An association between viral transformation and forssman antigen detected by immune adherence in cultured BHK21 cells

The changes in surface antigens accompanying the transformation of cells by viruses are of interest for a variety of reasons. The cell surface is intimately involved in a variety of cell functions and it is likely that the changes in morphology, contact inhibition, and control of growth and cell division which accompany transformation involve significant changes at the cell surface (Rubin, 1966). While such changes may be chemically complex, immunology offers a method for their characterization.

There is increasing evidence that transformation by viruses is accompanied by the appearance of new antigens (Sjogren, 1965). In some cases these antigens have been shown to be specific to the virus which induced the transformation. Much current evidence for the presence of new antigens on the surfaces of virus-transformed cells has been obtained by the use of transplantation rejection in syngeneic lines. With this system it has been possible to produce antisera specific to such antigens (Hellström & Sjögren, 1966), and it has been shown that transformation by polyoma virus results in the appearance on the surface of an antigen specific to this virus. The activities of such antisera are, however, low, and the detection of the antigen has been attended by difficulties. It has not been possible to detect the surface antigen associated with polyoma transformation in a non-syngeneic system. For these and other reasons it is necessary to explore the use of more sensitive, quantitative, methods.

The immune adherence method is exceptionally sensitive (Nelson, 1963). While it is not necessarily the most sensitive method available, it involves a relatively simple technique. A method of immunoassay by immune adherence is reported here which is applied to tissue culture cells in monolayer. With its aid the antigens appearing on the surfaces of the BHK21 line of hamster tissue culture cells (Macpherson & Stoker, 1962; Stoker & Macpherson, 1964) after transformation by polyoma virus have been studied.

Antisera were prepared by the injection of isolated plasma membrane fractions into rabbits. While the plasma membrane undoubtedly contains a variety of antigens, such fractions may be expected to be relatively enriched in surface antigens and the use of such material may increase the chances of obtaining active antisera. While rabbit antisera may be less specific than sera from hamsters in detecting virus-coded antigens, they permit the analysis of hamster antigens and are available in quantity.

Baby hamster kidney cells of the BHK21 line, clone C13 (Macpherson & Stoker, 1962) and polyoma-transformed derivatives of this line, PyJ, PyY and PyH 1, 3, 4, 3, 6, 7, 8, were obtained from Professor M. G. P. Stoker. Line Py Y had been isolated on glass (Stoker, 1962); the PyJ and PyH lines had been isolated more recently in agar suspension culture.

NIL cells were a clone, NIL-2E, derived by Dr I. A. Macpherson from NIL-2 (Diamond, 1967) cells. All the polyoma-transformed sublines used resulted from treatment with Toronto strain small-plaque virus.

The Rous virus-transformed cells used were derived from BHK21 clone 13 by treatment with the Schmidt-Ruppin strain of Rous sarcoma virus. Two such lines, designated SR1 and SR5, and revertant lines derived from them, SR-R1 and SR-R5, were obtained from Dr I. A. Macpherson.

Each cell fine was grown up for about thirty generations before storage at −70 °C. Individual aliquots were recovered at intervals for use and, after 1 month’s further growth, discarded. Cells were cultured on glass in modified Eagle’s medium containing 10% tryptose phosphate broth and 10% unheated calf serum, essentially as described by Stoker (1964). For mass culture, rotating Winchester bottles were used as described by House & Wildy (1965). Cells were collected by treating with 0·05 % trypsin plus 0·02 % EDTA (ethylenediaminetetra-acetic acid, disodium salt), centrifuged after the addition of 1 ml calf serum, and resuspended immediately in culture medium. Cell cultures were routinely tested for mycoplasma contamination (House & Waddell, 1967).

Human erythrocytes for immune adherence were collected from the same group O donor at monthly intervals, mixed with an equal volume of Alsever’s solution and stored at 4 °C. Immediately before use, the erythrocytes were centrifuged down for 5 min at 2000 rev/min and washed three times with saline albumin (138 HIM NaCl, 3·7 mM KCl, 8 mM Na₂HPO₄, 1·5 mM KH₂PO₄, 1·0 mM MgCl₂, 1·8 mM CaCl₂ and 0·2 % bovine serum albumin fraction V powder). Sheep erythrocytes were obtained from a commercial supplier (Burroughs Wellcome Ltd.).

