Detection of Hypoxic Cells by Monoclonal Antibody Recognizing 2-Nitroimidazole
Adducts I
E d i t h M . L o r d , 2 L e e H a r w e l l , a n d C a m e r o n J. K o c h
Cancer Center, University of Rochester, Rochester, New York 14642 [E. M. L., L. H.], and Radiation Oncology, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6072 [C. J. K.]
A B S T R A C T
Hypoxic cells in tissue pose many medical problems, and there is a need for more accurate measurements of tissue hypoxia. However, measure- ment of the pO2 and the extent of hypoxia within normal and tumor tissue have proven difficult. One of the most sensitive of the currently available methodologies involves the oxygen-dependent metabolic activation of ni- troheterocyclic drugs, leading to adducts between the drugs and cellular macromolecules. Limitations of the present drugs and adduct-detection methods prompted the present studies. A pentafluorinated derivative [EFS; 2-(2-nitro-1/-/-imidazol-l-yl)-N-(2,2,3,3~-pentafluoropropyl)acet- amide] of etanidazole was synthesized with the expectation of lessening some of the non-oxygen-dependent variability in adduct formation ob- served previously with other nitroaromatic compounds. EF5-protein con- jugates, prepared by radiochemical reduction, were found to be immuno- genie and allowed the development of monoclonal antibodies. One of these antibodies, ELK2-4, has been characterized and found to be highly spe- cific for the EF5 adducts whether produced radiochemically or by cellular bioreductive metabolism. 9L rat glioma cells pretreated with EF5 under hypoxic, compared with aerobic, conditions were readily discriminated immunochemically using fluorochrome-conjugated secondary antibodies which recognize the ELK2-4 antibody subtype (IgG1). Similarly, the cen- tral region of multicellular spheroids, composed of EMT6 mouse mam- mary sarcoma cells, was selectively visualized by immunohistochemistry after the spheroids were incubated for 4 h in 0.5 mM EF5. Tumor biopsy, preparation, and immunohistochemical staining 24 h after treatment of tumor-bearing animals with drug also demonstrated high contrast regions within EMT6 mouse or Morris 7777 hepatoma rat tumors. The use of this new compound and its highly specific monoclonai antibody may allow elucidation of bioreductive metabolism of the nitroheterocyclics and sig- nificantly improve technologies for the quantitation of tissue pOz.
INTRODUCTION
Oxygen consumption by cells provides their major source of en- ergy, through oxidative phosphorylation. Limitations in oxygen sup- ply from the tissue vasculature result in hypoxia. Hypoxia can occur rapidly, during the shutdown of blood flow from stroke or traumatic injury, or much more slowly in cases which arise from various other types of progressive vascular disease. Alternatively, the rapid growth of some tumors can outpace their host-provided blood supply causing regional hypoxia and necrosis. In these and other situations, the hy- poxic cells pose severe medical problems. For example, hypoxic tumor cells are radioresistant and can cause treatment failure. The same supply or diffusional barriers which prevent oxygen from reach- ing the hypoxic regions of tumors can prevent the diffusion of reactive chemotherapeutic drugs (1-3). Alternate therapies such as immuno- therapy may be similarly limited by the impaired ability of cytotoxic host cells or antibodies to reach the hypoxic areas.
Detection of hypoxia in tissue is thus important in many areas of disease, with the most obvious involving two of the leading causes of death, cardiovascular disease and cancer; it has become clear that
Received 5/24/93; accepted 9/30/93. The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 Work partially supported by Grant CA28332 (E. M. L.) from the National Cancer Institute and Grant DHP-27B (C. J. K.) from the American Cancer Society.
2 To whom requests for reprints should be addressed.
methods are required to identify hypoxia in individual patients. The presence of hypoxic cells in relatively large, previously diagnosed human tumors has been measured directly using polarographic oxygen needle electrodes (4-8) and has been inferred in numerous other ways (9, 10). Furthermore, direct comparisons of tumor hypoxia and radio- therapy resistance using needle oxygen sensors have shown a dramatic correlation (7). However, measurement of hypoxia in smaller tumors or regions of ischemia requires a detection system which can, in principle, monitor very low oxygen concentrations in small, inacces- sible tissue volumes. The range of oxygen concentrations which cause radiation or chemical resistance is greatly below the venous pO2 of about 35 torr (mm Hg; see Fig. 1).
Measurements of such low oxygen values are technically challeng- ing for two key reasons. The first is that many measurement tech- nologies have only a limited dynamic range, sometimes above and sometimes below the range of interest (Fig. 1). For example, changes in pyridine nucleotide fluorescence (11) are most sensitive at oxygen levels greatly below those which affect radiation damage, whereas changes in hemoglobin saturation occur at oxygen levels substantially above those which affect radiation sensitivity. One exception involves the oxygen dependence of radiation damage to DNA, which is thought to closely parallel that for survival (12). At present, technical limita- tions can sometimes prevent polarographic needle sensors from ac- curately monitoring the very low levels of oxygen characteristic of radiation resistance (8). Macroscopic Clark sensors (13) and phospho- rescence decay techniques (14) can adequately monitor oxygen over the entire range of oxygen concentrations shown but in any given application are limited to a dynamic range of about 1000. Only the latter can be used for imaging techniques in vivo. One of the newest monitoring techniques under current development is electron para- magnetic resonance (15). This technique may have the potential for an appropriate dynamic range, but elimination of interactions of the probes with biological materials is difficult.
