A HUMAN CHORIOCARCINOMA XENOGRAFT IN NUDE MICE; A MODEL FOR THE STUDY OF ANTIBODY LOCALIZATION

The successful development of the concept of linking cell-killing agents to tumour-specific antibodies will be largely determined by the extent to which the antibodies are preferentially localized in the malignant tissue. A xenograft of human choriocarcinoma (CC3) has been established in nude mice, and the relative distribution of affinity-purified specific antibodies to human chorionic gonadotrophin has been compared with that of nonspecific antibodies from the same species. Treatment of the nonspecific antibodies with ammonium thiocyanate appeared to be important to ensure that the distributions in normal nude mice were equivalent. Specificity indices, derived from the comparative distributions of isotope activity in the tumour and lung of labelled specific and nonspecific antibodies, ranged between 1-3 and 2 0. THERE HAS RECENTLY BEEN renewed interest in the use of antibodies as carriers for diagnostic agents and anti-tumour agents. A primary consideration is the extent to which the antibodies are retained in the malignant tissue, compared with their concentration in other, non-target tissues and body fluids. If a specific concentration of antibody can be achieved in the target tissue, further studies are needed to determine whether the antibodies are internalized into the malignant cell or retained at or near the cell surface since this would influence the nature of the conjugates required for chemotherapy. Antibodies may be directed at secreted tumour products as well as at membrane-bound antigens. Investigations of human colonic-cancer xenografts in hamsters have shown that 3-40 of an injected radioactive dose of iodinated, affinity-purified, specific anti-body to carcinoembryonic antigen (CEA) remained per gram of tumour after 8 days. Nonspecific labelled antibody cleared more rapidly from the tumour, giving a specific: nonspecific ratio of 7.72 + 1.47 (Primus et al., 1.977). In nude mice carrying human 10 colonic-cancer xenografts, optimal visual contrast between the tumour and normal tissue was seen at Day 3 by scanning animals given 2 jig of radio-labelled anti-CEA antibodies diluted with 200 ug of normal goat y-globulins (7S fractions of both). Localization ratios of 4.5:1 were obtained (Mach et al., 1974). It has been clearly shown by postoperative investigations in humans that antiCEA antibodies, whether intact or as fragments, can localize preferentially in tumour tissue as compared to normal tissues (Mach et al., 1980) but the actual recovery of antibody from the tissue is disappointingly small. After 6 days it is about one-thousandth of the administered dose, assuming that the radioisotope remains largely associated with the protein. …