Antisera were prepared by intravenous injection into rabbits of 10 μg protein per kg body weight. Three injections at 2-day intervals were followed after a further 7 days by bleeding from the ear. A minority of animals which did not give sera of adequate titre after the first series of injections were given a second course. The blood was allowed to clot overnight at 4 °C, freed from suspended cells, and heated for 20 min at 56 °C before storage in 4-ml vials at −20 °C.

The material for injection was plasma membrane fraction either from normal BHK cells (CPM) or from BHK cells transformed by polyoma virus (PyPM) obtained from cell homogenates as described below. In one case washed unfractionated microsomes (M) were used.

Antiserum was also prepared against sheep erythrocyte ghosts. These were prepared as described in Kabat & Mayer (1961) for ‘boiled sheep erythrocyte stromata’ and ultrasonically disrupted before injection in 1 mg quantities. Rabbit antiserum to sheep red cells was also obtained commercially and used without further treatment (Stayne’s Laboratories Ltd., High Wycombe, Bucks).

Cells were homogenized and fractionated to yield a product of characteristic equilibrium density and turbidity which by its immunological and enzymic properties has been identified as plasma membrane (Kamat & Wallach, 1965; Wallach, Kamat & Gail, 1966; D. Wallach, C. O’Neill & M. Stoker, in preparation) as follows. The cells were collected in a pellet by centrifugation at 1000 rev/min for 5 min and the volume measured. Pellets with volumes between 5 and 20 ml were washed in sucrose solution (0·25 M sucrose, 0·2M MgSO₄, 5 mM TRIS buffer pH 7·4) at 4 °C and resuspended in 5 to 10 times their volume of this medium before homogenization in a pressure homogenizer (Hunter & Comerford, 1961). This entailed exposure to an atmosphere of nitrogen at 1000 lb/in² in a pressure vessel (Artisan Industries Ltd., Waltham, Mass., U.S.A.) for 20 min with stirring. The cell suspension was then allowed to siphon out through a tube 1·7 mm diameter and EDTA added to a final concentration of 1 mM before centrifugation for 15 min at 16000 rev/min in the 40 head of a Spinco model L ultracentrifuge. The supernatant was collected and centrifuged for 45 min at 40000 rev/min. The pellet of microsomes resulting was resuspended by repeated passage through a hypodermic syringe in buffer (TRIS 10 mM, pH 8·6) to a protein concentration of 1–2 mg/ml and again centrifuged for the same time and speed. The pellet resulting was again resuspended in buffer of 1 mM concentration and centrifuged for the same time and speed. The final pellet resulting (M) was resuspended in 2 ml of 1 mM buffer containing 1 mM Mg²⁺ and dialysed for 2 h against the same medium. Volumes of the suspension (normally 1 ml) containing between 5 and 10 mg of protein were then added to centrifuge tubes containing 2 ml polysucrose (Ficoll, Pharmacia Ltd.) solution. This solution had a density of 1·094 at 20 °C and contained 1 mM TRIS pH 8·6, and 1 mM MgSO₄. The tubes were then centrifuged in the SW 39 head of the ultracentrifuge for 5 h at 39000 rev/min. The material remaining above the polysucrose layer as a compact band was collected, adjusted to a final protein concentration of 0·5 mg/ml, and stored at −70 °C (PM).

Protein was estimated by Lowry’s (Lowry, Rosebrogh, Farr & Randall, 1951) technique using Millipore filters (Bennett, 1967). All operations were carried out at 4 °C. All apparatus and media were maintained, as far as possible, sterile.

In some cases, antisera were exhaustively adsorbed with cells prior to testing. For use for adsorbing antisera, cells were grown in mass culture, and their volume measured in pellet form, as described above. The pellet was then washed in saline albumin free from divalent ions, and volumes of suspension equivalent to 1 ml of pellet were set out in screw-capped vials. These were centrifuged for 5 min at 2000 rev/min and the supernatants removed. To these prepared vials 1 ml of the appropriate antiserum was added and the cells gently resuspended. The vials were then tumbled for 30 min at 37 °C and the serum recovered by centrifugation for the same time and speed as before. In some cases guinea-pig kidney was used for adsorption. This was prepared as described in Cruikshank, Duguid & Swain (1965).

