EXPERIMENTAL
CELL
RESEARCH
1%,182-188
(1991)
Accessibility to Intracellular Antigens within Nutritionally Deprived Human Mammary Epithelial Cells SHAHNAZ
H. DAIRKEE,’ Per&a
LYNN PUETT, ANNA M. COUNELIS, Cancer
Research
Institute,
Inc.
INTRODUCTION We have recently demonstrated selective localization of a monoclonal antibody to intracellular, cytoskeletal keratin polypeptides in malignant breast epithelium under conditions of passive infusion of antibody in fresh surgical specimens of breast carcinoma [l]. These observations are closely analogous to those reported for the targeting of experimental myocardial infarcts using antibodies to another intracellular component, myosin [21, and may have important implications for tumor targeting. The uptake of antibody in the infarcts was reportedly greatest in regions of most severe blood flow reduction, and a subsequent histologic study correlated antimyosin binding with severe cellular necrosis [3], 1 To whom reprint keley Laboratory, 934.47A, University
requests should be addressed at Lawrence Division of Cell and Molecular Biology, of California, Berkeley, CA 94720.
Copyright All
rights
0
1991 by of reproduction
BerMS
182
0014.4827/91$3.00
Academic in any
Press, Inc. form
reserved.
California
94609
suggesting that loss of cell membrane integrity results in accessibility of macromolecules to intracellular sites. Our findings on antibody-infused solid tumor tissue have unexpectedly demonstrated antibody localization in the cytoplasm of tumor cells displaying visible cellular integrity without apparent signs of necrosis. One possible interpretation of this observation is that loss of membrane integrity is a progressive phenomenon and due to sustained metabolic, physiologic, and/or genetic limitations on the cell it eventually terminates in frank cellular necrosis and disintegration. The earliest signs of membrane damage thus may not be identifiable as specific morphologic changes but may possibly be measured in functional terms, simply by uncontrolled cellular uptake of macromolecules. In this regard, a two-dimensional, multicellular, in vitro model of a self-created gradient of nutrients and metabolic products, developed and described by Hlatky and Alpen [4] as the “sandwich system,” provides the unique opportunity for visualization and manipulation of cells at various stages in the entire sequence of events leading to frank necrosis. We have used here the sandwich system in conjunction with antibodies directed at intracellular antigens to serve as “indicator molecules” for determining the spectrum of cells in a diffusion gradient which can be potentially targeted by these reagents.
We have previously demonstrated immunolocalization of antikeratin antibodies in apparently random subpopulations of malignant cells in fresh surgical specimens of breast carcinoma (S. H. Dairkee and A. J. Hackett, 1988, J. Natl. Cancer Inst. 80, 1216-1220). The goal of the present study was to determine whether deficiencies in essential nutrients contribute toward cellular alterations in membrane integrity, consequently allowing antikeratin to bind to the cytoskeleton within live, unfixed cells. We have demonstrated here that in an in vitro model in which human mammary epithelial cells are subjected to an oxygen-glucose gradient, immunolocalization of antikeratin within the cells is observed in a dose-dependent manner in the depleted regions of the gradient, even though the cells appear to be morphologically unaltered. The potential use of antibodies to intracellular antigens for immunotargeting solid tumors and the use of this method in antibody-loading studies toward understanding functional aspects of specific cellular antigens, as well as determining differential response of various cell types under these culture conditions, are discussed. Q 1991 Academic Press.
Oakland,
AND ADELINE J. HACKETT
MATERIALS
AND METHODS
Cell culture. Normal human mammary epithelial cultures used in this study were initiated from reduction mammoplasty specimens. The methodology for culturing human mammary epithelial cells has been previously reported in detail [5]. The cells were propagated in the serum-free medium, MCDB 170, prepared as previously described [6]. Glucose concentration in fresh medium is 1.4 g/liter. Sandwich system. This in vitro assay was set up in our laboratory essentially as previously described (41 to serve as a model for studying the effects of a microenvironmental gradient on the membrane integrity of human mammary epithelial cells. Briefly, cells are grown as monolayers “sandwiched” between two glass microscope slides separated by a distance of 150 pm. A thin layer of growth medium covers the cells and fills the narrow gap between the slides. The sandwiching of cells in this manner causes all nutrients and waste products to move into or out of the local environment of the cells by diffusing through the narrow gap between the slides. Clean, sterilized microscope slides are first placed in 3.5 X 3.5-in. integrid petri dishes (Falcon). Cells are seeded at 2 X lo6 cells/dish in 12 ml of growth medium
NUTRIENT
DEFICIENCIES
STUDIED
and incubated at 37°C in 0.05% CO,. After 48-h of incubation, culture medium is renewed, 150.pm sterile glass spacers are placed at each end of the slide, and a top slide is laid down on the spacers. This setup is now called a “sandwich culture.” Time intervals for subsequent experiments described in this study represent the actual time after initiation of a sandwich culture. Immunocytochemical staining. For immunostaining, the top slide and culture medium of sandwich cultures were removed and test antibodies diluted in growth medium were directly added to the unfixed monolayer of cells on the glass slides and incubated for 1 h at 37°C. Reactivity of the antibodies was visualized by avidin-biotin amplified immunoperoxidase staining (Vector Labs., CA) which consisted of incubation with biotin-conjugated anti-mouse serum for 30 min at 37°C incubation with avidin-biotin-peroxidase conjugate for 30 min, and finally precipitation of the colored product using aminoethylcarbazole. After each incubation, the slides were washed on a shaker with two changes of PBS. The cells were counterstained with hematoxylin, dehydrated, and mounted in Crystal Mount. Undiluted tissue culture supernatant or a 1:lOO dilution of mouse ascites of an antikeratin monoclonal antibody 312C8-1 [7-91 and a pan-intermediate filament monoclonal antibody, TIB 131 (obtained from the American Type Culture Collection), was used in this study for targeting cellular components which are exclusively intracellular. To serve as negative controls, the following types of reagents were used: a nonspecific mouse IgM (same subclass as ab. 312C8-l), or a moab to carcinoembryonic antigen, designated B14B8 [l]. In our hands, this antibody is generally nonreactive with human mammary epithelial cells propagated in MCDB 170 medium. In experiments designed to study the consequences of metabolic poisoning, the sandwich cultures were exposed to 15 mM sodium azide for 30 min at 37”C, prior to immunostaining.
for Developing
Vxygen-
The sandwich is an ingenious concept representing in some ways, a dynamic cross section of a solid tumor. Details on the technical aspects of setup and evaluation, formation and measurement of spatial gradients leading to nutrient depletion, and theoretical analysis of the data supporting the role of oxygen-glucose deprivation in cellular necrosis in the sandwich system have been previously described in detail by Hlatky et al. [4, lo]. We have adopted this culture system in studies designed to investigate the relationship between cellular integrity and macromolecular uptake. For this purpose, a monoclonal antibody, designated 312C8-1, directed to an intracellular target of human mammary epithelial cells in culture, the cytoskeletal keratins, was used [7-91. Specific parameters of the sandwich system were optimized and consistently maintained throughout this study. Most important among these are growth medium composition, cell density, and gap height. The relevance of these parameters in the overall topography of the sandwich is illustrated in Fig. 1. The single layer of cells growing on a glass slide as a sandwich is subjected to self-created gradients of nutrients and metabolic products. Due to these gradients, the sandwiched monolayer reportedly gradually develops an inner region of ne-
SANDWICH
183
ASSAY
FIG. 1. Cross section of a single sandwich in a square culture dish. A, Top slide; B, bottom slide with cell monolayer; x, gap height; y-z, direction of decreasing nutrient supply; z, most depleted gradient zone; p, antibody-permeable zone; and imp, antibody-impermeable zone.
