Tests by Tissue Culture Methods on the Nature of Immunity to Transplanted Skin

When skin is grafted from one human being or one rabbit to another, a ‘defence’ mechanism is called into action which leads, in due course, to the complete destruction of the foreign grafted tissue. A quantitative analysis of the phenomenon (Gibson and Medawar, 1943; Medawar, 1944, 1945, 1946a, b) has shown that it conforms in broad outline with a reaction of actively acquired immunity. In other words, a human being at first submits to and later recovers from an ‘attack’ of foreign skin in much the same way as he recovers from an attack of measles: natural (i.e. ready-made) immunity is absent or ineffective; resistance develops in the course of exposure; and recovery is followed by a refractory or ‘immune’ state. This interpretation is not new, for certain critical tests bearing upon it were made by Peyton Rous in 1910; but it needs emphasizing, because Loeb (1921, 1930, 1945) has for many years denied that immunity in this technical sense has any important part to play in the organism’s reaction against tissue grafted to it from other members of its own species.

More specifically, a skin homograft builds up a systemic reaction against itself at a rate which varies with the antigenic relationship between donor and recipient and with the quantity of foreign tissue that is grafted. A second homograft, transplanted from the same donor to the same recipient when the reaction against its predecessor is complete, survives for a few days in a vegetative condition in which cell division is partly or wholly suppressed, and then undergoes accelerated breakdown (Medawar, 1946a).

Skin transplantation immunity is thus easily recognizable in vivo by the regression of foreign homologous grafts and the refractory state which follows it. If it conformed in detail, as it does in main outline, with the pattern of immunity created by bacterial and other crudely foreign antigens, then one might expect to find some immune body in the tissue or body fluids of an immunized animal which is inimical to the growth in vitro of the tissue responsible for generating the immune state. Many years’ experience of tumour transplantation has led to the belief that no such factors exist, and it is hard to be convinced by the few tests of the hypothesis that have been said to have had a positive outcome (cf. Phelps, 1937). The purpose of the present paper is to reinvestigate the problem systematically, by taking advantage of newly devised methods that make possible the cultivation in vitro of adult skin.

The tests to be described in this paper consist in the cultivation of adult rabbit skin epithelium in the presence of serum, tissues, and tissue extracts derived solely from a rabbit heavily and specifically immunized against it. Adult skin epithelium will proliferate and migrate promptly and vigorously when cultivated by flotation upon a stirred and properly aerated serous fluid medium (Medawar, 1948a). The advantages of fluid culture are its ease of execution, the rapidity with which cell division and migration begin and proceed, and the fact that the cells preserve their normal histological appearance and functional activity during cultivation (Text-figs. 1-5). Its only disadvantage is that, since migratory overgrowth or self-encystment takes the place of outgrowth as it normally occurs in vitro, histological analysis must, of necessity, turn upon the use of stained serial sections.

Each single test has made use of an independent pair of rabbits: a donor (D) and a recipient (R). In the initial immunizing operation, eight ‘pinch’ grafts, each about 10 mm. in diameter and together weighing 0·45—0·55 g., were transplanted from the thigh of D to the chest of R (Medawar, 1944, 1945). (In many trials the recipient was grafted on a second occasion from the same donor.) When the complete and long-standing necrosis of these immunizing grafts gave direct evidence of the completeness of immunization, but in no case earlier than the fifteenth day following their transplantation, a number of very thin 3 mm. × 3 mm. skin squares (minute ‘Thiersch’ grafts) were cut from a lightly vaselined area on D which had been toughened and rendered slightly hyperplastic by shaving 3 or 4 days beforehand (see Medawar, 1948a). The skin squares were thereupon cultivated for 4 or 8 days by flotation, raw side down, on serum freshly withdrawn from the median ear artery of R in dilatation (Medawar, 1946b), the serum being used either plain, or as an extractive for R tissues, or, most commonly, as a vehicle for the simultaneous cultivation of a variety of tissues from R. Although the D skin expiants usually coalesced with at least some of the R tissue fragments cultivated with them, it was important in some special trials to make quite certain that coalescence and intimate tissue union should take place. In these special trials, therefore, R tissue fragments were glued to the dermal sides of the D skin expiants with citrated and recalcified R plasma before cultivation began (see Medawar, 1948a).