To titrate antisera by immune adherence, sparse cell monolayers were formed by adding 0·5 ml of a suspension of 40000 BHK cells/ml in culture medium to 20-mm coverglasses to which stainless steel rings 5 mm in height had been attached with a heated mixture of petroleum jelly (vaseline) and paraffin wax. Groups of nine coverglass cultures of this kind were assembled in 90-mm Petri dishes and incubated at 37 °C in an atmosphere of 5 % CO₂, 95 % air for 16–24 h-Antiserum was prepared in groups of 9 serial doubling dilutions of 0·2 ml saline albumin in methacrylate (perspex) dilution plates, and added to individual coverglass cultures after they had been washed three times with saline albumin. Cells were allowed to react with serum for 30 min with gentle agitation. The washing step was then repeated and the culture chambers filled to the brim with a suspension of washed human erythrocytes (0·25 % erythrocytes in saline albumin) containing 1 % preserved guinea-pig complement (Burroughs Wellcome and Co., London). A further 30 min were allowed for reaction. The culture chambers were then individually removed from the Petri dish, covered with a 50-mm plastic Petri dish lid so as to exclude air bubbles, and inverted. The inverted chambers were gently tapped on the bench, and the erythrocytes detached in this way allowed to fall to the bottom before examination of the cell layer under the microscope. The titre of the antiserum was recorded as that greatest dilution with clearly distinguishable haemadsorption. Both cultures and media were maintained at 37 °C throughout; medium changes were done with the cultures resting on a plate controlled by thermostat to this temperature.

For fluorescein labelling, 1 ml antiserum was added to 3 mg fluorescein isothiocyanate adsorbed on celite in the presence of an equal volume of 0·2 M carbonate-bicarbonate buffer at 20 °C and the suspension shaken for 5 min. An equal volume of saturated ammonium sulphate was then added with stirring, and the precipitate collected by centrifugation, redissolved in water and passed through a 10 × 1 cm column of Sephadex G25. The globulin effluent from the column was collected, adjusted to 0·5 % protein concentration, sterilized by filtration and stored at − 20 °C.

Sera were assayed by complement fixation using the method of Grist, Ross, Bell & Stott (1966). A suspension containing 2·5 × 10⁶BHK cells per ml was used as the antigen, in 0·1 ml quantities.

For cytotoxicity, the ⁵¹Cr method of Wigsell (1965) was used, 10 × 10⁶BHK cells in 1·0 ml culture medium were labelled by the addition of 20 μc ⁵¹Cr as sodium chromate (50–150 μc/μg Cr) in sterile saline, and incubated with gentle shaking for 30 min. The cells were centrifuged down, washed twice, incubated in 5 vol. of medium at 4 °C for 30 min, again washed, and finally resuspended at the original concentration. Serial doubling dilutions of antiserum in 0·10 ml volumes were prepared in Perspex dilution plates. To 0·10 ml of labelled cell suspension 0·10 ml of antiserum was added. After 5 min, 0·10 ml of fresh guinea-pig serum (50 % in saline) was added and the suspension agitated at 5-min intervals during 30 min incubation at 37 °C. The suspensions were then centrifuged for 5 min at 2000 rev/min and the supernatant 0·7 ml assayed for γ activity in a shielded crystal scintillation counter.

Human erythrocytes, in the presence of complement, adhered to BHK21IC13 hamster cells previously treated with specific antiserum. This adherence was most conspicuous and in dilutions of antiserum as great as 1/1000 hamster cells were covered with a dense mass of erythrocytes (Fig. 4). A visual count from photographs showed that between 100 and 200 erythrocytes could adhere to a single hamster cell.

Use of medium free from divalent ions resulted in no adherence. Substitution of sheep erythrocytes for the human erythrocytes normally used also abolished the reaction completely. Samples of erythrocytes from ten different human individuals gave essentially the same titres with the same antiserum. Storage for 4 weeks in Alsever’s solution did not affect the titres obtained, but the erythrocytes gave low titres after storage in saline. Medium free from protein resulted in reduced and variable reaction; the substitution of gelatin for albumin also reduced the reproducibility of the reaction.

When complement was omitted from the medium, no reaction occurred. When concentrations of complement varying between 0·5 and 5·0 % were used, no difference in the intensity of the reaction could be detected. When fresh guinea-pig serum was substituted for the preserved guinea-pig complement normally used, the proportion of reacting cells was not increased. A trace of adherence, both to cells and to substratum, occurred in the absence of antiserum. This was not reduced by the substitution of fresh guinea-pig serum for preserved complement. Absorption of the guinea-pig serum with BHK cells did not alter the reaction in any way. These results are summarized in Table 1.