erotic cells surrounded by a border of viable cells, corresponding also to the necrotic center and viable rim, respectively, seen in cross sections of the in vitro multicellular spheroid model of solid tumors [ 111. In sandwiches of the specific cell type used in the present study, the culture conditions, and the length of time required to observe macromolecular uptake, did not affect gross cellular morphology or nuclear detail, particularly to the extent that they could be termed “necrotic.” In fact, the cells appeared healthy and viable. We therefore refer here to the inner versus outer region as “permeable” or “impermeable,” respectively. Antikeratin
RESULTS
Use of the Sandwich System Glucose Gradients in Vitro
USING
Uptake
versus Sandwich
Age
In the presence of the constant parameters of cell type, cell density, growth medium, and gap height, the uptake and binding of the antikeratin antibody 312C8-1 was measured in live, unfixed cultures of varying sandwich age. The kinetics of accumulation of antibody-permeable cells in human mammary epithelial sandwiches are shown in Fig. 2. Similar kinetic distribution was observed by measuring two different parameters in sandwiches of varying ages, namely, the percentage immunostained cells in the permeable region and the relative width of the permeable region, thus suggesting a nutrient dose-dependent association of this phenotype. As demonstrated by the data, the number of immunostained cells in the sandwich increases remarkably to 72 h, when the maximal number of immunostained cells is attained. Time points prior to 24 h did not show an appreciable number of immunostained cells in the sandwich. As illustrated in Fig. 3, the immunostained cells were localized only within the permeable region, whereas the edges of the sandwich which constitute the impermeable region were uniformly unstained. The width of the permeable region was found to increase up to 72 h, concurrently with the increasing number of immunostained cells (Fig. 2). The patterns of immunoreactivity within the permeable cells were quite interesting. In sandwiches of up to 55-h duration, at the junction of the permeable and impermeable regions, antikeratin lo-
184
DAIRKEE
0
24
48
72
96
120
144
ET
AL.
results were further confirmed in parallel experiments by a pan-intermediate filament monoclonal antibody, cross-reactive with a variety of cytoskeletal elements [ 121. Sandwiches of various ages set up to serve as negative controls received nonspecific mouse IgM instead of antikeratin. These cultures displayed no immunostaining and have thus led us to the conclusion that although in principle any antibody can enter a proportion of nutritionally deprived cells in the sandwich, yet it is not retained within the cell through the various steps of a standard immunolocalization procedure if it is not targeted at relatively immobilized cellular antigens. Data on sister sandwiches of varying ages, shown in Fig. 3 when stained with trypan blue, closely resembled the immunostained sandwiches in the distribution of stained cells, up to 48 h. However, a much larger area of the sandwich was trypan blue-positive at later stages as compared to immunostaining (Fig. 3). Effects of Metabolic Poisoning, Starvation, or Postgradient Recovery on Antikeratin Passage into Cells
0
24
48
72
96
120
144
Time (hours) FIG. 2. Accumulation of antibody-permeable human mammary epithelial cells as a function of time in sandwich culture. (1) For determining percentage immunostained cells in the antibody-permeable central zone, the number of immunostained versus nonstained cells was counted under 125~ magnification along the center of the 25.mm-wide and 55mm-long monolayer at l-cm intervals. (2) Relative width of antibody-permeable center = (W ~ d/W x 100 where W is the total width of monolayer (in this case, 25 mm), and d is the sum of distance to first immunostained cell seen from either edge.
calization was heterogeneous within the cells and appeared as small, focal areas of immunostaining (Fig. 3). In sister sandwiches within the same culture dish, where the cells were fixed and permeahilized prior to immunolocalization, all cells were uniformly reactive over the entire cytoplasm, as expected. This observation of heterogeneous intracellular reactivity in cells situated in areas proximal to the inner region of the sandwich is suggestive of antibody passage into the cytoplasm via focal gaps in the surface membrane of cells. In the innermost regions, antikeratin staining was homogeneously distributed over the entire cell, suggesting more uniformity in antibody passage throughout the cell surface. These
To address the possibility that antikeratin may be internalized, by human mammary epithelial cells in sandwich cultures, by active endocytosis, cells in sandwiches of various ages were exposed for 30 mins at 37°C to 15 mM sodium azide, which is sufficient to deplete cellular ATP by 60% in 2-3 min [ 131. The cells were then washed and incubated at 37°C for 1 h in antikeratin antibody diluted in growth medium. Internalization and cytoplasmic binding was visualized by routine immunostaining procedures used throughout this study. In sandwiches of greater than 48-h duration, sodium azide treatment had no obvious effect on the number or intensity of immunostained cells. However, prior to 48 h, exposure to sodium azide resulted in a significantly greater number of immunostained cells in the sandwich (Fig. 3). These observations suggest: (1) that antibody molecules enter nutritionally deprived cells by passive diffusion, most probably through regions of the cell surface which have suffered alterations in membrane integrity, and under these conditions, not by active endocytosis; and (2) that those cells which may be only mildly affected by nutrient deprivation and are in the initial stages of cellular damage, when exposed to a metabolic poison, such as sodium azide, rapidly deteriorate to the stages displaying extensive loss of cellular integrity, once again allowing passive diffusion of antibody molecules. In experiments aimed at evaluating the individual effects of glucose and oxygen, since both are simultaneously depleted in the sandwich assay, cell monolayers on glass slides were subjected to glucose “starvation.” This was achieved simply by setting up a series of replicate cultures for routine tissue culture, in which growth
NUTRIENT
DEFICIENCIES
STUDIED
USING
SANDWICH
ASSAY
185
FIG. 3. Immunoperoxidase localization of monoclonal antikeratin, 31X%1, in unfixed sandwich cultures of human mammary epithelial cells. All cultures, except where indicated, were counterstained with hematoxylin. (a) Macroscopic demonstration of the distribution of immunostained cells in a 72-h sandwich. The reddish-brown area represents the most nutritionally depleted, central zone of antibody-permeable cells. Unstained regions on either side predominantly consist of antibody-impermeable cells. (b-e) Photomicrographs of immunostained cells in various regions of the sandwich; (b) unstained cells in the outer, impermeable region; (c) heterogeneously stained cells in between the outer, impermeable and the central zone of permeable cells; (d) heterogeneously stained cells enlarged to show intracellular immunolocalization; (e) homogeneously stained central zone of permeable cells showing strong intracellular reactivity with the antikeratin antibody; (f) sister sandwich of the one shown in (b-e), stained with trypan blue only. No counterstain was applied to this slide. Note uniform blue staining of all cells. Nuclear staining appears more intense than the cytoplasmic staining. (g) A 24-h sandwich showing few immunoreactive cells in the central zone as compared to the 72-h sandwich shown in (e). (h) Sister culture of that shown in (g), which was treated with 15 mA4 sodium azide. Note significant increase in number of immunostained cells: (i) immunostaining of a 48-h sandwich with 312C8-1 aft,er 48 h of recovery under nongradient conditions. Note the persistence of a significant number of permeable cells. Magnification for b, c, e-i is 160~ and that for d is 400x. medium tures
was were
not
prepared
replenished for
for antikeratin
14 days.