Two culture methods were used: cultivation for 4 days in roller tubes, in a gas phase of air, or for 4 — 8 days in ‘rocker flasks’ of 250 ml. capacity in a gas phase of air-oxygen mixture (65 — 70 per cent. O₂). The design and use of the appropriate apparatus has been described in full elsewhere (Medawar, 1948a). The rocker-flask cultures were left undisturbed in the incubator for the run of each experiment. With roller-tube cultures the fluid part of the medium was replaced after the second day—with freshly withdrawn serum, if serum alone had been used, but with serum stored for 2 days in the refrigerator if the recipient animal had been killed at the beginning of the experiment to provide ‘immune’ tissue expiants. The main purpose of using high-O₂ rocker-flask cultures was to make certain of the continued functional survival of R spleen- and lymph-node tissue. Neither grows well, if at all, at low oxygen tensions (Parker, 1936, 1937).

Controls were achieved in two ways: either by the growth of D skin in normal homologous serum, tissues, and tissue extracts—i.e. in a medium derived from a rabbit immunized neither against D cells nor any others; or by cultivating R skin in the same vessel as D skin in media which, being derived solely from R, should in theory be inimical only to the growth of cells from D. (The R skin expiants were distinguished from their companions by being cut to a triangular shape.) Controls of these two types are adequate to reveal any non-specific action by the ingredients of the culture medium. It should be noted that a third type of control, the cultivation of skin from a second rabbit in media derived from a recipient specifically immunized against a first, is not acceptable; for although the skin immunity reaction is strongly donor-specific (Medawar, 19460), the second donor might well happen to share with the first some antigens not also present in R. Some degree of immunity would in that event be directed against it.

In all except the earlier roller-tube experiments (Table 1, exps. 1-16), 9 volumes of the chosen culture medium were mixed before use with 1 volume of streptomycin solution (200 u./ml.) in Ringer. In the rocker-flask tests (Table 2), 8 volumes of the culture medium were mixed with 1 volume of streptomycin solution and 1 volume of 5 per cent, glucose solution in water. With these exceptions, and with the substitution of Krebs-Ringer-bicarbonate for serum in two specific trials (Table 1, exps. 34, 35), the culture medium for D skin was derived wholly from the animal immunized against it.

Three separate routine tests and analyses have been made on roller-tube expiants of donor skin in immune media: (i)a test of their continued survival; (2) a measure of the frequency of cell divisions in the epidermis; and (3) a measure of the migratory activity of the epithelium as a whole. Rocker-flask cultures in a high-O₂ gas phase were used mainly for the histological analysis of donor skin and immune mesenchymal tissue growing in intimate union.

So much of importance turns upon whether donor skin survives cultivation in immune media that the diagnosis of survival has not been allowed to rest on histological evidence alone. Instead, each D expiant was duplicated; one was reserved for histological analysis and mitotic counting (see below), and the other tested for its continued survival by the simple process of grafting it back to a large raw area on the animal from which it originally came. Not less than two such tests of survival have been done for each type of medium to which the roller-tube expiants were subjected (Table 1, col. 5). The test has been described in full in earlier papers (Medawar, 1947, 1948a).

The other member of each pair of D skin expiants was rinsed in Ringer, fixed in Bouin’s solution, and embedded in wax. Alternate sections from strips of to taken at 5 evenly spaced vertical levels through the block were stained with Hance’s variant of Hcidenhain’s haematoxylin. All the mitoses in the 5 alternating 8μ sections from the strip of median level were counted and averaged, the figure entered in column 6 of Table 1 representing the mean number of mitoses per median 8μ section divided by the width of the explant in millimetres.

A median section stained with Ehrlich’s haema-toxylin was marked for degree of epidermal migration in accordance with the scale which follows (Table 1, col. 7):

* Overgrowth incipient: the ‘shoulders’ of the explant rounded off.

* * Overgrowth about 1/3rd complete.

* * * Overgrowth about 2/3rds complete.

* * * * Complete scif-encystmcnt (cf. Text-figs, 1, 3, 4, 5).

The results of 37 tests are summarized by Table 1 and amplified by the notes given below, which correspond to the references in column 4. The use of streptomycin at a final concentration of 20 u./ml. in the culture medium was adopted from exp. 17 onwards. The use of minute ‘Thiersch’ grafts as expiants (mean diameter after fixation: 2·9±0·1 mm.) was not adopted until exp. 14; small ‘pinch’ grafts, much thicker at the centre and somewhat broader (40+02 mm.) had been used until then, and this probably accounts for the fact that none of the expiants quite achieved complete self-encystment (* * * *) in 4 days. The total weight of the immune tissue expiants added to the culture medium in various experiments did not exceed 4 mg.; each was cut into a cube not exceeding, and so far as possible not much less than, mm. in length of side. In all the experiments described in this paper, the lymph-node expiants were taken from the axillary node receiving the lymphatics from the side of the chest carrying the immunizing homografts.