The second technical limitation involves the nature of the oxygen- dependent signal. True oxygen monitors, like polarographic oxygen sensors (irrespective of other design factors), have a signal which is proportional to the measured quantity (oxygen partial pressure) so they can become noisy and inaccurate at such low levels (13). Fur- thermore, they tend to average oxygen levels over relatively large tissue volumes. Thus measurement of a small region of extreme hypoxia is not feasible. Similarly, DNA damage, though closely re- lated to radiation sensitivity as described above, is greatly reduced at low oxygen levels and therefore limits detection.
What is required is a "no oxygen" detector. The signal from such a detector would be maximal in the absence of oxygen and would be inhibited by oxygen over the range of oxygen concentrations affecting the phenomena of interest. Examples include phosphorescence decay and the technique used in this report, nitroheterocyclic binding. It has been clearly shown that the metabolism of nitroheterocyclic drugs (RNO2), known as hypoxic cell-radiation sensitizing agents or radio- sensitizers, leads to the formation of stable adducts with cellular macromolecules ("binding") and that these adducts are formed at a much greater rate in hypoxic than in aerobic cells (1, 16-22). Thus, detection of these bound adducts can provide information on the relative oxygenation of tissue at a cell-to-cell resolution.
MoAb DETECTS HYPOXIC CELLS VIA BOUND 2-NITROIMIDAZOLE
SURVIVAL AFTER CONSTANT RADIATION EXPOSURE ( e )
1.00 . . . . .
0
" OAO-
0
> I ~ Hemoglobin So~tta I II~ = ////////////////'~rr / NeeOle Oxygen Sensors �9 -a J NiVoaromatic Bindinq / / / f / / / / / / / / ~ r j T ' ) ~ 4 C m . L tO ~ P y t . Nucl. Fluor. J / / / / / / / / ONA Damage J / / / / / d / E l e c t r o n Pinata?" , f f ~ ' w ' ~ ' ~
% o r ~ k P 8 n n r nun~n I n n , u u , i n I m n nmnun, I u , n , r i ~ m I J , , , n a u , ] 0.01 0.1 I 10 100
Fig. 1. Importance of oxygen in affecting treatment response and the dynamic range possible with various oxygen measurement techniques, e , survival of V79 Chinese hamster cells given 10 Gy, using the well-known effect of oxygen on the radiation response. Various oxygen measurement scales are provided along the Y-axis and, since some units involve concentration (which is temperature dependent), all scales have been adjusted for temperatures of 37~ and have been adjusted with allowance for an equilib- rium partial pressure of water (--45 mm Hg). Various measurement technologies have optimized response regions to oxygen concentration with regions of reduced accuracy indicated by diagonal lines.
Several detect ion modal i t ies exist for the ni t roaromat ic adducts: (a)
for mon i to r ing in situ, incorporat ion of radioact ive or other i sotopes
wi th P E T 3 (23), s ingle pho ton emiss ion c o m p u t e d tomography (24), and M R I or M R S (25, 26); and (b) for moni to r ing via b iopsy, /3-emit -
t ing radioact ive isotopes (22, 27), endogenous f luorescence of nitro-
heterocycl ic products (21), and polyclonal ant ibodies against the ad-
ducts (28). If cel lular adducts to ni t roheterocycl ics were accessible to
ant ibodies p rov ided by the b lood supply, ant ibodies labeled wi th suit-
able isotopes wou ld also provide an elegant in situ detect ion system.
Al though polyclonal ant ibodies against adducts o f a hexafluori-
propyl) -2-ni t roimidazole]) o f the 2-ni t roimidazole mison idazo le have
been m a d e (28), all a t tempts at mak ing M o A b s to adducts o f this and
other 2-ni t ro imidazoles have failed. M o A b techniques , wi th their high specifici ty and t r emendous variety of detect ion technologies , can pro-
v ide the mos t detai led informat ion on the cellular or tissue distr ibution
of the adducts, inc luding subcel lular distr ibution of sites of metabo-
l ism and/or binding, or even the ident if icat ion of adducts f rom indi-
vidual proteins (Western blott ing). An ideal detec t ion scheme migh t
include a water-soluble, f luor ine-conta in ing ni t roheterocycle , al low-
ing M R I / M R S or PE T imaging of the actual drug or drug adducts, and
a monoc lona l ant ibody to the drug or drug adducts which al lows all
other detect ion possibilit ies. This report p rovides the first demonst ra - t ion of such versati l i ty in de tec t ion techniques.
M A T E R I A L S A N D M E T H O D S
Drug Synthesis. A pentafluorinated derivative of etanidazole, EF5, ap- peared to have suitable properties based on some of our previous studies which have identified inconsistencies in the binding properties of several other 2-ni- troimidazoles (29). This drug, in unlabeled and labeled form (2-14C position; 43 /xCi/mg), was synthesized by Dr. M. Tracy and colleagues at Stanford
3 The abbreviations used are: PET, positron emission tomography; MRt, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; MoAb, monoclonal anti- body; EFS, pentafluorinated derivative of etanidazole; BBI, Bowman-Birk inhibitor; ELISA, enzyme-linked immunosorbent assay; AMCA, aminomethylcoumarin-acetic acid.
Research International, Palo Alto, CA, and is referred to as EF5 in this manu- script. Synthesis and detailed characterization of this drug will be published elsewhere.