Unfixed
immunostaining
culat
several time points during this period. Sister cultures fixed with acetone were used as controls. In contrast to the sandwich, under these conditions of nutrient depri-
vation, we did not observe any antikeratin-positive cells among the attached population of cells at any time point in the experiment. Sporadic immunostained cells seen were invariably rounded and most probably represented dead cells which expectedly would allow passive macro-
DAIRKEE
molecular diffusion. These observations thus support the conclusion that combined glucose-oxygen depletion dramatically alters membrane integrity of epithelial cells, which is not observed at least during 14 days of glucose starvation alone. However, our data do not exclude the possibility that other known and/or unknown agents also affect membrane integrity under these experimental conditions. To study the reversibility of the effects of nutrient depletion, sandwiches exposed to a 48-h nutrient gradient were allowed to recover for 8, 24, or 48 h under conventional nongradient culture conditions. Unfixed monolayers after these periods of recovery, when processed for antikeratin immunostaining, displayed the same degree and distribution of antikeratin-positive cells as the control sandwiches which were not allowed a recovery period. A 48-h sandwich after 48 hr of postgradient recovery is illustrated in Fig. 3. The simplest interpretation of these results is that cellular damage resulting from these environmental conditions is not rapidly repaired. Moreover, since we have not observed increased or renewed cell proliferation in this duration in either the impermeable or the permeable regions, it appears that a 48-h recovery period may not be sufficient to evoke a visible response in terms of cell division or cell repair as measured by this assay. Thus, the severity of this damage and the consequential effects on cellular metabolism, survival, and proliferation remain to be established. DISCUSSION
In this study of human mammary epithelial cells, we have employed the sandwich assay of Hlatky et al. [4, lo] to mimic physiologic conditions of nutrient availability or lack thereof, which often develop in viva in a variety of diseased or malignant states, including breast carcinoma. Using this system, it has been demonstrated that nutrient depletion initially affects cells at a submorphologic level. As illustrated here, these effects can be readily visualized microscopically by immunolocalization of antibodies to intracellular antigens which passively permeate into unfixed cells under these culture conditions. The simplicity and facility of the combined approach of growing cells in a sandwich and subsequently immunoprobing cellular integrity make it possible to investigate a variety of biologically important issues, for example, (a) the definition of specific compounds or conditions which can enhance or expedite passive macromolecular uptake, or those which counteract environmental conditions within a nutrient gradient such that cellular integrity is not compromised and passive macromolecular uptake is prevented, (b) the differences in the response of various histologic cell types, such as epithelial versus mesenchymal; and (c) the determination of differences in the response of normal
ET AL.
versus malignant cells. Properties of tumor-derived cells grown in other types of nutrient gradient models, such as spheroids, have been previously reported (for review, see [14]). However, unlike the sandwich, most anchorage-dependent cell types cannot be examined as spheroids. Furthermore, most of the widely used breast carcinoma cultures represent established cell lines derived from relatively late states in malignant progression, such as metastatic effusions [15]. These cells display accepted parameters of malignant transformation, such as immortality, reduced nutritional requirements, anchorage-independent growth, and tumor formation in immunosuppressed animals. However, human mammary epithelial cells cultured from primary breast carcinoma generally do not display these features. Instead, they closely resemble normal mammary epithelium derived from nonmalignant tissue [161. The latter two cell types also exhibit more stringent nutrient requirements and are consequently more sensitive to nutrient deprivation as compared to established cell lines. In this perspective, therefore, one would expect the kinetics of antibody permeability in normal human mammary epitheha1 cell sandwiches to differ considerably from the kinetics of sandwiches of established breast carcinoma cell lines but to closely resemble those of short-term cultures of primary breast carcinoma. The results of this study are also of direct relevance to consequences of microecological diversity of malignant cells in solid tumors, such as those caused by tumor hypoxia. There is extensive evidence documenting the existence of a significant subset of hypoxic cells within solid tumors. A recent study by Kallinowski et al. has demonstrated that xenografts from primary breast carcinoma were much more slow growing than lung cancer xenografts. It was found upon further investigation that eight of nine breast cancers were poorly perfused and exhibited low metabolic rates with evident tumor hypoxia and tissue acidosis [ 171. Generally, hypoxic cells lie between an inner necrotic zone and the peripheral wellnourished regions of the tumor [18-201. Cells in these nutrient-deficient regions are known to be resistant to radiation and conventional chemotherapy [21-231. Furthermore, experimental evidence suggests that the hypoxic state may be causally involved in over replicationrecombination events leading to spontaneous drug resistance, oncogene amplification, aneuploidy, and a variety of chromosomal rearrangements resulting in increased cellular heterogeneity and malignant progression [24]. Raleigh et al. [25] have reported an immunohistochemical assay for the estimation of hypoxic cells in solid tumors based on the detection of misonidazole adducts known to occur preferentially in hypoxic cells. Other related techniques involve the use of radiolabeled or brominated misonidazole [26, 271 or fluorescent nitroheterocyclics and fluorescent dyes [28, 291. Such assays done on tumor biopsies could be a very valuable
NUTRIENT
DEFICIENCIES
STUDIED
USING SANDWICH
187
ASSAY
adjunctin determiningthe degreeof tumor hypoxia and exclusionproceduresasshownin Fig. 3,and asreported reoxygenation during therapy, Designed to meet a simi- by others, arenot reliableindicators of cell viability and lar objective, the assay for assessingpassive cellular in- metabolicactivity underconditionswhich result in gaps fusion of antibodies, proposed by us [ 11,could also serve
in the plasma membrane [40, 411. However, such cells
to identify tumor cells affected by hypoxia. In fact, since the assay is based primarily on the identification of cells with compromised membrane integrity, it could serve to estimate cells affected in this manner by other genetic, metabolic, or environmental conditions as well. It is generally believed that in the elimination of hypoxic tumor cells, lack of efficacy of conventional therapeutic strategies is mainly due to the fact that hypoxic cells are not actively proliferating [30] or are slowed in their progression through the cell cycle [31]. Therefore, a therapeutic modality which is independent of these factors, such as the use of monoclonal antibodies to tumor-associated antigens, as toxic immunoconjugates, has the potential to overcome these limitations. As demonstrated in this report and in our earlier study [ 11, antibodies to intracellular cytoskeletal proteins such as keratins, offer several possible advantages for targeting carcinoma cells with compromised membrane integrity, for example, the availability of large quantities of target antigen present per cell as compared to tumor-associated surface antigens, which are often shed from the cell surface. Previous reports have demonstrated successful localization of tumors in uiuo with predominantly intracellular, tumor-associated antigens such as a-fetoprotein [32] and c-myc oncogene product [33]; secretion or extracellular accumulation of these proteins has not been ruled out in these studies. Furthermore, localization of xenografts using antibodies to exclusively intracellular keratin-like antigens [34] and nuclear antigens [35] has also been achieved, as predicted by results obtained by the passive cellular infusion of antibody into surgical specimens of malignant, tissue [l]. In the present study, we have provided direct evidence for one of possibly several underlying mechanisms, namely oxygen-glucose depletion, which gives rise to a population of tumor cells that can be potentially targeted and eliminated by immunoconjugates directed at intracellular antigens. As succinctly stated in a recent review [36], “given the existence of hypoxia in tumors, it is sensible to develop strategies to take advantage of its presence.” Passive uptake and retention of monoclonal antibodies due to reduced membrane integrity of cells under conditions of nutrient depletion demonstrate that these conditions could also be utilized for loading exogenous macromolecules into the cell cytoplasm. A number of methods exist for introducing antibodies into cells. These include microinjection [37], scrape loading [38], and osmotic permeabilization [39]. A major objective of introducing antibodies into cells is the functional analysis of the target antigen. In this regard maintenance of cell viability is an important factor. Conventional dye
reportedly are capable of clonal proliferation gesting that the transient loss of membrane often readily repaired by the cell.