Rarely a uniform, rather low, level of adherence was observed in the absence of antiserum, which involved all cells equally. This adherence was independent of the presence of complement, and thus is more properly termed haemadsorption. Medium from cultures showing this phenomenon invariably gave evidence of mycoplasma infection. Experimental infection (with M. hominis, M. ferment ans, and with M. arthriditis) produced a similar uniform haemadsorption. Treatment of such cultures with antisera specific to the mycoplasma strain used for infection produced a strong immune adherence reaction. Haemadsorption due to the presence of mycoplasma could readily be recognized by these characteristics, and cultures showing such haemadsorption were discarded.

The cells were cultured in medium containing bovine serum. When sparse cell cultures were treated with rabbit antiserum specific to bovine serum, immune adherence occurred to the substratum alone (Fig. 6). While bovine serum components had clearly adsorbed to the coverglass on which the cells were cultured, no evidence was obtained for the presence of bovine material, or of any other component of the culture medium, on the cells themselves.

Occasionally a proportion of the cells failed to react with erythrocytes even at high antiserum concentrations. On any one plate cells could be divided into two classes, those which reacted strongly and those which reacted little or not at all. Commonly the proportion of cells reacting was at least 50 % of the total. When cells were plated out at low densities and allowed to grow into colonies, the same variation was apparent between the cells forming a single colony and no difference could be detected between the colonies in this respect. Fewer cells reacted when the temperature was allowed to fall below 37 °C during the reaction and the difference appeared to depend on the physiological state of the cells.

The reaction was normally performed on sparse monolayers of cells covering approximately 10 % of the total substrate surface. When the number of cells plated was increased, patches of confluent cell sheet appeared in the centre of the coverglasses. Such patches showed total absence of reaction in immune adherence at any serum concentration. Figures 4 and 5 show different areas from the same coverglass culture, which was reacted in immune adherence (antiserum CM 8423) in the standard manner. Figure 4 represents an area near the edge of the coverglass in which the cells were widely separate, and shows a normally intense adherence. More than 100 cells are adherent to the single cell on the right. Figure 5 represents an area at the centre of the coverglass, which in this case was occupied by a continuous cell sheet. No adherence is evident. More than 50 cells can be distinguished, and less than 15 erythrocytes are attached. This inhibition of confluent cell layers was regularly observed with a variety of cell strains, both virus-transformed and normal, and with a variety of antisera. It was not observed when cultures infected with mycoplasma were reacted with antimycoplasma sera.

In the standard method, cells were plated 24 h before use in order to ensure full spreading. At 5 h after plating, spreading is incomplete, but the cells are at this time quite firmly attached to the glass. In one case the reaction was performed on cultures plated only 5 h previously. When cultures had been plated in numbers sufficient to produce confluent monolayers, it was found that all cells in contact with one another failed to react, and that reaction was limited to cells exposed at gaps in the monolayer, and to rounded cells overlying it. Sparse cultures plated at the same time before reaction showed adherence at a level only slightly reduced from normal. Thus the inhibition of confluent cells took less than 5 h to develop.

As the concentration of antiserum was increased the intensity of adherence fell off. This effect was only observed at concentrations above that at which a cytotoxic effect could also be detected by the chromium elution method. In one experiment, antiserum (PyPM 8940) at concentrations between 1:20 and 1:2560 was reacted with BHK cells. In this case the end-point proved to be 1:1280. After completion of the titration the cultures were washed, resuspended in trypsin, and individually plated out in culture medium. After 7 days incubation the plates were counted for colonies. In all cases where the dilution of the applied antiserum had been greater than 1:100 the colony counts represented 20 % or more of the applied cells. This is equivalent to the plating efficiency normally obtained with these cells. Where the dilution of antiserum applied had been less than 1:100 the colony counts were variable and low. A dilution of 1:100 corresponds to the greatest dilution at which any cytotoxic effect can be detected with this serum (see below). It may be concluded that the fall-off in intensity of the reaction observed at high concentrations of antiserum is due to the cell damage which occurs at these concentrations. Cells must be viable if they are to react.