Binding of EF5 to Cells in Tissue Culture under Defined Oxygen Con- ditions. The cells used were derived from 9L rat glioma (30) or EMT6-Ro mouse mammary carcinoma (31). The cells were thawed from frozen stock on a roughly semiannual basis, and tests were made routinely to ensure that the cultures were free from Mycoplasma and other contaminants. The cells were cultured (37~ 95% air + 5% carbon dioxide; 100% relative humidity) in the exponential phase of growth by twice-weekly transfers using Eagle's minimal essential medium containing 12.5% v/v of either newborn calf serum (9L) or fetal calf serum (EMT6-Ro). Penicillin and streptomycin were also routinely included (all culture solutions were from Sigma Chemical Co.). On the day before an experiment, cells were trypsinized and plated onto glass Petri dishes; -250,000 cells were confined to the central area of the dish followed by overnight incubation at 37~ as described previously (32). The dishes were then removed from the incubator, cooled to 0--4~ and their medium was replaced with drug containing medium, first as a rinse (1 ml) which was simply aspirated and then as the actual medium used for the experiment (also 1 ml). Dishes were then placed in leak-proof aluminum chambers which were con- nected to a manifold, allowing them to be deoxygenated with a series of gas exchanges taking approximately 30 min. The confinement of cells to the central area of the dish and the use of a small volume of medium allow very rapid equilibration of the gas and liquid phase to improve the control of oxygen concentration (32). After gas exchange, the chambers were quickly warmed to 37~ by immersion in a water bath and then dried and transferred to a warm room, also at 37~ To prevent minor gradients of oxygen or potentially larger gradients of nutrients/metabolites, the chambers were also shaken gently (1 Hz; 2.5-cm stroke).
Binding of radioactive nitroheterocyclics after incubation under defined experimental conditions was assessed as described previously (33).
Some cells were treated as above but nonradioactive EF5 was used instead. After incubation in air or nitrogen, the cells were removed from the glass dishes with trypsin and then allowed to attach to microscope slides where they were then stained using conventional immunohistochemical techniques (see below).
Preparat ion of Antigen. Protein conjugates of EF5 were prepared using radiochemical reduction methods in a two-step process. It had been shown previously (34) that the chosen protein must contain a high mol fraction of reduced cysteine residues for this process to be efficient, i.e., to achieve a high fraction of (bound drug:reduced drug) and to achieve a density of adducts per protein which was as high as possible. The cysteine residues of most readily available proteins are either scarce, not accessible for adduct formation (e.g., alcohol dehydrogenase; data not shown), or are oxidized as cystine dimers which are often important in determining the structure of the protein (e.g., albumin). It was not possible to reduce protein cystines by the addition of excess quantities of reducing agents such as dithiothreitol or mercaptoethanol, which could simultaneously reduce and stabilize cystine-containing proteins. Drug-adducts would then preferentially form with the excess low molecular weight thiol, and separation of the proteins from the low molecular weight thiols inevitably led to large protein losses (data not shown). To circumvent this difficulty, we exploited the discovery that the cystine dimers of many proteins can be very efficiently reduced via a radiochemical chain reaction, i.e., low doses of ionizing radiation in an oxygen-free, formate-containing solution (35). However, the modified protein is often relatively insoluble (possibly because of the formation of disulfide bridges between molecules). 4 Thus it was nec- essary to identify a protein with high cystine content and having relative freedom from precipitation after radiochemical reduction. BBI (36), a trypsin/ chymotrypsin inhibitor from soybeans (7 cystine bridges; molecular mass 7800), was found to have near optimal characteristics from this point of view, and reduction of up to an average of 8 cysteine residues was possible while maintaining protein solubility and thiol stability. The EF5-BBI conjugates were then made in a second radiochemical reduction step which could be accom- plished in the same solution.
Preparat ion of Monoclonai Antibodies. C57Br/cdJ • SJL/Br-H-2(k) mice were given injections of 15 /~g of protein antigen emulsified with an equal volume of Freund's complete adjuvant by i.m. injection at each of two
MoAb DETECTS H Y P O X I C CELLS VIA B O U N D 2-NIq-ROIMIDAZOLE
sites. An identical booster injection of antigen in Freund's incomplete adjuvant was given i.p. 2 months later. Three weeks after the booster injection, a serum sample was tested for antibody activity against protein-EF5 adducts v e r s u s
protein alone using a standard ELISA assay, t= A mouse whose antiserum reacted with the EF5, whether it was conjugated
to either the immunizing protein (BBI) or other proteins, was selected. This ..a .a
mouse received additional i.p. injections of EF5-BBI conjugate (80/xg) at 4, m 3, 2, and 1 day before sacrifice and spleen removal. The fusion was performed o
Ilg as described previously (31). Polyethylene glycol (P7777; Sigma) was pre- m
a . pared at 35% in serum-free media containing 5% dimethytsulfoxide (tissue culture grade; Sigma) with pH adjusted to 7.4 with 4 r~ NaOH. All cells and _z
a reagents were warmed to 37~ prior to the fusion. After the fusion, the cells ~,
m were bulk-cultured in a roller bottle for 2 days. The cells were then centrifuged, m resuspended in 10 ml of freezing medium (culture medium with 40% fetal bovine serum and 10% dimethyl sulfoxide), and frozen at 1 ml per vial.
One vial was subsequently thawed and cultured by splitting the cells into 5 96-well microtiter plates in 10% fetal bovine serum and HAT-containing cul- ture medium (1.36 mg/ml hypoxanthine, 0.038 mg/ml aminopterin, and 0.0176 mg/ml thymidine) to select for fused cells, with 2000 irradiated mouse peri- toneal cells/well as feeder cells. Wells were screened for specific antibody- producing cells using an EL1SA assay and BBI v e r s u s BBI-EF5 conjugate as antigens. Six hybridomas were selected based on the differentiation of these two antigens and were further screened against a panel of additional antigens. One of these (ELK2-4) was selected based on its strong selective reaction against protein-EF5 conjugates and lack of reactivity against any other antigens tested, including whole-cell protein preparations. This clone was slowly adapted to growth in serum-free medium (EXCELL 300; JRH Biosciences), and relatively pure MoAbs were purified by a simple uttrafiltration using Amicon Centriprep ultrafilters (M, 30,000 cutoff). Some MoAb protein was made autofluorescent by the addition of AMCA using standard protein modi- fication kits available from Pierce.
