A comparison of the expression of the DIDS-binding proteins from normal and tumorigenic human cells.

A monoclonal antibody was prepared against DIDS, an inhibitor of anion transport, and used to compare the occurrence and distribution of DIDS-binding sites of tumorigenic and non-tumorigenic human somatic-cell hybrids. The monoclonal antibody (E8) was produced by the fusion of the mouse myeloma (NS-1) with mouse spleen cells and is of the IgG1 subclass. The apparent half-saturation of DIDS for HEp-2 cells is 16 microM and the reaction is rapid. The number of binding sites on tumorigenic and non-tumorigenic hybrid cells was the same. The DIDS-binding protein occurs homogeneously on all cells, a characteristic which distinguishes it from the possible tumour antigen recognised by the M/27 monoclonal antibody.

The work reported herein is the result of a search for the identity of the 4,4'-diisothiocyano-2,2'disulphonic stilbene-binding protein (DIDS-BP) of the cell membrane of non-erythroid cells. This search originated from the observation that DIDS could modify the binding of a monoclonal antibody (M/27) which recognises a membrane protein associated with glucose transport (Gingrich et al., 1981a, b;Banyard et al., 1982) which may be the Bramwell-Harris glycoprotein specific for tumorigenic cells (Bramwell & Harris, 1978). Thus it is important to know if, and how, the DIDSbinding protein is related to the glucose transport protein and the 105 Mr Bramwell-Harris glycoprotein.
In addition to these empirical reasons for wishing to know more about the DIDS-binding proteins of non-erythroid cells, there is reason to think that the removal of lactate and the buffering of intracellular protons by HCOQ transported from extracellular fluid ought to assume particular importance in malignant cells. The maintenance of intracellular pH depends, essentially, on three pathways: (i) the Na+/H+ (amiloride-sensitive) exchanger, (ii) the inorganic anion transporter and (iii) the organic anion transporter (Pace & Travin, 1983). It might be expected that anion-transport would be rapid in malignant cells because the high rate of glycolysis, which is characteristic of malignant cells, will result in the production of large amounts of lactic acid (Vaupel et al., 1981). If a drop of intracellular pH is to be avoided, exchange of Cl-/HCOQ, Na+/H+ and the efflux of lactate must be rapid (Roos & Boron, 1981). In addition anion transport may be an important modulator of cell growth and Received 26 July 1984; accepted 6 September 1984.
proliferation. It has recently been shown (L'Allemain et al., 1984) that if the HCOQ is removed from the growth medium of Chinese hamster lung fibroblasts then blockade of the Na+/H+ exchanger with amiloride prevents serumstimulated initiation of growth. Since DIDS specifically inhibits the flux of inorganic anions at micromolar concentration (Cabantchik & Rothstein, 1974;Deuticke et al., 1982) and at higher concentrations can also block lactate transport (Jennings & Adams-Lackey, 1982) it was chosen as the most appropriate probe to investigate the structure and regulation of the anion transporter of normal and malignant cells.
This paper describes the preparation of a monoclonal antibody (E8) against DIDS and describes the application of this antibody to study the amount and heterogeneity of expression of the DIDS-binding proteins of normal and tumorigenic human cells. This information is used to compare and contrast the DIDS-binding protein with the M/27 antigen.

Materials and methods
Cell culture Cells were maintained in Dulbecco's Modified Eagle Medium (Gibco, No. 430-2100) in an incubator maintained at 5% (v/v) CO2 in air at 37°C. The medium was supplemented with either 10% (v/v) foetal calf serum (FCS) or 5% foetal calf serum and 5% (v/v) newborn calf serum (NCS) together with 1 mM L-glutamine, 20 mM sodium pyruvate, 50 Iu ml -1 penicillin G, 50 Iu ml -1 streptomycin sulphate and 80 Iu ml-1 neomycin sulphate. The cultures were checked at regular intervals for mycoplasma by the method of Chen (1977). Cells to be harvested for the analysis of DIDS-binding proteins were detached from the glass bottles with 7 mM Na2HPO4-2H20, 3 mM NaH2PO4-2H20, 137 mM NaCl (PBS) containing 0.02% (w/v) EDTA. A suspension culture of HeLa cells was grown in Minimal Essential Medium modified for suspension culture (S-MEM, Gibco No. 410-1800). pH was maintained with 5% (v/v) CO2 in air and 25 mM NaHCO3. The medium was supplemented as described above using 5% (v/v) FCS/5% (v/v)