(a) The use of leucocytes (exps. 17, 20). Leucocytes were present in the immune serum at times their physiological concentration, having been separated from 6 to 8 ml. citrated whole blood by repeated slow spinnings and then resuspended in 3 ml. serum. The serum preserved a strong cellular opalescence during cultivation, though a majority of the leucocytes ‘agglutinated’ and formed compact cell clusters.

(b) Adjuvant immunization with chicken plasma (exps. 18, 19, 21, 24). The recipients were injected with 2 ml. chicken plasma 12 days before and 4 days after the transplantation of the immunizing dose of homografts, the first injection being intradermal and the second intra-venous. The antigen dilution titre of precipitins in the R serum exceeded 1,000 when the culture tests were begun 12 days after the second injection.

(c) Tissue extractions (exps. 21, 24, 34, 35). About 1 g. of each type of tissue was chopped into a fine mush with scissors and extracted with 10 ml. serum (exps. 21, 24) or 10 ml. Krebs-Ringer-bicarbonate (exps. 34, 35). In each case 2 7 ml. of the supernatant fluid, clarified by spinning, served as the culture medium.

(d) ‘Reverse’ tests (exps. 25, 26). The skin of the immunized recipient was cultivated in serum and tissue fragments derived from its donor.

(e) Defibrinated blood (exps. 29, 30, 31, 36, 37). It proved best to dilute the defibrinated whole blood with 4—5 volumes of its own serum.

(f) Anaerobic cultivation (exps. 32, 33). These experiments, though entered here for ease of reference, made use of the anaerobic culture apparatus described by Medawar (1947), and not of roller tubes. In this apparatus the expiants were incubated for 4 days at 38° C. in 5 ml. serum under an atmosphere of H₂ from which the last traces of O₂ had been removed by catalysis with heated palladium (indicator: methylene blue). Skin epithelium incubated anaerobically neither moves nor proliferates in any degree; hence the zero entries in the last two columns. The survival test shows, however, that in immune serum as in normal serum (Medawar, 1947), the skin cells remain alive and resume normal growth and activity when oxygen is restored to them.

(g) Adjuvant immunization with a cellular antigen (exps. 36, 37). A fine suspension in Ringer of about too mg. adult mouse spleen, lymph node, and submaxillary gland fragments was injected intradermally and intraperitoneally into R on the day of its receiving the immunizing grafts from its homologous donor. In exp. 37 the heterologous cell suspension was, in addition, injected intradermally into the graft donor area of D immediately before the immunizing grafts were cut from it for transplantation to R.

Special interest attaches to an additional and independent series of rollertube cultures in which fragments of ‘immune’ tissue were glued on to the dermal surfaces of the skin expiants before cultivation began (see above). Text-figs. 4 and 5 illustrate the complete encapsulation of lymph-node fragments by skin epithelium that has migrated from the donor expiants. In Text-fig. 5 the very pronounced thickening of the epibolic skin epithelium in the immediate neighbourhood of the encapsulated explant shows that, so far from its having had any inhibitory action, the node tissue seems actually to have stimulated the growth of the adjacent epithelial cells. Text-fig. 3 in low power and Text-fig. 8 in higher power illustrate complete confluence between donor skin epithelium and epithelium from the immunized rabbit’s kidney medulla. It is hard to tell where the one ends and the other begins.

The tests using lymph-node expiants (and others using spleen) are not decisive, because cultivation in a gas phase of air is not adequate to maintain the complete functional survival of tissues of these two types. The experiments to be described in the next section were done mainly to repair this shortcoming.

The cultures to be described here were done in 250 ml. gas perfusion rocker flasks containing the following culture ingredients: (a) 4·15 ml. immune serum containing streptomycin and glucose at final concentrations of 20 u./ml. and 0·5 per cent, respectively; (6) two square or rectangular donor skin expiants of the same size as those used in roller tubes; (c) two skin expiants from the recipient, cut to triangular shape for ease of identification and serving as controls against the possibility of any non-specific action by the ingredients of the medium; and (d) up to 88 mg. immune spleen or lymph-node tissue (mainly the former) cleanly chopped into cubical fragments of the size habitually used in tissue cultivation.