[39], sugintegrity is
This research was supported by the Office of Health and Environmental Research (OHER), U.S. Department of Energy, under Contract DE-AC03-76SF00098, and by Research Grant R813784 from the Environmental Protection Agency (S. H. Dairkee). We thank Dr. Lynn Hlatky, Harvard Medical School, for advice and assistance in setting up sandwich cultures.
REFERENCES 1.
Dairkee, S. H., and Hackett, 1216-1220.
2.
Khaw, B. A., Beller, G. A., and Haber, E. (1978) Circulation 57, 743-750. Khaw, B. A., Fallon, J. T.. Belier, G. A., and Haber, E. (1979) Circulation 60, 1527-1531.
I3 4.
Hlatky, 611.
L., and Alpen,
A. J. (1988)
E. L. (1985)
J. N&l.
Cell Tissue
Cancer
Kinet.
Inst. 80,
18,
597-
5.
Stampfer, M. R., Hallowes, R. C., and Hackett, A. J. (1980) In Vitro 16, 415-425. 6. Hammond, S. L., Ham, R. G., and Stampfer, M. R. (1984) Proc. ! Natl. Acad. Sci. USA 81,5435-5439. I 7.1 Dairkee, S. H., Blayney, C. M., Smith, H. S., and Hackett, A. J. I (1985) Proc. Natl. Acad. Sci. USA 82, 7409-7412. S. H., Blayney-Moore, C. M., Smith, H. S., and Hack8.1 Dairkee, ett, A. J. (1986) Differentiation 32, 93-100. / S. H., Ljung, B. M., Smith, H. S., and Hackett, A. J. 9 ./ Dairkee, (1987) Breast Cancer Res. 7’reat. 10, 11-20. 10.1 Hlatky, L., Sachs, I 134,167-178.
R. K., and Alpen,
11.1 Sutherland, R. M., and Cancer Hes. 95, 24-49.
Durand,
E. L. (1988) R. E. (1984)
J. Cell. Physiol. Recent
Results
12./
Pruss, R. M., Mirsky, R., Raff, M., Thorpe, R., Dowding, A. J., / and Anderton, B. H. (1981) Cell 27,419-428. / 13.; Clarke, B. L., and Weigel, P. H. (1985) J. Riol. Chem. 260,128133. 14.1 Sutherland, B. M. (1988) Science 240, 177-184. 15.1 Smith, H. S., Wolman, S. R., Dairkee, S. H., Hancock, M. C., Lippman, M., Leff, A., and Hackett, A. J. (1987) J. Natl. Cant. Inst. 78,611-615. 16.,
17. 18. 19. 20. 21. 22.
Smith, H. S., Dairkee, S. H., Ljung, B. M., Mayall, B., Sylvester, S. S., and Hackett, A. J. (1987) in Cellular and Molecular Biology of Experimental Mammary Cancer (Medina, D., Kidwell, W., Heppner, G., and Anderson, E., Eds.), pp. 437-452, Plenum Publishing. Kallinowski, F., Schlenger, K. H., Kloes, M., Stohrer, M., and Vaupel, P. (1989) Znt. J. Cancer 44, 266-272. Thomlinson, R. H., and Gray, L. H. (1955) Bit. J. Cancer 9, 539-549. Tannock, I. F. (1968) Brit. J. Cancer 22, 258-273. Goldacre, R. J., and Slyven, B. (1962) B-it. J. Carzcer 16, 306322. Hill, R. P., and Stanley, J. A. (1975) Cancer Sutherland, R. M., Eddy, H. A., Bareham,
Res. 35,1147-1153. B., Reich, K., and
23.
Vanantwerp, D. (1979) Int. J. Rod. On&. Hiol. 1230. Tannock, I. (1982) Cancer Res. 42, 4921-4926.
Phys.
Proc.
DAIRKEE
ET Al,.
5, 12255
42.
24.
Rice, G. C., Hoy, C., and Schimke, Sci. USA 83, 5978-5982.
R. ‘I’. (1986)
Natl.
Acad.
25.
Raleigh, J. A., Miller, G. G., Franko, A. F., and Kelly, D. A. (1987) Hrit.
A. d., Koch, C. J., Fuciarelli, J. Cancer 56, 395-400.
26. 21.
Ho, M. K., Geist, C., Murray, ,J., Kato, K., Delellis, R., Wolfe, H., Carney, W., Rabin, H., and Durda, 0. (1988) Cancer Res. 48, 4969-4975.
Urtasun, R. C., Chapman, J. D., Raleigh, .J. A., Franko, A. J., and Koch, C. J. (1986) Int. J. Rod. Oncol. Biol. Phys. 12, 12631267.
35.
Epstein, A. L., Chen, 48, 5842-5848.
Rasey, Radiut.
36.
Coleman,
37.
Miller, M. R., Ulrich, R. G., Shu-Fong, W. T., and Korn, D. (1985) J. Biol. Chem. 260, 134-138. Freeman, J. W., Hazlewood, J. E., Auerbach, P., and Busch, H. (1988) Cancer Res. 48, 5246-5250. Okada, C. Y., and Rechsteiner, M. (1982) Cell 29, 33-41. Beckers, C. J. M., Keller, D. S., and Balch, W. E. (1987) Cell 50, 523-534. Charkabarti, R., Pf’eiffer, N. E., Wylie, D. E., and Schuster, S. M. (1989) J. Biol. Chem. 264, 8214-8221.
J. S., Krohn, K., Freauff, Res. 87, 455-459. P. L., and
Durand,
S., and
Res. 43,
(1981)
Begg, A. C., Engelhardt, E. I,., McNally, N. d., Terry, N. H. A., and Wardman, P. (1983) Brit. J. Radial. 56, 97Ob973. Valeriote, F., and Vanputten, L. M. (1975) Cancer Res. 35,26192630.
327638.
Koch, C. J., Kruuv, J. Radiat. Riol. 23,
April version
Cancer
S. M.
29.
Received Revised
R. E. (1983)
Larson,
Olive, 3280.
31.
K., Fairweather, D. S., Bradwell, A. R., Blackburn, P. W., Howell, A., Reeder, A., and Hine, K. R. Med. J. 283, 942-944. T., Evan, G. I., Ritson, A., Watson, J., Wraight, P., K. (1984) Bit. J. Cancer 54, 761-769.
34.
28.
30.
:33.
Halsall, A. J. C., Dykes, (1981) Bit. Chan, S. Y. and Sikora,
20, 1990. received
J., Frey, 67-74.
July
N. E., and Sayder,
18, 1990.
R. A. (1973)
Int.
39. 40. 41.
C. N. (1988)
F. M., and Taylor, J. N&l.
Cancer.
C. R. (1988) Inst.
80,
Cancer
Res.