Histological Staining. Cells (9L; incubation at 37~ in the presence of 0.5 mM EF5 for 4 h in air or nitrogen) were allowed to attach to microscope slides. The slides were air dried, briefly exposed to a 0.1% solution of Triton X-100, and then rinsed to remove the detergent. The slides were treated overnight at 4~ with ELK2-4 antibody and rinsed for 10 min in phosphate-buffered saline. They were then stained with fluorescein-tabeled goat anti-mouse IgG (The Jackson Laboratory), rinsed to remove unbound second antibody, and observed using a fluorescent microscope.
EMT6-Ro spheroids were grown as described previously and then placed in individual spinner flasks to allow development of an equilibrium fraction of hypoxic cells (37). EF5 was added at a final concentration of 0.5 mM and the spheroids were incubated at 37~ for 4 h. The spheroids were removed, washed, embedded in OCT compound, and frozen on dry ice. Cryostat sections (10/xm thick) were cut and placed on poly-L-lysine coated glass slides, and the sections were fixed in freshly prepared 1% paraformaldehyde for 10 rain. After rinsing, blocking antiserum (2% normal goat serum) was added and sections were incubated for 30 min at room temperature. The slides were treated with ELK2-4 overnight at 4~ washed 3 times, and then treated with fluorescein- labeled second antibody as indicated above. The slides were then washed and observed with a fluorescent microscope. ~.
A BALB/c mouse (25 g) bearing an EMT6 tumor was injected i.p. with 1.2 I--- >
ml of 2 nun EF5. Forty-eight h later, the mouse was sacrificed, and the tumor U-- o removed, embedded in OCT compound, and frozen on dry ice. Cryostat sec- ,~
tions of - 2 0 p,m thick were cut, and the sections were treated as above for the m spheroid sections except that only one antibody was used, namely the AMCA- _>
I - conjugated ELK2 described above. A similar experiment was performed in a < ..a Buffalo rat bearing a Morris 7777 hepatoma tumor, with the drug dose in- tu creased in proportion to the weight of the larger animal, n-
R E S U L T S
The oxygen dependency o f binding was tested using 20 tzra and 500
~,M drug for EMT6-Ro cells in tissue culture. The rate of binding of
EF5 in ni trogen was reduced 2-fold for cells at intermediate oxygen
levels ( - -3 torr) and 10-fold for cells growing in 15 torr oxygen (Fig.
2). Note that at 37~ with gas phase in equil ibrium with water vapor,
the absolute oxygen concentrat ion in micromolar is approximately 1.4
t imes the partial pressure in torr, where torr is the standard abbrevia-
10 -2
10 -3
10 .4 -
10-5
._... & 1
�9 ..... , ---- J
j �9
I [ ]
O I
INCUBATION TIME AT 37 ~ (hr)
Fig. 2. Inhibition of binding by oxygen in EMT6-Ro ceils. EF5 concentrations were 500 /.tM (A, I , O) and 20 /xM (A, 17, C)). Binding occurs to a much greater extent in hypoxic cells (~, A) than cells at intermediate oxygen levels ( -3 torr or 0.4%; [7, I) , and a 15-fold reduction in binding is seen for cells growing in 15 torr (2%) oxygen (C), O). A log-log scale has been used to accommodate the large range of binding. Binding is linear with time, and the lines have been drawn to indicate this.
tion for m m of Hg (Fig. 1). For cells at even higher oxygen levels (air;
data not shown), binding was reduced 50-fold.
Having determined a clear oxygen dependency o f binding for EF5,
it was then necessary to demonstrate the specificity of the monoclonal
antibody f rom the ELK2-4 hybr idoma clone. Using standard ELISA
techniques as described above, a variety of other ni troaromatic com-
pounds were tested for their ability to competi t ively inhibit the bind-
ing of antibody to EF5-bovine serum albumin adsorbed to the ELISA
plates (Fig. 3). These data demonstrate that the only compound to
inhibit binding effect ively was EF5 itself. Even the most reactive o f
the other compounds (CCI-103F, a derivative o f misonidazole which
contains two CF 3 terminal residues on its sidechain) required at least
a 100-fold higher concentrat ion for inhibition.
Immunohis tochemica l staining techniques can be simplified signifi-
cantly if detection molecules can be added to the monoclonal directly.
Two types of protein modif icat ion chemistr ies available f rom Pierce
were used to add the f luorophor A M C A to the ELK2-4 antibody. In
one type o f chemistry (NHS-sulfo), adducts are fo rmed via pr imary
Fig. 3. Competition ELISA assay using various nitroaromatic compounds to compete against EF5-protein adduct adsorbed onto ELISA plates. The compounds tested were EF5 (O), etanidazole (O), misonidazole (A), dimethylmisonidazole (IS]), NP-6 [2-(2,4-dinitro- 1H-pyrrol-I-yl)ethanol] (*), and CCI-103F (ll). Note that etanidazole did not compote with EF5 at 1000-fold excess concentration and that the most active competitor (CCI- 103F) was at least 100-fold less reactive.
MoAb DETECTS HYPOXIC CELLS VIA BOUND 2-NITROIMIDAZOLE
amines on the antibody. In the second, the modification is effected via protein sulfhydryl groups (HPDP); since antibodies do not have free sulfhydryl groups but contain a number of cystine disulfides, some of the latter were first reduced using mild radiochemical reduction (see "Materials and Methods"; Ref. 35). Addition of AMCA by either method did not alter the EF5 binding capacity of ELK2-4 antibody (Fig. 4).