NCS.
Conjugation of DIDS to proteins DIDS (Pierce Chemical Company, No. 20047) was covalently attached to proteins and unbound DIDS removed by dialysis. The protein solutions were prepared at 1-10mgml-1 in PBS and DIDS was added to a 10-100 molar excess. The solution was mixed for 1.5-4 h at 20°C, in the dark, and the reaction continued overnight at 4°C. The solution was then extensively dialyzed against PBS and stored at -20°C. Proteins used were: bovine serum albumin, BSA (Sigma No. A7638), ovalbumin (Sigma No. A5503), human gammaglobulin, HGG (Sigma No. G2011) and carbonic anhydrase, CA (Sigma No. C7500).
Preparation of monoclonal anti-DIDS antibodies Female BALB/c mice were initially immunized subcutaneously with 50,ug DIDS-HGG in Freund's complete adjuvant. The immunization was repeated 3 times over 3 months and a final intraperitoneal injection of 200,pg of the antigen was given 4 days before removing the spleen for fusion. The nonsecreting myeloma NS-1 was used for fusion and standard techniques were followed (Kohler & Milstein, 1975).

Solid-phase radioimmunoassay
The presence of antibodies against DIDS was measured by radio-immunoassay of culture supernatants on Immunolon 2 Removawell strips (Dynatech Laboratories, No. 011-010-6302) which were coated with a DIDS-ovalbumin conjugate at lSjigmlin PBS for 1 h at 37°C. BSA (5%, w/v) was used to block the remaining binding sites. Primary and secondary antibodies were incubated in the well for 1 h each and free antibody removed by washing with PBS. Ovalbumin was used as a negative control.
In experiments to demonstrate the specificity of E8 for DIDS (Figure 1), the protein solutions were absorbed to the wells at 100 jg ml -1 in PBS for 1 h at 37°C and blocked with BSA, as before, but thereafter additional washing steps were used. Following the BSA-blocking the wells were washed with 25 mM Tris/HCl pH 7.4, 140 mM NaCl (TBS) for 5min then with TBS containing 0.05% (v/v) Triton x 100 for 10 min and finally washed in TBS again for 5min. Fifty jl of the primary antibody solution at 25gml-1 in TBS/1% FCS was applied per well. Unbound primary antibody was removed, after 1 h incubation at 20°C, by washing as before (2min each step). ['251]-rabbit antimouse antibody (RAM) was added at 2.6x 104 cpm per 50,u1 in TBS/1% FCS. After a further 1h incubation the wash procedure was repeated (5, 10, 5min) and the dry wells counted.

The binding of DIDS to the cell membrane
The concentration of DIDS necessary to saturate the high affinity binding sites was determined by a saturation-binding assay. Increasing concentrations of freshly prepared DIDS were added to 4 x 106 HEp-2 cells suspended in S ml of PBS. The reaction proceeded in the dark for 10min at 4°C with frequent mixing. The reaction was stopped by adding an equal volume of TBS containing 5mM 1,2-diamino ethane (Fluka, K3596-/25/2/4).
The cells were pelleted at 400g for 3 min and resuspended in S ml of TBS. The cells were pelleted again and resuspended to 2 x 107 cells ml -1 in PBS.
The amount of DIDS bound to the cell membrane was determined by saturation-binding assay.
Saturationand Trace-binding assays The basic method of Letarte-Muirhead et al., (1975) was followed. One hundred jil of E8 culture supernantant or 50 jil of purified E8 at 25mg mlwas used for the primary antibody incubation. The [1251]labelled rabbit anti-mouse IgG (H + L) (Miles No. 65.157) was used at 25 jgml-1 or higher. Cells were resuspended to 2 x 107Mmlin PBS and 50 pi was used per well. Determinations were made in triplicate or quadruplicate.

Flow cytometry
HeLa spinner cells were washed 4 x in PBS and DIDS conjugated to the membrane as described previously. Two hundred and fifty ,ul of cells at 2x 107 cellsml-1 were resuspended in saturating amounts of ascites (M/27) or purified antibody (E8) prepared in TBS/l % FCS/0.02% NaN3. (TBS/FCS/Az). Following an incubation of I h, the cells were washed with TBS/FCS/Az and incubated with 1:40 dilution of FITC labelled RAM (Dako, No. F232) for a further hour. The cells were washed 3 times and resuspended to 1 ml in PBS for analysis on the FACS (Beckton Dickinson, FACS IV). The cells were maintained at 4°C until the time of analysis.