310-317.
CELL
RESEARCH
1%,182-188
(1991)
Accessibility to Intracellular Antigens within Nutritionally Deprived Human Mammary Epithelial Cells SHAHNAZ
H. DAIRKEE,’ Per&a
LYNN PUETT, ANNA M. COUNELIS, Cancer
Research
Institute,
Inc.
INTRODUCTION We have recently demonstrated selective localization of a monoclonal antibody to intracellular, cytoskeletal keratin polypeptides in malignant breast epithelium under conditions of passive infusion of antibody in fresh surgical specimens of breast carcinoma [l]. These observations are closely analogous to those reported for the targeting of experimental myocardial infarcts using antibodies to another intracellular component, myosin [21, and may have important implications for tumor targeting. The uptake of antibody in the infarcts was reportedly greatest in regions of most severe blood flow reduction, and a subsequent histologic study correlated antimyosin binding with severe cellular necrosis [3], 1 To whom reprint keley Laboratory, 934.47A, University
requests should be addressed at Lawrence Division of Cell and Molecular Biology, of California, Berkeley, CA 94720.
Copyright All
rights
0
1991 by of reproduction
BerMS
182
0014.4827/91$3.00
Academic in any
Press, Inc. form
reserved.
California
94609
suggesting that loss of cell membrane integrity results in accessibility of macromolecules to intracellular sites. Our findings on antibody-infused solid tumor tissue have unexpectedly demonstrated antibody localization in the cytoplasm of tumor cells displaying visible cellular integrity without apparent signs of necrosis. One possible interpretation of this observation is that loss of membrane integrity is a progressive phenomenon and due to sustained metabolic, physiologic, and/or genetic limitations on the cell it eventually terminates in frank cellular necrosis and disintegration. The earliest signs of membrane damage thus may not be identifiable as specific morphologic changes but may possibly be measured in functional terms, simply by uncontrolled cellular uptake of macromolecules. In this regard, a two-dimensional, multicellular, in vitro model of a self-created gradient of nutrients and metabolic products, developed and described by Hlatky and Alpen [4] as the “sandwich system,” provides the unique opportunity for visualization and manipulation of cells at various stages in the entire sequence of events leading to frank necrosis. We have used here the sandwich system in conjunction with antibodies directed at intracellular antigens to serve as “indicator molecules” for determining the spectrum of cells in a diffusion gradient which can be potentially targeted by these reagents.
We have previously demonstrated immunolocalization of antikeratin antibodies in apparently random subpopulations of malignant cells in fresh surgical specimens of breast carcinoma (S. H. Dairkee and A. J. Hackett, 1988, J. Natl. Cancer Inst. 80, 1216-1220). The goal of the present study was to determine whether deficiencies in essential nutrients contribute toward cellular alterations in membrane integrity, consequently allowing antikeratin to bind to the cytoskeleton within live, unfixed cells. We have demonstrated here that in an in vitro model in which human mammary epithelial cells are subjected to an oxygen-glucose gradient, immunolocalization of antikeratin within the cells is observed in a dose-dependent manner in the depleted regions of the gradient, even though the cells appear to be morphologically unaltered. The potential use of antibodies to intracellular antigens for immunotargeting solid tumors and the use of this method in antibody-loading studies toward understanding functional aspects of specific cellular antigens, as well as determining differential response of various cell types under these culture conditions, are discussed. Q 1991 Academic Press.
Oakland,
AND ADELINE J. HACKETT
MATERIALS
AND METHODS
Cell culture. Normal human mammary epithelial cultures used in this study were initiated from reduction mammoplasty specimens. The methodology for culturing human mammary epithelial cells has been previously reported in detail [5]. The cells were propagated in the serum-free medium, MCDB 170, prepared as previously described [6]. Glucose concentration in fresh medium is 1.4 g/liter. Sandwich system. This in vitro assay was set up in our laboratory essentially as previously described (41 to serve as a model for studying the effects of a microenvironmental gradient on the membrane integrity of human mammary epithelial cells. Briefly, cells are grown as monolayers “sandwiched” between two glass microscope slides separated by a distance of 150 pm. A thin layer of growth medium covers the cells and fills the narrow gap between the slides. The sandwiching of cells in this manner causes all nutrients and waste products to move into or out of the local environment of the cells by diffusing through the narrow gap between the slides. Clean, sterilized microscope slides are first placed in 3.5 X 3.5-in. integrid petri dishes (Falcon). Cells are seeded at 2 X lo6 cells/dish in 12 ml of growth medium
NUTRIENT
DEFICIENCIES
STUDIED
and incubated at 37°C in 0.05% CO,. After 48-h of incubation, culture medium is renewed, 150.pm sterile glass spacers are placed at each end of the slide, and a top slide is laid down on the spacers. This setup is now called a “sandwich culture.” Time intervals for subsequent experiments described in this study represent the actual time after initiation of a sandwich culture. Immunocytochemical staining. For immunostaining, the top slide and culture medium of sandwich cultures were removed and test antibodies diluted in growth medium were directly added to the unfixed monolayer of cells on the glass slides and incubated for 1 h at 37°C. Reactivity of the antibodies was visualized by avidin-biotin amplified immunoperoxidase staining (Vector Labs., CA) which consisted of incubation with biotin-conjugated anti-mouse serum for 30 min at 37°C incubation with avidin-biotin-peroxidase conjugate for 30 min, and finally precipitation of the colored product using aminoethylcarbazole. After each incubation, the slides were washed on a shaker with two changes of PBS. The cells were counterstained with hematoxylin, dehydrated, and mounted in Crystal Mount. Undiluted tissue culture supernatant or a 1:lOO dilution of mouse ascites of an antikeratin monoclonal antibody 312C8-1 [7-91 and a pan-intermediate filament monoclonal antibody, TIB 131 (obtained from the American Type Culture Collection), was used in this study for targeting cellular components which are exclusively intracellular. To serve as negative controls, the following types of reagents were used: a nonspecific mouse IgM (same subclass as ab. 312C8-l), or a moab to carcinoembryonic antigen, designated B14B8 [l]. In our hands, this antibody is generally nonreactive with human mammary epithelial cells propagated in MCDB 170 medium. In experiments designed to study the consequences of metabolic poisoning, the sandwich cultures were exposed to 15 mM sodium azide for 30 min at 37”C, prior to immunostaining.
for Developing
Vxygen-
The sandwich is an ingenious concept representing in some ways, a dynamic cross section of a solid tumor. Details on the technical aspects of setup and evaluation, formation and measurement of spatial gradients leading to nutrient depletion, and theoretical analysis of the data supporting the role of oxygen-glucose deprivation in cellular necrosis in the sandwich system have been previously described in detail by Hlatky et al. [4, lo]. We have adopted this culture system in studies designed to investigate the relationship between cellular integrity and macromolecular uptake. For this purpose, a monoclonal antibody, designated 312C8-1, directed to an intracellular target of human mammary epithelial cells in culture, the cytoskeletal keratins, was used [7-91. Specific parameters of the sandwich system were optimized and consistently maintained throughout this study. Most important among these are growth medium composition, cell density, and gap height. The relevance of these parameters in the overall topography of the sandwich is illustrated in Fig. 1. The single layer of cells growing on a glass slide as a sandwich is subjected to self-created gradients of nutrients and metabolic products. Due to these gradients, the sandwiched monolayer reportedly gradually develops an inner region of ne-
SANDWICH
183
ASSAY
FIG. 1. Cross section of a single sandwich in a square culture dish. A, Top slide; B, bottom slide with cell monolayer; x, gap height; y-z, direction of decreasing nutrient supply; z, most depleted gradient zone; p, antibody-permeable zone; and imp, antibody-impermeable zone.