In the above examples, the EFS-protein antigen used was derived by radiochemical reduction techniques. It was necessary to confirm that ELK2-4 antibody would also recognize macromolecular adducts of EF5 produced by cellular bioreduction pathways as would occur in vitro or in vivo. EMT6 tumor cells were pretreated with 0.5 mM EF5 under varying levels of oxygen or with etanidazole under extreme hypoxia. Protein extracts of whole cells were then made by homog- enizing the cells in 0.1% Nonidet P-40 detergent (Sigma), and these proteins were then used as competitive inhibitors, as described above. Proteins from EF5-treated hypoxic cells were much more reactive than those at intermediate oxygen levels (1% oxygen). Proteins from EFS-treated aerobic cells or etanidazole-treated hypoxic cells were nonreactive (Fig. 5). The latter result confirms the specific require- ment for EF5 and demonstrates that the reactivity described does not result from specific protein changes which could be associated with hypoxic growth conditions.
A final requirement was to determine whether the ELK2-4 antibody would recognize EF5-adducts in situ, using whole cells or thin sec- tions, with or without fixation (Fig. 6, A-D). In the first example, 9L rat glioma cells were incubated for 4 h in air versus nitrogen in the presence of 0.5 mM EF5. Methods used were as described above to assess binding. The cells were trypsinized and allowed to attach to microscope slides; then they were air dried and stained overnight at 4~ with ELK2-4. Antibody binding was visualized using a fluores- cein-labeled second antibody (The Jackson Laboratory) recognizing the isotype (mouse IgG1) of ELK2-4. No antibody binding was ob- served for the aerobic cells but the hypoxic cells fluoresced brightly (Fig. 6A). A second example illustrates the binding pattern observed after whole spheroids growing in aerobic medium in vitro were incu- bated for 4 h in the presence of 0.5 mM EF5 (Fig. 6B). It can be seen that the outer rim of the spheroid was completely unstained, but the staining intensity increases for cells in the spheroid interior which are expected to be hypoxic. There was no staining in the necrotic center
1.2
1.0'
0.8 ~
0.6'
0.4
0.2
0.0 ,4,. l l
�9 m �9 *w - �9 . r r
0-s 10-a
i
,/o ii... ~ / ~ . . . . . . . . i
10 .7
"/0" /o
0
i . . - - - - - �9
6 0 0
. . . . . . . . | . . . . . . . .
10-6 10.5
ANTIBODY CONCENTRATION (g/ml)
Fig. 4. ELK2-4 antibody can be modified by directly conjugated fluorophors without loss of activity. The ability of modified antibody to react with protein-EF5 adducts was demonstrated by dkect antibody titrations using either untreated antibody (�9 or antibody to which fluorescent AMCA molecules were added using the HPDP (ll') or NHS sulfo (&) kits available from Pierce.
Fig. 5. Competition ELISA assay using whole-cell protein preparations of EMT6-Ro cells exposed to 0.5 mM EF5 at varying oxygen concentrations. Protein from cells treated with EF5 in nitrogen (O) were much more reactive than those from cells treated with EF5 in 1% oxygen (1). Protein from EF5-treated aerobic cells (&) or etanidazole-treated hypoxic cells ((3) was not reactive.
of the spheroids, indicating the necessity of having viable ceils to form adducts with the EF5 molecules.
The ability of the antibody to detect regions of high binding of EF5 given to tumor-bearing animals was also confirmed (Fig. 6, C and D). Tumors typically contain large amounts of host (mouse) antibody so that a second antibody could not be used to detect the ELK2-4 mouse monoclonal in the case of mouse tumors. To circumvent this problem, AMCA-conjugated EKL2-4 (Fig. 4) was used. High contrast areas were found both in EMT6 mouse and Morris hepatoma 7777 rat tumor tissue (Fig. 6, C and D, respectively).
DISCUSSION
The data presented in this report have demonstrated several quali- tative aspects of EF5 metabolism which are now expected for 2-ni- troimidazoles and other nitroaromatics, i.e., dramatic inhibition with increasing oxygen concentration (Fig. 2) and the ability to discrimi- nate hypoxic from aerobic tissue (Fig. 6). In more quantitative terms, an oxygen-sensing drug should have the same characteristic depen- dence of its metabolism on oxygen concentration in all cells or tissues of interest. We have previously demonstrated conclusively that this is not the case for misonidazole (27, 29, 38, 39). Although the rate of bioreductive metabolism of misonidazole always has a large oxygen- dependent component, it is, in absolute terms, extremely variable from one cell/tissue type to the next, both in terms of absolute binding rate and in the oxygen dependence of the process. At this time, the cause of the variability in absolute binding rate for misonidazole is not clear. Possibly, there are enzymes (like DT-diaphorase) which can donate two electrons to misonidazole, thus bypassing the oxygen-dependent step. It is known that the DT-diaphorase activity of various cell types is highly variable (40). We have found that the nonoxygen-dependent variability associated with the metabolism of misonidazole is not shared with closely related compounds such as etanidazole (29, 38).