With progressive reduction in antiserum concentration, the number of reacting cells fell off, though individual cells continued to react strongly. Eventually an end-point was reached. This reduction was not proportional to the concentration of antibody, but rather fell more or less abruptly as a limiting concentration was approached. In some cases the number of reacting cells fell several hundred fold over the space of two doubling dilutions. An end-point is illustrated in Figs. 7–9. Making use of this end-point, it was possible to perform a quantitative titration of the antiserum.

Polyoma-transformed cell lines derived from BHK2ijCig also reacted with antiserum in an essentially similar fashion. Transformed cells were, however, readily distinguishable from normal cells in immune adherence by the rounded form they adopted when reacting with erythrocytes. Transformed cells which had only partially reacted bore rings of erythrocytes and were thus very conspicuous. Similar cells on the same plate which had failed to react did not round up but were normally extended. Transformed cells, like normal cells, failed to react when confluent. They were not observed to differ from normal cells in any other aspect of the reaction and the reaction could be used for immunoassay with the same degree of reproducibility for either.

An antiserum CPM8938 was prepared by the injection of plasma membrane fraction from BHK cells and titrated by immune adherence on sparse BHK monolayers. The end-point occurred at dilutions varying between 1:2500 and 1:5000 on different occasions. Titres of this order were frequently obtained from a single course of injections of quantities of membrane corresponding to only 10 μg/kg body weight. In a few cases it was necessary to repeat the injection course.

Fluorescein labelled samples of this and other antisera were applied to both live and acetone-fixed BHK cells. In neither case was it possible to distinguish any specific staining on fully extended cells. Occasionally in these cultures rounded cells were seen, presumably in the course of division. Such cells, when not fixed, could clearly be seen to react with the stain. They appeared as rings of bright fluorescence and the effect was presumably due to the tangential view of the cell surface afforded by these cells. Normal rabbit serum labelled in the same way showed no such effect. Specific staining could be detected in this way at antiserum dilutions up to 1:10.

Antiserum CPM8938 was also assayed against BHK cells by complement fixation. Using reaction volumes of 0·1 ml, an average titre of 1:250 was obtained.

Antiserum was also assayed against BHK cells by cytotoxicity, using the elution of isotopically labelled chromium as a quantitative indicator of antibody activity (Wigsell, 1965). In this case antiserum PyPM894O was used. This antiserum showed an average titre of 1 : 1000 in immune adherence on BHK cells. When assayed by the cytotoxicity method the corresponding titre was 1:40. This figure is the antiserum concentration resulting in 50 % maximum elution; the greatest antiserum dilution showing detectable effect was slightly less than 1:100.

In summary, the sensitivity of the immune adherence method proved to be more than a hundred fold greater than that of immunofluorescence, 20-fold greater than cytotoxicity, and 10-fold greater than complement fixation.

Antisera prepared in rabbits were assayed for activity by immune adherence on various strains of BHK cells transformed by polyoma virus. When antisera against normal cells of the parent clone Cig (CPM) were reacted in immune adherence against either normal or polyoma-transformed cells the average titres in the two cases differed by less than a factor of two (Fig. 1). Titres of the order of 1 : 1000 were regularly obtained for a variety of BHK fines transformed by polyoma virus. Similarly, antisera against polyoma-transformed cells (PyPM) showed no difference between their reactivity to normal and transformed cells. Thus transformed and normal cells have, as is to be expected, common antigens.

Exhaustive adsorption of both types of antiserum with the cell line from which they were produced reduced their titres regularly to an undetectable level. Thus the trypsinization which is necessary in preparing the cells used for adsorption did not affect their effectiveness as adsorbing agents. Similarly, adsorption of several samples of CPM antisera with polyoma-transformed BHK cells of several different lines abolished their reactivity to normal cells. Transformed cells were as effective as normal cells in adsorbing the reactivity of antisera to normal cells.