erotic cells surrounded by a border of viable cells, corresponding also to the necrotic center and viable rim, respectively, seen in cross sections of the in vitro multicellular spheroid model of solid tumors [ 111. In sandwiches of the specific cell type used in the present study, the culture conditions, and the length of time required to observe macromolecular uptake, did not affect gross cellular morphology or nuclear detail, particularly to the extent that they could be termed “necrotic.” In fact, the cells appeared healthy and viable. We therefore refer here to the inner versus outer region as “permeable” or “impermeable,” respectively. Antikeratin
RESULTS
Use of the Sandwich System Glucose Gradients in Vitro
USING
Uptake
versus Sandwich
Age
In the presence of the constant parameters of cell type, cell density, growth medium, and gap height, the uptake and binding of the antikeratin antibody 312C8-1 was measured in live, unfixed cultures of varying sandwich age. The kinetics of accumulation of antibody-permeable cells in human mammary epithelial sandwiches are shown in Fig. 2. Similar kinetic distribution was observed by measuring two different parameters in sandwiches of varying ages, namely, the percentage immunostained cells in the permeable region and the relative width of the permeable region, thus suggesting a nutrient dose-dependent association of this phenotype. As demonstrated by the data, the number of immunostained cells in the sandwich increases remarkably to 72 h, when the maximal number of immunostained cells is attained. Time points prior to 24 h did not show an appreciable number of immunostained cells in the sandwich. As illustrated in Fig. 3, the immunostained cells were localized only within the permeable region, whereas the edges of the sandwich which constitute the impermeable region were uniformly unstained. The width of the permeable region was found to increase up to 72 h, concurrently with the increasing number of immunostained cells (Fig. 2). The patterns of immunoreactivity within the permeable cells were quite interesting. In sandwiches of up to 55-h duration, at the junction of the permeable and impermeable regions, antikeratin lo-
184
DAIRKEE
0
24
48
72
96
120
144
ET
AL.
results were further confirmed in parallel experiments by a pan-intermediate filament monoclonal antibody, cross-reactive with a variety of cytoskeletal elements [ 121. Sandwiches of various ages set up to serve as negative controls received nonspecific mouse IgM instead of antikeratin. These cultures displayed no immunostaining and have thus led us to the conclusion that although in principle any antibody can enter a proportion of nutritionally deprived cells in the sandwich, yet it is not retained within the cell through the various steps of a standard immunolocalization procedure if it is not targeted at relatively immobilized cellular antigens. Data on sister sandwiches of varying ages, shown in Fig. 3 when stained with trypan blue, closely resembled the immunostained sandwiches in the distribution of stained cells, up to 48 h. However, a much larger area of the sandwich was trypan blue-positive at later stages as compared to immunostaining (Fig. 3). Effects of Metabolic Poisoning, Starvation, or Postgradient Recovery on Antikeratin Passage into Cells
0
24
48
72
96
120
144
Time (hours) FIG. 2. Accumulation of antibody-permeable human mammary epithelial cells as a function of time in sandwich culture. (1) For determining percentage immunostained cells in the antibody-permeable central zone, the number of immunostained versus nonstained cells was counted under 125~ magnification along the center of the 25.mm-wide and 55mm-long monolayer at l-cm intervals. (2) Relative width of antibody-permeable center = (W ~ d/W x 100 where W is the total width of monolayer (in this case, 25 mm), and d is the sum of distance to first immunostained cell seen from either edge.
calization was heterogeneous within the cells and appeared as small, focal areas of immunostaining (Fig. 3). In sister sandwiches within the same culture dish, where the cells were fixed and permeahilized prior to immunolocalization, all cells were uniformly reactive over the entire cytoplasm, as expected. This observation of heterogeneous intracellular reactivity in cells situated in areas proximal to the inner region of the sandwich is suggestive of antibody passage into the cytoplasm via focal gaps in the surface membrane of cells. In the innermost regions, antikeratin staining was homogeneously distributed over the entire cell, suggesting more uniformity in antibody passage throughout the cell surface. These
To address the possibility that antikeratin may be internalized, by human mammary epithelial cells in sandwich cultures, by active endocytosis, cells in sandwiches of various ages were exposed for 30 mins at 37°C to 15 mM sodium azide, which is sufficient to deplete cellular ATP by 60% in 2-3 min [ 131. The cells were then washed and incubated at 37°C for 1 h in antikeratin antibody diluted in growth medium. Internalization and cytoplasmic binding was visualized by routine immunostaining procedures used throughout this study. In sandwiches of greater than 48-h duration, sodium azide treatment had no obvious effect on the number or intensity of immunostained cells. However, prior to 48 h, exposure to sodium azide resulted in a significantly greater number of immunostained cells in the sandwich (Fig. 3). These observations suggest: (1) that antibody molecules enter nutritionally deprived cells by passive diffusion, most probably through regions of the cell surface which have suffered alterations in membrane integrity, and under these conditions, not by active endocytosis; and (2) that those cells which may be only mildly affected by nutrient deprivation and are in the initial stages of cellular damage, when exposed to a metabolic poison, such as sodium azide, rapidly deteriorate to the stages displaying extensive loss of cellular integrity, once again allowing passive diffusion of antibody molecules. In experiments aimed at evaluating the individual effects of glucose and oxygen, since both are simultaneously depleted in the sandwich assay, cell monolayers on glass slides were subjected to glucose “starvation.” This was achieved simply by setting up a series of replicate cultures for routine tissue culture, in which growth
NUTRIENT
DEFICIENCIES
STUDIED
USING
SANDWICH
ASSAY
185
FIG. 3. Immunoperoxidase localization of monoclonal antikeratin, 31X%1, in unfixed sandwich cultures of human mammary epithelial cells. All cultures, except where indicated, were counterstained with hematoxylin. (a) Macroscopic demonstration of the distribution of immunostained cells in a 72-h sandwich. The reddish-brown area represents the most nutritionally depleted, central zone of antibody-permeable cells. Unstained regions on either side predominantly consist of antibody-impermeable cells. (b-e) Photomicrographs of immunostained cells in various regions of the sandwich; (b) unstained cells in the outer, impermeable region; (c) heterogeneously stained cells in between the outer, impermeable and the central zone of permeable cells; (d) heterogeneously stained cells enlarged to show intracellular immunolocalization; (e) homogeneously stained central zone of permeable cells showing strong intracellular reactivity with the antikeratin antibody; (f) sister sandwich of the one shown in (b-e), stained with trypan blue only. No counterstain was applied to this slide. Note uniform blue staining of all cells. Nuclear staining appears more intense than the cytoplasmic staining. (g) A 24-h sandwich showing few immunoreactive cells in the central zone as compared to the 72-h sandwich shown in (e). (h) Sister culture of that shown in (g), which was treated with 15 mA4 sodium azide. Note significant increase in number of immunostained cells: (i) immunostaining of a 48-h sandwich with 312C8-1 aft,er 48 h of recovery under nongradient conditions. Note the persistence of a significant number of permeable cells. Magnification for b, c, e-i is 160~ and that for d is 400x. medium tures
was were
not
prepared
replenished for
for antikeratin
14 days.