Etanidazole was developed with the principal objective of reducing patient neurotoxicity associated with misonidazole (41), but as a marker of hypoxia, it suffers from two disadvantages (29). First, most of the bioreductive metabolites of etanidazole are acid-soluble, pre- sumably low-molecular-weight compounds which would be lost dur- ing typical histological processing. This property is thought to be caused by its extreme hydrophilicity (42). Secondly, etanidazole has no specific marker atom which could be used in noninvasive assays
MoAb DETECTS HYPOXIC CELLS VIA BOUND 2-NITROIMIDAZOLE
Fig. 6. lmmunohistochemical staining of cells and tissues by ELK2-4 after treatment with EFS. A (t~)p left), compares the binding of antibody to 9L cells treated with 0.5 mM EF5 in the presence of either air (dark) versu.s nitrogen (light). l'hc cxposnrcs of drugs, antibodies, and photography were all identical ira the two halves of the figure. Only the oxygen concentration during incubation ,.vas varied. B (upper rig, ht), re'suits of ~t similar experiment ,.̀ , hcrc whole spheroids (EMT6-Ro ceils) gro`,ving in spinner flasks were incubated with 0.5 na'~l EF5 for 4 h. Ten-/,tm frozen sections are illustrated. A aml LI. slides were treated using ELK2-4 and a second fluorcscein-containing antibody as described in "Materials and Methods." C (lower left), another example showing 20-/tin sections from an actual tumor-bearing ztnimal. A BALB/c mouse bearing an EM'I6 tumor received i.p. injections of 1.2 ml of 2 mM EFS. Section,, were treated as above for the spheroid sections, except th~t only one :mtibod,, was used. namely the AMCA conjug;aled ELK2-4 described in "Materials and Methods." The AM('A dye fluoresces in the blue, using UV excitation, but as above, the bright areas are those expected Io cont~dn hypoxic cells. D (lower right), results of a similar experiment using a buffalo rat bearing a Morris 7777 Iunlt_)r,
such as PET, MRI-MRS, or single photon emission computed tomog- raphy. We initiated this work hoping that fluorine-containing deriva-
tives of etanidazole would, like the parent drug, have limited nonoxy-
gen-dependent variations in binding, while allowing the detection
techniques offered by the fluorine atoms (PET; MRS-MRI). Addition-
ally, the 5 fluorines of EF5 substantially decrease its hydrophilicity
compared with etanidazole, thus causing a greatly decreased propor-
tion of acid-soluble adducts. Al though EF5 is much less polar than
etanidazole, it maintains a reasonably high solubility in physiological
saline at 37~ ( - 1 4 raM).
It was also reasoned that the pentafluorinated side chain of EF5
might be sufficiently nonphysiological as to allow the formation of highly specific antibodies. This may have been a contributing factor in
the successful production of polyclonal antibodies to CCI-103F in the
pioneering studies of Raleigh e t al . (28, 43). This reasoning may have
been justified, considering the current successful production of mono-
clonal antibodies to the present drug, EF5.
There are two aspects of the binding characteristics of MoAbs which are relevant to their experimental and/or clinical use, specificity
and affinity. In determining the relative specificity of ELK2-4, a large number of antibodies from various hybridoma clones were examined.
It was clear that the antibodies from most clones readily discriminated
the antigenic adduct-containing proteins from untreated protein (data
not shown); however, it appears that the considerable manipulations involved in protein disulfide reduction and adduct formation allowed
the expression of many new protein epitopes, most of which had little
to do with the actual EF5 drug adduct. We therefore realized that a
very general screening mechanism was required and settled on whole-
cell protein preparations prepared from either hypoxic cells (control)
or hypoxic EF5-treated cells (experimental). This selection technique
has allowed the choice of the ELK2-4 antibody which, to date, has been found to show no binding to non-EF5-treated tumor tissue. Much
more work is presently required to test for this specificity in a broad
variety of tissues. The affinity of an antibody for its antigen determines the useful
range of sensitivity and to some extent affects the optimization of its use ( i . e . , a low affinity antibody might typically show more rapid
dissociation kinetics, and therefore its optimal use would require
shorter times for rinsing, nonuse o f secondary antibodies, etc.). The
affinity of ELK2 (based on equilibrium dialysis experiments; data not shown) is of the order of 1 0 6 for the parent drug and 5 to 10 t imes
higher for protein adducts of EF5. Thus, the present antibody should
be very usable at adduct concentrations of the order of 10 /x i and higher. From the data in Fig. 2, it is clear that dramatically higher
MoAb DETECTS HYPOXIC CELL"; VIA BOUND 2-NITROIMIDAZOLE
levels can easi ly be achieved . For example , in hypox ic E M T 6 - R o cells
t reated wi th 500 t~r, drug, the adduct level i s more than 2 mM after
on ly 2 h o f incuba t ion ( i .e . , 0.003 pmol /ce l l ; cel ls have a size o f - 1 . 5
pl). Prev ious s tudies wi th re la t ively nonpo la r n i t ro imidazo les l ike
mi son idazo l e sugges t that 500/~M drug levels w o u l d not be acceptable
in humans . However , at d rug levels w h i c h are l ikely to be comple t e ly
non tox ic (e .g . , 20 /~M) E M T 6 - R o cells still a ccumula t e 1 5 - 3 0 0 /xM
drug over the range o f o x y g e n levels typ ica l ly expec ted f rom normal
to ex t r eme h y p o x i a (Fig. 2). Fur thermore , it is qui te l ikely that anti-
bod ies can be f o u n d wi th even h igher aff ini t ies s ince on ly 10% of the
or iginal fus ion has been screened. O f course, one o f the ch i e f advan-
tages o f m o n o c l o n a l ant ibodies is that they can easi ly be pur i f ied and
can be m a d e to inc lude a de tec t ion mo ie ty w i th in the molecu le . This
e l imina tes the two levels o f op t imiza t ion required w h e n us ing mul t ip le
ant ibodies in i m m u n o h i s t o c h e m i c a l techniques . Indeed, our da ta dem-
onstrate no reduc t ion in aff in i ty w i th the addi t ion o f A M C A f luoro-
phor us ing two di f ferent coup l ing chemis t r ies , and s imi lar results have
been ob ta ined wi th b io t iny la ted an t ibody (data not shown) . Further-
more , the present an t ibody appears able to detect an t igen us ing air-
dr ied v e r s u s p a r a f o r m a l d e h y d e - f i x e d t issue, a l lowing for the possibi l -
i ty o f no rmal t issue f ixa t ion techniques .