Adsorption of PyPM antisera with normal cells, in contrast, reduced their titre against polyoma-transformed cells by a factor of less than 1:10. In many cases the titres of such antisera before and after adsorption in this way were not detectably different. The results of a representative experiment of this kind are shown in Fig. 2 in which antiserum PyPM 8940 was titrated against the normal cell clone and nine lines derived from it by transformation with polyoma virus. It can be seen that most of these transformed lines gave titres of the order of 1:1000, while the parent cell line gave no detectable reaction. None of these transformed lines gave titres less than 1:100. Antiserum PyPM 8940 was obtained after the injection of material from the polyoma-transformed line PyJ. Similar results were obtained with antiserum obtained after the injection of material from PyH6, and duplicate antisera prepared in different rabbits also gave the same result. Antisera prepared by the injection of transformed cells were, after adsorption with normal cells, specifically reactive to all the polyoma-transformed cell lines studied.

The reaction of PyPM 8940 (ads.) was also examined by the cytotoxicity technique, which is more precisely quantitative. In this technique there may be some variation between individual experiments in the amount of labelled chromium taken up by the cells. These variations can be discounted if the point of maximum slope on the curve is recorded as the end-point. As shown in Fig. 3, for PyPM 8940 this point corresponds to a dilution of somewhat less than 1:100; for PyPM 8940 (ads.) the corresponding figure is close to 1:50. The titre was thus reduced by less than a factor of two after exhaustive adsorption with normal cells. It can be concluded that more than half of the activity of such adsorbed antisera is directed against antigens specific to the virus-transformed cells.

While most of the transformed lines gave titres of the order of 1:1000 PyPM with (ads.) antisera, two lines, PyY and PyH8, differed in that they gave lower titres of the order of 1:100. Antiserum was prepared against PyY cells (PyYPM9859) and proved to have a titre of the order of 1:2500 on several different BHK lines, both transformed and normal. After exhaustive adsorption with C13 cells its titre with transformed cells was reduced to 1:100. The same low titre, of the order of 1:100, was found when it was reacted with each of the nine transformed lines studied previously, including PyY and PyH8. Thus these two cell lines also differ from the remainder in their capacity to induce antiserum. It was less easy to distinguish specific antigens in these two transformed lines.

Antisera against transformed cells, after adsorption (PyPM (ads.)), were also titrated against BHK cells transformed by Rous sarcoma virus. Two separately derived Rous-transformed lines gave titres of 500 or more with PyPM 8940 (ads.). Rous-transformed lines of BHK cells occasionally produce revertant colonies, which have the morphology of the normal cell and no longer grow when introduced into agar suspension. Revenant lines derived from the two Rous-transformed lines previously tested gave no detectable reaction with PyPM8940 (ads.). Their reaction with normal antiserum (CPM8938) remained intense. Thus transformed lines derived both from polyoma and from Rous virus have an antigen in common, and the presence of this antigen is dependent upon the maintenance of the transformed state.

Polyoma-transformed cells of another maintained line (NIL-2 (Diamond, 1967)) derived from hamster tissue were available. When these were tested with PyPM (ads.) antisera, they were found to give titres of the order of 1:1000. In this case, however, the untransformed parent line also reacted strongly and there was no distinction between transformed and normal cells. Both normal and transformed NIL cell lines reacted positively with PyPM (ads.) antisera.

Primary hamster embryo cells, subcultured on to coverglass cultures and tested after 24 h, reacted strongly with antisera to normal BHK cells (CPM). This reaction was effectively abolished by adsorption. When such cells were tested with PyPM (ads.) antisera, they were found to give titres of the order of 1:1000. Thus the antigen appearing in BHK cells after virus transformation is also present on hamster embryo cells.

These results are summarized in Table 2. They show that an antigen can be identified in hamster cells which is absent from the BHK21 line but reappears on transformation by either polyoma or by Rous virus. There is some quantitative variation in the amount of antigen present. Revertants appearing in Rous-transformed lines, which no longer show the transformed morphology, have also lost the antigen.

When PyPM antiserum was reacted with washed sheep erythrocytes a significant degree of haemolysis was observed which did not occur with the control CPM antiserum. Antiserum induced by boiled sheep erythrocyte ghosts reacted with titres of over 1:1000 against several fines of polyoma-transformed BHK cells, but showed no trace of reaction with normal BHK cells. A sample of PyPM (ads.) antiserum was further adsorbed with guinea-pig kidney. This treatment abolished its reactivity. Commercial antiserum to sheep erythrocytes (stated to have a haemolytic titre of between 1:3000 and 1:4000) was titrated against polyoma-transformed BHK cells (PyH6) in immune adherence, and the titre was found to be of the order of 1:100000 (varying between 1:64000 and 1:128000 in different experiments). No reaction was detected with normal cells at dilutions greater than 1:20. The antigen appearing on BHK cells after transformation is present on sheep red cells even after boiling. It is also present on guinea-pig kidney and will react with antisera prepared in rabbits.