Unfixed
immunostaining
culat
several time points during this period. Sister cultures fixed with acetone were used as controls. In contrast to the sandwich, under these conditions of nutrient depri-
vation, we did not observe any antikeratin-positive cells among the attached population of cells at any time point in the experiment. Sporadic immunostained cells seen were invariably rounded and most probably represented dead cells which expectedly would allow passive macro-
DAIRKEE
molecular diffusion. These observations thus support the conclusion that combined glucose-oxygen depletion dramatically alters membrane integrity of epithelial cells, which is not observed at least during 14 days of glucose starvation alone. However, our data do not exclude the possibility that other known and/or unknown agents also affect membrane integrity under these experimental conditions. To study the reversibility of the effects of nutrient depletion, sandwiches exposed to a 48-h nutrient gradient were allowed to recover for 8, 24, or 48 h under conventional nongradient culture conditions. Unfixed monolayers after these periods of recovery, when processed for antikeratin immunostaining, displayed the same degree and distribution of antikeratin-positive cells as the control sandwiches which were not allowed a recovery period. A 48-h sandwich after 48 hr of postgradient recovery is illustrated in Fig. 3. The simplest interpretation of these results is that cellular damage resulting from these environmental conditions is not rapidly repaired. Moreover, since we have not observed increased or renewed cell proliferation in this duration in either the impermeable or the permeable regions, it appears that a 48-h recovery period may not be sufficient to evoke a visible response in terms of cell division or cell repair as measured by this assay. Thus, the severity of this damage and the consequential effects on cellular metabolism, survival, and proliferation remain to be established. DISCUSSION
In this study of human mammary epithelial cells, we have employed the sandwich assay of Hlatky et al. [4, lo] to mimic physiologic conditions of nutrient availability or lack thereof, which often develop in viva in a variety of diseased or malignant states, including breast carcinoma. Using this system, it has been demonstrated that nutrient depletion initially affects cells at a submorphologic level. As illustrated here, these effects can be readily visualized microscopically by immunolocalization of antibodies to intracellular antigens which passively permeate into unfixed cells under these culture conditions. The simplicity and facility of the combined approach of growing cells in a sandwich and subsequently immunoprobing cellular integrity make it possible to investigate a variety of biologically important issues, for example, (a) the definition of specific compounds or conditions which can enhance or expedite passive macromolecular uptake, or those which counteract environmental conditions within a nutrient gradient such that cellular integrity is not compromised and passive macromolecular uptake is prevented, (b) the differences in the response of various histologic cell types, such as epithelial versus mesenchymal; and (c) the determination of differences in the response of normal
ET AL.
versus malignant cells. Properties of tumor-derived cells grown in other types of nutrient gradient models, such as spheroids, have been previously reported (for review, see [14]). However, unlike the sandwich, most anchorage-dependent cell types cannot be examined as spheroids. Furthermore, most of the widely used breast carcinoma cultures represent established cell lines derived from relatively late states in malignant progression, such as metastatic effusions [15]. These cells display accepted parameters of malignant transformation, such as immortality, reduced nutritional requirements, anchorage-independent growth, and tumor formation in immunosuppressed animals. However, human mammary epithelial cells cultured from primary breast carcinoma generally do not display these features. Instead, they closely resemble normal mammary epithelium derived from nonmalignant tissue [161. The latter two cell types also exhibit more stringent nutrient requirements and are consequently more sensitive to nutrient deprivation as compared to established cell lines. In this perspective, therefore, one would expect the kinetics of antibody permeability in normal human mammary epitheha1 cell sandwiches to differ considerably from the kinetics of sandwiches of established breast carcinoma cell lines but to closely resemble those of short-term cultures of primary breast carcinoma. The results of this study are also of direct relevance to consequences of microecological diversity of malignant cells in solid tumors, such as those caused by tumor hypoxia. There is extensive evidence documenting the existence of a significant subset of hypoxic cells within solid tumors. A recent study by Kallinowski et al. has demonstrated that xenografts from primary breast carcinoma were much more slow growing than lung cancer xenografts. It was found upon further investigation that eight of nine breast cancers were poorly perfused and exhibited low metabolic rates with evident tumor hypoxia and tissue acidosis [ 171. Generally, hypoxic cells lie between an inner necrotic zone and the peripheral wellnourished regions of the tumor [18-201. Cells in these nutrient-deficient regions are known to be resistant to radiation and conventional chemotherapy [21-231. Furthermore, experimental evidence suggests that the hypoxic state may be causally involved in over replicationrecombination events leading to spontaneous drug resistance, oncogene amplification, aneuploidy, and a variety of chromosomal rearrangements resulting in increased cellular heterogeneity and malignant progression [24]. Raleigh et al. [25] have reported an immunohistochemical assay for the estimation of hypoxic cells in solid tumors based on the detection of misonidazole adducts known to occur preferentially in hypoxic cells. Other related techniques involve the use of radiolabeled or brominated misonidazole [26, 271 or fluorescent nitroheterocyclics and fluorescent dyes [28, 291. Such assays done on tumor biopsies could be a very valuable
NUTRIENT
DEFICIENCIES
STUDIED
USING SANDWICH
187
ASSAY
adjunctin determiningthe degreeof tumor hypoxia and exclusionproceduresasshownin Fig. 3,and asreported reoxygenation during therapy, Designed to meet a simi- by others, arenot reliableindicators of cell viability and lar objective, the assay for assessingpassive cellular in- metabolicactivity underconditionswhich result in gaps fusion of antibodies, proposed by us [ 11,could also serve
in the plasma membrane [40, 411. However, such cells
to identify tumor cells affected by hypoxia. In fact, since the assay is based primarily on the identification of cells with compromised membrane integrity, it could serve to estimate cells affected in this manner by other genetic, metabolic, or environmental conditions as well. It is generally believed that in the elimination of hypoxic tumor cells, lack of efficacy of conventional therapeutic strategies is mainly due to the fact that hypoxic cells are not actively proliferating [30] or are slowed in their progression through the cell cycle [31]. Therefore, a therapeutic modality which is independent of these factors, such as the use of monoclonal antibodies to tumor-associated antigens, as toxic immunoconjugates, has the potential to overcome these limitations. As demonstrated in this report and in our earlier study [ 11, antibodies to intracellular cytoskeletal proteins such as keratins, offer several possible advantages for targeting carcinoma cells with compromised membrane integrity, for example, the availability of large quantities of target antigen present per cell as compared to tumor-associated surface antigens, which are often shed from the cell surface. Previous reports have demonstrated successful localization of tumors in uiuo with predominantly intracellular, tumor-associated antigens such as a-fetoprotein [32] and c-myc oncogene product [33]; secretion or extracellular accumulation of these proteins has not been ruled out in these studies. Furthermore, localization of xenografts using antibodies to exclusively intracellular keratin-like antigens [34] and nuclear antigens [35] has also been achieved, as predicted by results obtained by the passive cellular infusion of antibody into surgical specimens of malignant, tissue [l]. In the present study, we have provided direct evidence for one of possibly several underlying mechanisms, namely oxygen-glucose depletion, which gives rise to a population of tumor cells that can be potentially targeted and eliminated by immunoconjugates directed at intracellular antigens. As succinctly stated in a recent review [36], “given the existence of hypoxia in tumors, it is sensible to develop strategies to take advantage of its presence.” Passive uptake and retention of monoclonal antibodies due to reduced membrane integrity of cells under conditions of nutrient depletion demonstrate that these conditions could also be utilized for loading exogenous macromolecules into the cell cytoplasm. A number of methods exist for introducing antibodies into cells. These include microinjection [37], scrape loading [38], and osmotic permeabilization [39]. A major objective of introducing antibodies into cells is the functional analysis of the target antigen. In this regard maintenance of cell viability is an important factor. Conventional dye
reportedly are capable of clonal proliferation gesting that the transient loss of membrane often readily repaired by the cell.