A l t h o u g h w e are in on ly the p re l imina ry s tages o f charac te r iz ing
this d rug and have on ly tes ted two o f the poss ib le de tec t ion schemes ,
it appears that EF5 and E L K 2 - 4 m a y a l low the de tec t ion o f hypox i a
in v i v o at h igh sens i t iv i ty and resolut ion.
R E F E R E N C E S
1. Chapman, J. D., Baer, K., and Lee, J. Characteristics of the metabolism-induced binding of misonidazole to hypoxic mammalian cells. Cancer Res., 43: 1523-1528, 1983.
2. Moulder, J. E., and Rockwell, S. C. Hypoxic fractions of solid tumors: experimental techniques, methods of analysis and a survey of existing data. Int. J. Radiat. Oncol. Biol. Phys., 10: 695-712, 1984.
3. Kennedy, K. A., Rockwell, S., and Sartorelli, A. C. Preferential activation of mito- mycin C to cytotoxic metabolites by hypoxic tumor cells. Cancer Res., 40: 2356- 2360, 1980.
4. Cater, D. B., and Silver, I. A. Quantitative measurements of oxygen tensions in normal tissues and in the tumors of patients before and after radiotherapy. Acta Radiol, 23: 233-256, 1960.
5. Wendling, P., Manz, R., Thews, G., and Vaupel, P. Heterogeneous oxygenation of rectal carcinomas in humans. A critical parameter for pre-operative irradiation. Adv. Exp. Med. Biol., 180: 293-300, 1984.
6. Gatenby, R. A., Kessler, H. B., Rosenblum, J. S., Cola, L. R., Moldofsky, P. J., Hartz, W. H., and Broder, G. J. Oxygen tension in human tumors: in vivo mapping using cq'-guided probes. Radiology, 156: 211-214, 1985.
7. Gatenby, R. A., Kessler, H. B., Rosenblum, J. S., Cola, L. R., Moldofsky, P. J., Hartz, W. H., and Broder, G. J. Oxygen distribution in squamous cell carcinoma metastases and its relationship to outcome of therapy. Int. J. Radiat. Oncol. Biol. Phys., 14: 831-838, 1988.
8. Kallinowski, E, Zander, R., Hoeckel, M., and Vaupel, P. Tumor tissue oxygenation as evaluated by computerized-pO2-histography. Int. J. Radiat. Oncol. Biol. Phys., 19: 953-961, 1990.
9. Bush, R. S., Jenkins,. R. D. T., Allt, W. E. C., Beale, F. A., Bean, H., Dembo, A. J., and Pringte, J. E Definitive evidence for hypoxic ceils influencing cure in cancer therapy. Br. J. Cancer, 37 (Suppl II1): 302-306, 1978.
10. Thomlinson, R. H., and Gray, L. H. The histological structure of some human lung cancers and the possible implications for radiotherapy. Br. J. Cancer, 9: 539-579, 1955.
11. Sugano, T., Oshino, N., and Chance, B. Mitochondrial functions under hypoxic conditions: the steady states of cytochrome c reduction and of energy metabolism. Biochim. Biophys. Acta, 347: 340-358, 1974.
12. Olive, P., Banath, J., and Durand, R. Heterogeneity in radiation-induced DNA damage and repair in tumor and normal cells measured using the "Comet" a~say. Radiat. Res., 122: 86-94, 1990.
13. Koch, C. J. Polarographic Oxygen Sensor. United States Patent 5030336. July 9, 1991.
14. Rumsey, W. L., Vanderkooi, J. M., and Wilson, D. F. Imaging of phosphorescence: a novel method for measuring oxygen distribution in perfused tissue. Science (Wash- ington DC), 241: 1649-1651, 1988.
15. Glockner, J. F., and Swartz, H. M. In vivo oximetry using two novel probes: fusinite and lithium phthalocyanine. Adv. Exp. Med. Biol., 317: 229-234, 1992.
16. Taylor, Y. C., and Rauth, A. M. Differences in the toxicity and metabolism of
2-nitroimidazole misonid~ole (Ro-07-0582) in HeLa and Chinese hamster ovary cells. Cancer Res., 38: 2745-2752, 1978.
17. Chapman, J. D. Hypoxic sensitizers--implications for radiation therapy. N. Eng. J. Med., 301: 1429-1432, 1979.
18. Varghese, A. J., and Whitmore, G. E Binding to cellular macromolecules as a possible mechanism for the cytotoxicity of misonidazole. Cancer Res., 40: 2165-2169, 1980.
19. Taylor, Y. C., and Rauth, A. M. Oxygen tension, cellular respiration and redox state as variables influencing the cytotoxicity of the radiosensitizer misonidazole. Radial. Res., 91: 104-123, 1982.
20. Rasey, J. S., Krohn, IC A., and Freanff, S. Bromomisonidazole: synthesis and char- acterization of a new radiosensitizer. Radial. Res., 91: 542-554, 1982.
21. Olive, P. L., and Durand, R. E. Fluorescent nitroheterocycles for identifying hypoxic cells. Cancer Res., 43: 3276-3280, 1983.
22. Urtasun, R. C., Koch, C. J., Franko, A. J., Raleigh, J. A., and Chapman, J. D. A novel technique for measuring human tissue hypoxia at the cellular level. Br. J. Cancer, 54: 453-457, 1985.
23. Rasey, J. S., Grunbaum, Z., Magee, S., Nelson, N. J., Olive, P. L., Durand, R. E., and Krohn, K. A. Characterization of radiolabelled fluoromisonidazole as a probe for hypoxic cells. Radiat. Res., 111: 292-304, 1987.
