, E., Taylor-Papadimitrou, J., Stein, H. and Mason, D.Y. (1984) The differential diagnosis of routinely processed anaplastic tumors using monoclonal antibodies. Am. J. Clin. Pathol. 82, 33. Hancock, W.W., Becker, G.J. and Atkins, R.C. (1982) A comparison of fixatives and immunohistochemical technics for use with monoclonal antibodies to cell surface antigens. Am. J Clin. Pathol. 78, 825. Hastings, G.Z., Burt, D.S. and Stanworth, D.R. (1988) A monoclonal antibody raised against a synthetic peptide representative of part of the amino acid sequence of rat immunoglobulin E detects thermally induced changes in that region of the IgE molecule. Immunology. 65, 149. Klibanov, A.M. and Ahern, T.J. (1987) Thermal Stability of Proteins. In: D.L. Oxender and C.F. Fox (Eds.), Protein Engineering. Alan R. Liss, New York, p. 213. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 277, 680. Linsenmayer, T.F., Gibney, E., Fitch, J.M., Gross, J. and Mayne, R. (1984) Thermal stability of the helical structure to type IV collagen within basement membranes in situ: determination with a conformation-dependent monoclonal antibody. J. Cell Biol. 99, 405. Lun, H., Isselbacher, K.J., Wands, J.R., Goodman, H.M., Shih, C. and Quaroni, A. (1984). Establishment and characterization of a new human hepatocellular carcinoma cell line. In Vitro 20, 493. Oda, Y. and Yasai, K. (1984). Monoclonal antibodies against fl-galactoside-binding lectin of chick embryo recognizes conformational change of the antigen. Biochem. Biophys. Res. Commun. 23, 1179. Ozturk, M., Motte, P., Takahashi, H., Frohlich, M. Wilson, B., Cummings, W., Bressac, B. and Wands J.R. (1989) Identification and characterization of a M r 50,000 in human hepatocellular carcinoma. Cancer Res. 49, 6764-6773. Pollard, K., Lunny, D., Holgate, C.S., Jackson, P.K. and Bird, C.C. (1987) Fixation, processing, and immunochemical reagent effects on preservation of T-lymphocyte surface membrane antigens in paraffin-embedded tissue. J. Histochem. Cytochem. 35, 1329.
Tainer, J.A., Getzoff, E.D., Alexander, H., Houghten, R.A., Olsen, A.J., Lerner, R.A. and Henderickson, W.A. (1984) The reactivity of anti-peptide antibodies is a function of the atomic mobility of sites in a protein. Nature 312, 127. Takahashi, H., Ozturk, M., Wlson, B., Maki, A., Ozawa, K., Koizumi, M., Endo, K., Strauss, W., Shouval, D. and Wands J.R. (1988) In vivo expression of two novel tumor associated antigens and their use in immunolocalization of human hepatocellular carcinoma. Hepatology 9, 625. Takahashi, H., Ozturk, M., Wlson, B., Maki, A., Ozawa, K., Koizumi, M., Endo, K., Strauss, W., Shouval, D. and Wands J.R. (1989a) Radioimmunolocation of hepatic and pulmonary metastasis of human colon adenocarcinoma. Gastroenterology 96,1317. Takahashi, H., Wilson, B., Ozturk, M., Motte, P., Strauss, W., Isselbacher, K.J. and Wands, J.R. (1989b) In vivo localization of colon adenocarcinoma by monoclonal antibody binding to a highly expressed cell surface antigen. Cancer Res. 48, 6573. Tanford, C. (1968) Protein denaturation. Adv. Prot. Chem. 23,121. Teipel J.W., Koshland, Jr., D.E. (197I) Kinetic aspects of conformational changes in proteins. If. Structural changes in renaturation of denatured proteins. Biochemistry 10, 798. Vita, C., Fontana, A. and Chaiken, I.M. (1985) Folding of thermolysin fragments. Eur. J. Biochem. 151,191. Vitetta, E.S., Fulton, R.J., May, R.D., Till, M. and Uhr, J.W. (1987) Redesigning nature's poisons to create anti-tumor reagents. Science 238, 1098. Wands, J.R. and Zurawski, Jr., V.R. (1981) High affinity monoclonal antibodies to hepatitis B surface antigen (HBsAg) produced by somatic cell hybrids. Gastroenterology 80, 225. Westhof, E., Altschuh, D., Moras, D., Bloomer, A.C., Mondragon, A., Klug, A. and Van Regenmortel, M.H.V. (1984) Correlation between segmental mobility and the location of antigenic determinants in proteins. Nature 311. 123. Wilson, B., Ozturk, M., Takahashi, H., Motte, P., Kew, M.C., Isselbacher, K.J. and Wands, J.R. (1988) Cell surface changes associated with transformation of human hepatocytes to the malignant phenotype. Proc. Natl. Acad. Sci. U.S.A. 85, 3140. Zaks, A. and Klibanov, A.M. (1984) Enzymatic catalysis in organic media at 100 o C. Science 224, 1249. Zaks, A. and Klibanov, A.M. (1988) Enzymatic catalysis in nonaqueous solvents. J. Biol. Chem. 263, 3194.