[39], sugintegrity is
This research was supported by the Office of Health and Environmental Research (OHER), U.S. Department of Energy, under Contract DE-AC03-76SF00098, and by Research Grant R813784 from the Environmental Protection Agency (S. H. Dairkee). We thank Dr. Lynn Hlatky, Harvard Medical School, for advice and assistance in setting up sandwich cultures.
REFERENCES 1.
Dairkee, S. H., and Hackett, 1216-1220.
2.
Khaw, B. A., Beller, G. A., and Haber, E. (1978) Circulation 57, 743-750. Khaw, B. A., Fallon, J. T.. Belier, G. A., and Haber, E. (1979) Circulation 60, 1527-1531.
I3 4.
Hlatky, 611.
L., and Alpen,
A. J. (1988)
E. L. (1985)
J. N&l.
Cell Tissue
Cancer
Kinet.
Inst. 80,
18,
597-
5.
Stampfer, M. R., Hallowes, R. C., and Hackett, A. J. (1980) In Vitro 16, 415-425. 6. Hammond, S. L., Ham, R. G., and Stampfer, M. R. (1984) Proc. ! Natl. Acad. Sci. USA 81,5435-5439. I 7.1 Dairkee, S. H., Blayney, C. M., Smith, H. S., and Hackett, A. J. I (1985) Proc. Natl. Acad. Sci. USA 82, 7409-7412. S. H., Blayney-Moore, C. M., Smith, H. S., and Hack8.1 Dairkee, ett, A. J. (1986) Differentiation 32, 93-100. / S. H., Ljung, B. M., Smith, H. S., and Hackett, A. J. 9 ./ Dairkee, (1987) Breast Cancer Res. 7’reat. 10, 11-20. 10.1 Hlatky, L., Sachs, I 134,167-178.
R. K., and Alpen,
11.1 Sutherland, R. M., and Cancer Hes. 95, 24-49.
Durand,
E. L. (1988) R. E. (1984)
J. Cell. Physiol. Recent
Results
12./
Pruss, R. M., Mirsky, R., Raff, M., Thorpe, R., Dowding, A. J., / and Anderton, B. H. (1981) Cell 27,419-428. / 13.; Clarke, B. L., and Weigel, P. H. (1985) J. Riol. Chem. 260,128133. 14.1 Sutherland, B. M. (1988) Science 240, 177-184. 15.1 Smith, H. S., Wolman, S. R., Dairkee, S. H., Hancock, M. C., Lippman, M., Leff, A., and Hackett, A. J. (1987) J. Natl. Cant. Inst. 78,611-615. 16.,
17. 18. 19. 20. 21. 22.
Smith, H. S., Dairkee, S. H., Ljung, B. M., Mayall, B., Sylvester, S. S., and Hackett, A. J. (1987) in Cellular and Molecular Biology of Experimental Mammary Cancer (Medina, D., Kidwell, W., Heppner, G., and Anderson, E., Eds.), pp. 437-452, Plenum Publishing. Kallinowski, F., Schlenger, K. H., Kloes, M., Stohrer, M., and Vaupel, P. (1989) Znt. J. Cancer 44, 266-272. Thomlinson, R. H., and Gray, L. H. (1955) Bit. J. Cancer 9, 539-549. Tannock, I. F. (1968) Brit. J. Cancer 22, 258-273. Goldacre, R. J., and Slyven, B. (1962) B-it. J. Carzcer 16, 306322. Hill, R. P., and Stanley, J. A. (1975) Cancer Sutherland, R. M., Eddy, H. A., Bareham,
Res. 35,1147-1153. B., Reich, K., and
23.
Vanantwerp, D. (1979) Int. J. Rod. On&. Hiol. 1230. Tannock, I. (1982) Cancer Res. 42, 4921-4926.
Phys.
Proc.
DAIRKEE
ET Al,.
5, 12255
42.
24.
Rice, G. C., Hoy, C., and Schimke, Sci. USA 83, 5978-5982.
R. ‘I’. (1986)
Natl.
Acad.
25.
Raleigh, J. A., Miller, G. G., Franko, A. F., and Kelly, D. A. (1987) Hrit.
A. d., Koch, C. J., Fuciarelli, J. Cancer 56, 395-400.
26. 21.
Ho, M. K., Geist, C., Murray, ,J., Kato, K., Delellis, R., Wolfe, H., Carney, W., Rabin, H., and Durda, 0. (1988) Cancer Res. 48, 4969-4975.
Urtasun, R. C., Chapman, J. D., Raleigh, .J. A., Franko, A. J., and Koch, C. J. (1986) Int. J. Rod. Oncol. Biol. Phys. 12, 12631267.
35.
Epstein, A. L., Chen, 48, 5842-5848.
Rasey, Radiut.
36.
Coleman,
37.
Miller, M. R., Ulrich, R. G., Shu-Fong, W. T., and Korn, D. (1985) J. Biol. Chem. 260, 134-138. Freeman, J. W., Hazlewood, J. E., Auerbach, P., and Busch, H. (1988) Cancer Res. 48, 5246-5250. Okada, C. Y., and Rechsteiner, M. (1982) Cell 29, 33-41. Beckers, C. J. M., Keller, D. S., and Balch, W. E. (1987) Cell 50, 523-534. Charkabarti, R., Pf’eiffer, N. E., Wylie, D. E., and Schuster, S. M. (1989) J. Biol. Chem. 264, 8214-8221.
J. S., Krohn, K., Freauff, Res. 87, 455-459. P. L., and
Durand,
S., and
Res. 43,
(1981)
Begg, A. C., Engelhardt, E. I,., McNally, N. d., Terry, N. H. A., and Wardman, P. (1983) Brit. J. Radial. 56, 97Ob973. Valeriote, F., and Vanputten, L. M. (1975) Cancer Res. 35,26192630.
327638.
Koch, C. J., Kruuv, J. Radiat. Riol. 23,
April version
Cancer
S. M.
29.
Received Revised
R. E. (1983)
Larson,
Olive, 3280.
31.
K., Fairweather, D. S., Bradwell, A. R., Blackburn, P. W., Howell, A., Reeder, A., and Hine, K. R. Med. J. 283, 942-944. T., Evan, G. I., Ritson, A., Watson, J., Wraight, P., K. (1984) Bit. J. Cancer 54, 761-769.
34.
28.
30.
:33.
Halsall, A. J. C., Dykes, (1981) Bit. Chan, S. Y. and Sikora,
20, 1990. received
J., Frey, 67-74.
July
N. E., and Sayder,
18, 1990.
R. A. (1973)
Int.
39. 40. 41.
C. N. (1988)
F. M., and Taylor, J. N&l.
Cancer.
C. R. (1988) Inst.
80,
Cancer
Res.
310-317.