Characterization of the monoclonal antibody
The specificity of the E8 monoclonal antibody was determined by comparing the binding of E8 to four different proteins to which DIDS had or had not been conjugated. DIDS was mixed at 100:1 molar ratio (DIDS:protein) to samples of ovalbumin, bovine serum albumin, human gamma globulin and carbonic anhydrase as described in materials and methods. Another monoclonal antibody, JCS-2, was used as a negative control. JCS-2 is an IgG2a monoclonal anti-glycophorin (unpublished results). In all cases the binding of E8 to the unconjugated proteins was similar to the binding of the negative control and significantly lower than the binding of E8 to the DIDS-conjugated proteins ( Figure 1). Hence, within the scope of the comparison E8 is specific for DIDS. The epitope or immunodominant region of DIDS is unknown but E8 is not absorbed by any of a panel of sulphated compounds including chondroitin 4-sulphate, heparin, fucoidon and carrageenan.
The E8 monoclonal antibody was shown to be of the IgGl isotype by double diffusion against monospecific commercial antisera (data not shown). The kinetics of binding of DIDS for the cell membrane The rate of binding of DIDS to HeLa cells followed biphasic kinetics, a rapid initial phase, which was completed in 5-10 min, was followed by a phase with a slower rate which did not reach saturation during the 60 min of the time-course studied. A separate experiment of the kinetics of the 0-10 min period conflrmed the rapid initial binding of DIDS (unpublished observations). The biphasic rate of binding observed in these experiments is very similar to the binding of H2 DIDS to the erythrocyte membrane at 20°C (Lepke et al., 1976, Figure 3).
A measure of the affinity of binding of DIDS for its receptor was examined in HEp-2 cells (a human laryngeal carcinoma cell line). A concentrationdependent saturation was reached at 50-100 M with half-saturation at 16 M (Figure 2). These experiments measure only the rapid phase of binding since the reaction was terminated at 10min. The value of 16yM is intermediate between the Ki of DIDS for the inorganic anion transporter of the human erythrocyte, (Cabantchik & Rothstein, 1974;Lepke et al., 1976)  with DIDS (at saturation) and the amount of DIDS bound 10%6 cells was measured by the saturation-binding assay. Initial experiments showed that HEp-2 had significantly (P<I0i0l) more DIDS-binding sites than the non-tumorigenic MRC-5 (a human foetal fibroblast) cells. A more stringent comparison was then made. The number of DIDS-binding sites was compared on a series of hybrids between D98/AH-2 (a HeLa variant deficient in HGPRT) and normal human fibroblasts (Stanbridge et al., 1982). No difference was found in the number of DIDS-binding sites between the tumorigenic variants (SE and 39E) and those in which tumorigenicity had been suppressed (5L and ESH) and (Figure 3). monoclonal antibody have demonstrated the antigen to be expressed heterogeneously; a small proportion of the cell population is labelled much more heavily than the bulk of the population (Gingrich et al., 1981b;Banyard & White, 1984). The cells heavily labelled with M/27 monoclonal antibody have two features in common, they are in S-phase of the cell cycle (Gingrich et al., 1981b) and they transport glucose more rapidly than the cell population as a whole (Banyard & White, 1984). Because the binding of DIDS to the cell-surface increases the number of M/27 binding sites it was suggested that the DIDS-binding protein and the M/27 antigen must either be the same, or be clearly associated in the cell membrane (Banyard et al., 1982). But since the binding of DIDS does not compete with M/27 (Banyard et al., 1982) the binding sites cannot be identical nor within the molecular radii of M/27 plus DIDS. They could occur at distant sites on the same molecule; in which case binding of DIDS could exert an allosteric effect at a distant site of the same molecule which facilitates the binding of M/27 to its epitope.
The fluorescence activated cell sorter (FACS) was used to distinguish between the two possibilities (i) that the binding sites occur on the same molecule, in which case their expression should be coregulated or (ii) that the binding sites occur on different molecules so that non-coordinate regulation might occur. The heterogeneity of binding of M/27 is recognised on the FACS as a skewing of the fluorescence distribution to the right. If all cells within the population bind the monoclonal equally, a single peak, which follows a Poisson distribution, should be recorded. Figure 4 shows E8 labels a single peak of cells homogeneously whereas the cells labelled with M/27 antibody demonstrate a skewed distribution of more fluorescent cells. The proportion of cells in the skewed tail (more fluorescent than channel 59) is -20% of the total cells counted, a figure in agreement with previous work (Gingrich et al., 1981b;Banyard & White, 1984). The absolute number depends, of course, on the arbitrary point of cut-off which was, in this instance, taken as the point of divergence between the E8 and the M/27 profile ( Figure 4). The value of 20% in the skewed tail was determined by subtracting the percent of  (Table I).