The flasks were perfused with one measured volume of filtered cylinder 0₂ at the timed rate of too ml. per minute, giving a final concentration of 65-70 per cent. O₂, and then set to rock for from 4 to 8 days without further attention in a dry-air incubator at 38° C. At the end of the experiment the pH of the culture medium was tested with the glass electrode and the cultivated tissues were fixed in formal-HgCl₂ for histological analysis. Independent survival tests were done in only two cases, with positive results. Both spleen and lymph-node fragments showed obvious proliferation under these conditions of culture (Text-figs. 6, 7), but the results from the cultivation of spleen were characteristically variable (cf. Parker, 1936, 1937). In most cultures the medium became strongly opalescent with mononuclear cells that had migrated from the spleen expiants during the first day or two of cultivation. The occurrence or failure of this response was in no way correlated with the degree of survival of the spleen expiants as judged by their histological appearance at the end of the experiment. Cells migrating from the lymph-node expiants showed a strong tendency to infiltrate the dermis of donor and control skin expiants alike and to bring about some degree of collagen dissolution.

The behaviour of the skin expiants was also variable: there were occasionally greater differences among the members of the two pairs than between them. For this reason many more cultures (Table 2) had to be done than would otherwise have been necessary. In high-O₂ media the migratory activity of the skin epithelium was far less pronounced than with cultivation in a gas phase of air, and complete self-encystment was achieved more slowly and less often. Cell division, on the other hand, was rapid and led to the formation of a deeply stratified epidermis (Text-fig. 2).

The cultures which were analysed are set out in summary in Table 2. There was no appreciable difference between the experimental cultures and the controls run in media of autologous origin.

In the experimental group the donor skin expiants were quite clearly superior to their controls (i.e. the skin expiants taken from the immunized recipient itself) in exps. 15 and 18, and clearly inferior in exps. 12,14, 22, and 23. (In exp. 14, as a result of infection, the growth of both skin pairs was very poor.) There was no ground for supposing that these differences were anything but fortuitous in origin.

There was almost invariably some degree of union between the skin expiants and the fragments of spleen or lymph-node tissue cultivated with them. Text-fig. 9 illustrates a particularly intimate union between strongly proliferating donor skin epithelium and an immune lymph-node fragment almost wholly encapsulated by it. Mitotic figures, everywhere abundant, were not less frequent in the epibolic epithelium abutting immediatelyagainst the node tissue. In short, the entire series of experiments affirms that the mitotic and migratory activity of skin epithelium was in no way influenced by cultivation in media, and in intimate association with tissues, derived from a rabbit heavily immunized against it.

It has already been established (Medawar, 1946b) that the grafting of skin from one rabbit to another does not elicit the formation of red-cell isoagglutinins. It might conceivably elicit the formation of agglutinins for cell suspensions of donor skin epithelium itself; and to complete the present series of in vitro tests the possibility was examined as follows.

About I cm.² of the thinnest possible shavings from the vaselined hyperplastic skin of a donor rabbit was incubated for one hour in a Seitz-filtered 0·5 per cent, solution of commercial trypsin powder in Ringer-bicarbonate at pH 7·6 (Medawar, 1941). The shavings were thoroughly rinsed in Ringer, and from each one the dermal layer was lifted off with fine forceps, leaving behind a pure epidermal sheet which transplantation tests have shown to be still living (Billingham and Medawar, 1948). The sheets were flattened down on their vaselined cuticular surfaces and then smoothly and firmly scraped to separate the cells of the deeper layers from those of the cuticle. The clumps so formed were rendered into a fine suspension in Ringer by sucking them repeatedly in and out of a fine-bore pipette. The suspensions were then mixed in agglutination tubes with or 1 volume of serum from a rabbit specifically and effectively immunized against the donor skin. Five such tests, run for 3-6 hours at body temperature and overnight at room temperature, were uniformly negative in outcome, and the cell suspension which in due course settled in the bottom of the agglutination tubes could be evenly redispersed by light tapping or shaking.

The presence of precipitins for skin extracts was tested simultaneously by the ring method, a water-clear extract being secured by freezing thin skin slices with CO₂ snow, cutting them into 15 μ sections on a freezing microtome, grinding the sections so cut in a mortar under Ringer, and then spinning down the collagenous and cuticular debris. No ring of precipitation developed between immune serum and the supernatant skin extract, and the subsequent mixing of the two solutions with further incubation produced no turbidity perceptible by direct or oblique illumination.