24. Parliament, M. B., Chapman, J. D., Urtasun, R. C., McEwan, A. J., Golberg, L., Mercer, J. R., Mannan, R. H., and Wiebe, L. 1. Non-invasive assessment of human tumour hypoxia with 123I-iodoazomycin arabinoside: preliminary report of a clinical study. Br. J. Cancer, 65: 90-95, 1992.
25. Raleigh, J. A., Franko, A. J., Treiber, E. O., Lunt, J. A., and Allen, E S. Covalent binding of a fluorinated 2-nitroimidazote to EMT-6 turnouts in BALB/c mice: detec- tion by F-19 nuclear magnetic resonance at 2.35 T. Int. J. Radiat. Oncol. Biol. Phys., I0: 1337-1340, 1984.
26. Li, S-J., Jin, G-Y., Fish, B. L., and Moulder, J. E. Correlation of hypoxic fraction with CCI-103F retention and (31)P magnetic resonance spectroscopy across multiple tu- mor lines. Radiat. Res., in press, 1993.
27. Franko, A. J., Koch, C. J., Garrecht, B. M., Sharplin, J., and Howorko, J. Oxygen dependence of binding of misonidazole to rodent and human tumors in vitro. Cancer Res., 47: 5367-5376, 1987.
28. Raleigh, J. A., Miller, G. G., Franko, A. J., Koch, C. J., Fuciarelli, A. E, and Kelley, D. A. Fluorescence immunohistochemical detection of hypoxic cells in spheroids and tumours. Br. J. Cancer, 56: 395-400, 1987.
29. Koch, C. J. The reductive activation of nitroimidazoles: modification by oxygen and other redox-active molecules in cellular systems. In: G. E. Adams, A. Breccia, E. M. Fielden, and P. Wardman (eds.), Selective Activation of Drugs by Redox Processes, NATO Series A, pp. 237-247. New York: Plenum Publishing Corp., 1990.
30. Franko, A. J., Koch, C. J., and Boisvert, D. P. J. Oxygen distribution in 9L tumors and spheroids and KHU tumor cords. Cancer Res., 52: 1-7, 1992.
31. Harwell, L. W., Bolognino, M., Bidlack, J. M., Knapp, R. J., and Lord E. M. A freezing method for cell fusions to distribute and reduce labor and permit more thorough early evaluation of hybridomas. J. lmmunol. Methods, 66: 59~57, 1984.
32. Franko, A. J., Freedman, H. I., and Koch, C. J. Oxygen supply to spheroids in liquid overlay and spinner culture. Recent results in cancer research 95. In: H. Acker, J. Cartsson, R. Dnrand, and R. M. Sutherland (eds.) Spheroids in Cancer Research, pp. 162-167. Berlin: Springer Verlag, 1984.
33. Koch, C. J., Stobbe, C. C., and Baer, K. A. Metabolism induced binding of a4C- misonidazole to hypoxic cells: kinetic dependence on oxygen concentration and misonidazole concentration. Int. J. Radiat. Oncol. Biol. Phys., 10: 1327-1332, 1984.
34. Raleigh, J. A., and Koch, C. J. The importance of thiols in the reductive binding of 2-nitroimidazoles to macromolecules. Biochem. Pharmacol., 40: 2457-2464, 1990.
35. Koch, C. J., and Raleigh, J. A. Radiolytic reduction of protein and non-protein disulfides in the presence of formate: a chain reaction. Arch. Biochem. Biophys., 287: 7544, 1991.
36. Birk, Y. The Bowman-Birk inhibitor. Int. J. Peptide Protein Res., 25: 113-131, 1985. 37. Franko, A. J., and Koch, C. J. The radiation response of hypoxic cells in EMT6
speroids in suspension culture does model data from EMT6 tumors. Radial. Res., 96: 496-504, 1983.
38. Koch, C. J., Giandomenico, A. R., and Iyengar, C. W. L. Bioreductive metabolism of AF-2 [2(2-furyl)-3-(5-nitro-2-furyl)acrylamidet combined with 2-nitroimidazoles: implications for use as hypoxic cell markers. Biochem. Pharmacol., 46: 1029--1036, 1993.
39. Franko, A. J., and Koch, C. J. Binding of misonidazole to V79 spheroids and fragments of Dunning rat prostate and human colon carcinoma in vitro: diffusion of oxygen and reactive metabolites. Int. J. Radiat. Oncol. Biol. Phys., 10: 1333-1337, 1984.
40. Dulhanty, A. M., and Whitmore, G. E Chinese hamster ovary ceil lines resistant to mitomycin C under aerobic but not hypoxic conditions are deficient in DT-diaphorase. Cancer Res., 51: 1860-1865, 1991.
41. Brown, J. M., Yu, N. Y., Brown, D. M., and Lee, W. W. SR-2508: a 2-nitroimidazole amide which should be superior to misonidazole as a radiosensitizer for clinical use. Int. J. Radiat. Oncot. Biol. Phys., 7: 695-703, 1981.
42. Born, J. L., Smith, B. R., Harper, N., and Koch, C. J. Metabolism and radiosensiti- zation of 4,5-dimethylmisonidazole, a ring substituted analog of misonidazole. Bio- chem. Pharmacol., 43: 1337-1344, 1992.
43. Miller, G. G., Best, M. W., Franko, A. J., Koch, C. J., and Raleigh, J. A. Quantitation of hypoxia in multicellular spheroids by video image analysis. Int. J. Radiat. Oncol. Biol. Phys., 16: 949-952, 1989.
1993;53:5721-5726. Cancer Res Edith M. Lord, Lee Harwell and Cameron J. Koch Recognizing 2-Nitroimidazole AdductsDetection of Hypoxic Cells by Monoclonal Antibody