Discussion
The concentration at which DIDS half-saturates HEp-2 cells was found to be 16,M, a value which is intermediate between the Ki of DIDS for the inorganic anion transporter (Band 3) of the human erythrocyte, which is about 1 uM (Cabantchik & Rothstein, 1974;Lepke et al., 1976) and the Ki of DIDS for the organic anion transporter which Jennings & Adams-Lackey (1982) found to be inhibited by 80% at 10pMH2 DIDS. It should be noted that the DIDS and H2 DIDS have slightly different requirements for binding to their substrates (Lepke et al., 1976) but both effectively inhibit inorganic anion transport. The apparently simple kinetics of binding over 5-10min suggests a single major binding site but the use of an indirectantibody assay to determine this point in detail is inappropriate because of the complex kinetics which characterise the antigen-antibody reaction. Since the apparent Ki is intermediate between the inorganic and the organic anion transporter no rational guess can be made about which one is the major DIDS-binding protein of malignant cells but the molecules appear to be structurally distinct. The lactate transporter of the rabbit erythrocyte membrane labelled by [3H]H2-DIDS has a relative molecular mass (Mr) on SDS-PAGE of 4.3 x 104 (Jennings & Adams-Lackey, 1982) whereas the inorganic anion transporter has a relative molecular mass of 9.5 x 104 (Cabantchik & Rothstein, 1974).
The data comparing the number of DIDSbinding sites on tumorigenic and non-tumorigenic cells suggests that the Vma, for lactate and inorganic anions is not rate-limiting in tumorigenic cells. This finding supports the work of Johnson et al., (1980) which shows that lactate transport had to be almost completely blocked before intracellular pH fell. Spencer & Lehninger (1976) also found that the maximal rate of lactate transport was far in excess of the production of lactate by glycolysis in Ehrlich ascites tumour cells.
Three independent pieces of information suggest that the DIDS-binding protein and the M/27 antigen are different molecules. Firstly, the DIDSbinding protein, measured indirectly with the E8 monoclonal antibody, is expressed homogeneously on all cells within the population whereas M/27 is not. Secondly, DIDS does not compete with M/27 but in fact enhances binding of M/27 (Banyard et al., 1982). This observation suggests that when the DIDS-binding protein and M/27 antigen coexist in the same cell there is a close topological association between them. The importance of such membrane complexes is supported by the observation that DIDS and ouabain both independently cause an alteration of the NMR-resonance of [31P]-2,3diphosphoglycerate through the membrane (Fossel & Solomon, 1983) suggesting that long-range allosteric effects may operate between apparently unrelated membrane molecules. Finally, the dissimilar levels of expression of E8 and M/27 on the same cell also supports the idea that the proteins are distinct entities. A direct comparison of the number of sites per cell should not however be overinterpreted since the stoichiometry of [1251]-RAM binding to an IgM monoclonal antibody (M/27) and to a IgG1 monoclonal antibody (E8) is likely to be different. Furthermore since the M/27 determinant is heterogenously expressed the number of sites per cell can only be expressed as an arithmetic mean of the binding to the whole population.
The possible topological assocation between the M/27 antigen and the DIDS-binding protein is currently being studied directly by the purification of the DIDS-binding component. Previous, extensive attempts to purify the M/27 determinant in quantity have been unsuccessful. It is hoped that this new approach will finally resolve the relationship between the DIDS-binding protein and the Bramwell-Harris 1 x 105 Mr glycoprotein (Bramwell & Harris, 1978).
The technical help of Miss R. Stokes is gratefully acknowledged. The help of Mrs K. Rabl in typing the manuscript is also gratefully acknowledged. I thank Dr. R.B. Ashman and Mr. D. Light for their help with the FACS analysis.