Immune adherence has been shown to be a simple, rapid, and effective way of detecting antigens on the surfaces of individual cells. Using sparse monolayers on coverglasses, an exceptionally sensitive assay of serum activity can be performed without difficulty. Individual assays performed on different days did not normally differ by more than one doubling dilution, and replicate assays performed at the same time showed excellent agreement. The method is some twenty-fold more sensitive than cytotoxicity and a hundred-fold more sensitive than immunofluorescence. The observation that the cells remain viable and may be recovered for further growth after testing gives reason for confidence in the method as an assay of surface antigens and may prove to be of practical use. The method bears similarities to the mixed haemadsorption method (Barth, Espmark & Fagraeus, 1967), which differs in that two special antisera, specific antiglobulin and antiserum specific to red cells, were used to provide the link between the antibody and the erythrocyte. These authors quoted a sensitivity, relative to the chromium elution method, rather greater than that found here for immune adherence.

In most cultures, a proportion of cells, of the order of 50 %, failed to react even at high antiserum concentrations. While this proportion varied between individual cultures, the antiserum concentration at which reacting cells were no longer apparent remained relatively constant, and for this reason it was possible to perform quantitative assays.

The reaction was routinely performed on sparse cell monolayers, in which attached cells occupied approximately 10 % of the total surface area. When the number of cells plated was increased, the proportion of reacting cells fell off. When the number of cells plated was such that confluent monolayers were formed, the proportion of reacting cells fell to zero. It is apparent that the reaction is inhibited at high cell densities. It is in principle possible that this inhibition could be due to the accumulation of some substance secreted by individual cells at a constant rate, and in consequence accumulating in regions of high cell density. However, experiments performed on confluent monolayers as soon as was practicable after plating showed the same inhibition, and the effect must therefore depend directly on cell contact. It appears to be a contact inhibition of adherence, and thus has analogies with contact inhibition of movement.

The abrupt fall-off in the reaction as limiting antibody concentration is approached also requires explanation. It is known that a suspension of antigen in equilibrium with its antibody shows at least some degree of proportionality between the amount of bound antibody and its concentration, until saturation is reached. It has been observed here, however, that the number of red cells bound to each fibroblast may fall from a maximum of over 100 to near zero in the space of only two doubling dilutions. A concentration effect of this general type would be expected if the reaction required that two molecules of antibody should bind at adjacent sites. This has been shown to be the case for the lysis of red cells by IgG antibody (Humphrey, 1967), and while it is not known if immune adherence follows the same laws as haemolysis, both involve the binding of complement. However, this hypothesis alone seems inadequate to explain the magnitude of the effect observed. Furthermore, the injection schedules used in these experiments favoured the production of IgM, which will cause haemolysis as a single molecule. It therefore appears likely that bound antibody has some secondary effect, in addition to its function as a site for immune adherence.

Contact inhibition (Abercrombie & Heaysman, 1954) is the phenomenon whereby a cell will not move over its fellows. It has been observed as a cessation of pseudopodal movement (referred to as ‘ruffled membranes’) at the advancing end of the cell when it comes into contact with a neighbouring cell. When a cell is completely surrounded, all pseudopodal movement of this sort ceases, and the cell becomes stationary. If pseudopodia disappear from the periphery of the cell it is possible that they may also disappear from the remaining exposed cell surface. It is suggested here that immune adherence to tissue cells only occurs at the tips of pseudopods, and that the inhibition of immune adherence in confluent cell monolayers is a consequence of the withdrawal of all pseudopods from such monolayers.

There are a variety of reasons why the close contact of cell surfaces should require the performance of relatively large amounts of work. One of the simplest possibilities is that the viscous drag of the medium which must be drained from the intercellular gap may be sufficient to severely decrease the speed of apposition. It is also possible that the electrostatic repulsive forces occurring between surfaces of like charge may indefinitely limit apposition to a minimum distance. The total force required to produce contact would in either case be reduced to a minimum if only the small area of surface at the tip of the pseudopod was involved. It may be relevant in this connexion that red blood cells, which are exceptionally smooth, are also mutually non-adhesive to an exceptional degree. It appears very probable that adhesion cannot be established between tissue cells by the apposition of absolutely smooth surfaces, and that bonds of any sort are generally initiated at pseudopods.