The experimental results make it clear that immunity to foreign homologous skin grafts is not demonstrable by an in vitro reaction of any type described in this paper. Notwithstanding statements that have from time to time been made to the contrary, they justify the provisional conclusion that the occurrence of free serum antibodies is not a sufficient explanation of the destruction of foreign-homologous tissue. The tests involved the cultivation of not more than 5 mg. donor skin for up to 8 days in so much as 15 ml. of serum and 88 mg. of mesenchymal tissue derived solely from a rabbit specifically and demonstrably immunized against it. Within these limits the experimental results are quite decisive: skin epithelium survives, moves, and proliferates with unaffected vigour in these presumptively immune media.

The results are consistent with those of Harris (1943), which have not yet been fully reported on; and, also, with what is known of the relationship between the regression of tumour homografts and the occurrence in their hosts of antibodies revealed by complement fixation. Actively induced complement fixing iso-antibodies appear in the sera of rabbits bearing a variety of transplanted tumours (Friedewald and Kidd, 1945) and they react in vitro with the fine particulate matter thrown down from saline tissue extracts by spinning at about 20,000 g for 2—3 hours. An antibody elicited by at least one such tumour, the Brown-Pearce carcinoma, has since been shown to be specific to the tumour in question (MacKenzie and Kidd, 1945; Kidd, 1946). The lack of correlation between tumour regression and specific antibody titre makes it clear, however, that ‘factors other than the specific antibody are probably responsible for regression of growth in the majority of cases’ (Kidd, 1946).

Two recent attempts have been made to elicit the formation of iso-antibodies by injecting extracts or homogenates of foreign homologous tissues together with toxins of known antigenic power (Schwentker and Comploier, 1939; Hecht, Sulzberger, and Weil, 1943). In neither case was the homologous tissue by itself effective.

Four possibilities that indicate the limitations of the present preliminary analysis may now be briefly discussed. First, it is possible that cultivated tissues will survive for long periods, but not indefinitely, in media derived from animals immunized against them. What is known already about the tempo and vigour of the immune reaction in vivo gives one no ground for supposing this to be true. A second possibility is that the dosage relationships between the cultivated tissues and their media were inappropriate to the demonstration of an immune reaction. When a skin graft weighing about 0 · 05 g. is transplanted to a 2 kg. rabbit, it ‘competes’ with 40,000 times its own mass of native tissue. The corresponding dosage ratio for the tests in vitro reported here has not exceeded 20 at the outside. This second possibility will therefore be investigated further, though it depends upon an interpretation of the immune reaction—in terms of a theory of ‘competition’—which is supported only by perhaps remote analogies with the behaviour of microorganisms.

A third possibility, which preliminary tests suggest to be the one most worthy of further inquiry, is that immune media affect skin epithelium in vitro only in such a way as to expedite its breakdown on subsequent grafting. Woglom’s (1933, 1937) experiments with transplanted rat sarcomata support this view, and Kidd (1946) has found that a preliminary incubation of Brown-Pearce tumour tissue in media containing its specific antibody reduced or suppressed its growth on transplantation. Finally, it may be that the hypo-thetical antibodies are cell-bound, and that an accessory immune mechanism is needed to bring them into action, viz. the destruction of the host’s antibody containing cells in the immediate neighbourhood of the graft. (The regression of skin grafts is accompanied by the wholesale destruction of such native lymphocytes and other leucocytes as may have penetrated them: cf. Medawar, 1944.)

The biological significance of the tissue homograft reaction deserves some mention. It is widely believed that tissue transplantation, like blood transfusion, is an act that has no counterpart in nature. This is clearly untrue: the mammalian foetus is a tissue homograft, though it is normally protected from the consequences of that fact by its strictly independent circulation. (The behaviour of skin homografts in the anterior chamber of the eye is to some extent a model of the mother-foetus relationship: such grafts, if not vascularized, are not destroyed: cf. Medawar, 19486.) It is at present thought likely (cf. Penrose, 1946; Kalmus, 1947) that a variety of foetal and placental abnormalities are to be attributed to immunological incompatibilities between foetus and mother rather than, for example, to endocrine disorder. If this view is correct, the interpretation of some forms of foetal abnormality is likely to turn on the analysis of the tissue homograft reaction.

Text-figs. 1-5 are from photographs taken by Mr. D. A. Kempson. The cost of the experimental animals used in this work was met by a grant from the Medical Research Council; of the special apparatus, by a grant from the Department of Plastic Surgery, Oxford University.