Marcus (1962) has observed that haemadsorption in HeLa cells infected with Newcastle disease virus is localized at pseudopods (termed ‘microvilli’). He found that 1–5 microvilli attached to each red cell, and that 30–40 red cells bound to each HeLa cell. From these figures it may be concluded that up to 200 microvilli are present on the upper surface of each HeLa cell. This conclusion is now supported by the observations of Fisher & Cooper (1967), who examined the upper surfaces of HeLa cells in the electron microscope after shadowing, and found that each cell bore just over 200 microvilli. Each microvillus was between 2 and 4 μ, in length and between 800 and 1200 Å in diameter. They also found, by vertical sectioning of cultures on plastic films, that the initial stage of adhesion of cells in the process of settling on to a substratum involved the formation of fine pseudopodia of this type. The figures found by these authors for the number of pseudopods on the surfaces of HeLa cells agree rather closely with the figures reported in this paper for the number of red cells bound to BHK cells at optimum antibody concentrations.

It has been observed (Carey & Pettengill, 1967) that the application of antibody to HeLa cells results in cessation of pseudopodal movement. Using time-lapse cinematography, they found that, on application of antibody, pseudopods straightened out and that the movement of withdrawal and extension previously observed no longer occurred. Thus antibody has a profound effect on pseudopodal movement. It may be that this is the reason why the intensity of immune adherence falls off so rapidly with falling antibody concentration. It is therefore proposed that, at antibody concentrations above a certain limiting level, pseudopodia are altered in some way so as to render them available for immune adherence. This alteration may be simply the stabilization of existing pseudopods, so that they are not withdrawn and therefore accumulate on the surface of the cell. Alternatively, the formation of pseudopods may be actively stimulated, or the mechanical properties of existing pseudopods may be altered. It is not at present possible to distinguish between these alternatives. An investigation is in progress into the effects of antibody on pseudopodia by electron microscopy.

Whatever the mechanism of immune adherence may be in this system, it has proved to be a sensitive and reproducible immunoassay. Antisera against BHK membranes gave essentially the same titres with several strains of BHK cells, and variations between individual determinations were normally not more than one doubling dilution.

Both normal and several clones of virus-transformed cells gave similar titres, and thus no loss of surface antigen as a consequence of transformation has been detected. Antisera against the membrane of transformed BHK cells, however, after adsorption with normal cells, did show a clear specificity. Such adsorbed sera showed consistently very high titres with transformed cells, while their titres against the parent, normal, BHK line had been abolished. This specificity was found in more than ten transformed lines. In two lines the specificity was present in relatively smaller amount, but in all cases the titres exceeded 1:100. Transformation, whether by polyoma or by Rous virus, produced a similar result. All transformed lines bore an antigen not found in the parent line.

This antigen must be of the Forssman type. Antisera prepared against transformed cells also reacted with hamster embryo cells and with sheep red blood cells, in which Forssman is known to be present. Transformed cells could also be very easily distinguished from normal BHK cells by means of anti-sheep red cell serum.

The presence of Forssman antigen in this cell line offers an additional criterion of transformation which is important because unlike colonial morphology or growth in suspension, it can be detected without the need for cell division. The loss of this antigen which accompanies the reversion of Rous virus-transformed cells gives ground for believing that it is closely associated with transformation. Individual transformed BHK cells can be readily detected with this antiserum. In view of the very high titres obtained with immune adherence, it is to be expected that they will also be detected by immunofluorescence.

Fogel & Sachs (1964) reported the appearance of Forssman antigen on hamster embryo cells after one subculture. They also found it to be present on mouse cells in tissue culture, and it appears unlikely that the appearance of this antigen as a concomitant of transformation can be a general phenomenon. It does however offer the possibility of following more closely the time course of transformation in the BHK cell line and experiments are in progress with this object.

I am much indebted to Professor M. G. P. Stoker for his support and helpful criticism. I also wish to thank Dr I. A. Macpherson for kindly providing cell lines, Dr R. J. Fallon for providing antimycoplasma sera, and Mr W. House for providing cells infected with known strains of mycoplasma. Miss C. Laird gave very competent technical assistance.