Summary.-The successful development of the concept of linking cell-killing agents to tumour-specific antibodies will be largely determined by the extent to which the antibodies are preferentially localized in the malignant tissue. A xenograft of human choriocarcinoma (CC3) has been established in nude mice, and the relative distribution of affinity-purified specific antibodies to human chorionic gonadotrophin has been compared with that of nonspecific antibodies from the same species. Treatment of the nonspecific antibodies with ammonium thiocyanate appeared to be important to ensure that the distributions in normal nude mice were equivalent. Specificity indices, derived from the comparative distributions of isotope activity in the tumour and lung of labelled specific and nonspecific antibodies, ranged between 1-3 and 2 0. THERE HAS RECENTLY BEEN renewed interest in the use of antibodies as carriers for diagnostic agents and anti-tumour agents. A primary consideration is the extent to which the antibodies are retained in the malignant tissue, compared with their concentration in other, non-target tissues and body fluids. If a specific concentration of antibody can be achieved in the target tissue, further studies are needed to determine whether the antibodies are internalized into the malignant cell or retained at or near the cell surface since this would influence the nature of the conjugates required for chemotherapy. Antibodies may be directed at secreted tumour products as well as at membranebound antigens.
Investigations of human colonic-cancer xenografts in hamsters have shown that 3-40 of an injected radioactive dose of iodinated, affinity-purified, specific antibody to carcinoembryonic antigen (CEA) remained per gram of tumour after 8 days.
Nonspecific labelled antibody cleared more rapidly from the tumour, giving a specific: nonspecific ratio of 7.72 + 1.47 (Primus et al.,1.977). In nude mice carrying human 10 colonic-cancer xenografts, optimal visual contrast between the tumour and normal tissue was seen at Day 3 by scanning animals given 2 jig of radio-labelled anti-CEA antibodies diluted with 200 ug of normal goat y-globulins (7S fractions of both). Localization ratios of 4.5:1 were obtained (Mach et al., 1974). It has been clearly shown by postoperative investigations in humans that antiCEA antibodies, whether intact or as fragments, can localize preferentially in tumour tissue as compared to normal tissues (Mach et al., 1980) but the actual recovery of antibody from the tissue is disappointingly small. After 6 days it is about one-thousandth of the administered dose, assuming that the radioisotope remains largely associated with the protein.
Another model to study the localization of antibodies is provided by the human choriocarcinoma xenograft. Antisera to human chorionic gonadotrophin (hCG) have been well characterized (Bagshawe et al., 1979). Hertz (1959) reported the successful growth of metastatic choriocarcinoma (Strains BO, MA and WO in the cheek pouch of conditioned hamsters and in conditioned rats. Further lines were established in hamsters (Ehrmann & Glisermann, 1964;Galton et al., 1963;Hertz, 1967) while Lewis reported the heterotransplantation of choriocarcinoma in the monkey (Lewis et al., 1968). Serial transplantation of an hCG-producing testicular tumour into immune-deprived mice produced a xenograft (HX36) which retained characteristics of the original tumour, but there was a loss in hCGproducing cells after prolonged passage (Selby et al., 1979).
The relative distribution of anti-hCG y-globulins and non-immune y-globulins in Syrian hamsters carrying cheek-pouch tumours of human gestational choriocarcinoma has been compared by totalbody scans (Quinones et al., 1971). Unfortunately, in the ammonium sulphatederived fractions of antisera used by these authors for double isotope experiments, nonspecific immunoglobulin in the specific antiserum would have been labelled inappropriately. Also the expression of results as tissue/blood ratios may depend heavily on the differential clearance of free and complexed circulating antibodies.
The nude mouse has been shown to be a suitable host for human choriocarcinoma, morphological and biological characteristics of the tumour being maintained over serial passages (Kameya et al., 1976). In the present paper we report studies on a xenograft of human choriocarcinoma which has been established in nude mice. Preliminary studies were undertaken to show that specific and nonspecific antibodies could be paired to ensure that they behaved identically in non-tumour-bearing nude mice. The distribution of these antibodies in animals carrying the xenograft was then studied.

METHODS AND) RESULTS
The CC3 xenograft A fresh surgical specimen of human uterine choriocarcinoma was diced into -2mm3 fragments and washed twice in culture medium containing penicillin and streptomycin (Wellcome Reagents, TC199 Single strength). One fragment was implanted s.c. into 2 male and 4 female nude mice, 10 weeks old, at each of 2 sites. Tumours developed in the anterior site (left flank, level with last rib) in all animals after 10-32 days. Tumours developed in the posterior site (over the sacrum) in the female animals only. (It has been suggested previously that regional differences exist in the incidence of successful xenograft growth in nude mice (Auerbach et al., 1978).) The tumour has been passaged 6 times, remaining morphologically stable (Figs. 1 & 2). The CC3 tumour grows as a non-invasive encapsulated nodule in the s.c. space and does not appear to metastasize. Histologically it can be classified as a poorly differentiated choriocarcinoma containing both cytotrophoblastic and syn-** * * w ** . * *s w * *2;:.

Measurement of hCG in mouse serum
Weekly serum-hCGvalues were measured for 10 animals implanted with tumour material of the third passage and 10 sham-operated controls, using an automated f-subunit-directed assay (Kar-~.  (Looney et al., 1973). In individual tumour-bearing animals, the rate of change of serum hCG closely paralleled the rate of change of estimated tumour volume; a typical result is shown in Fig. 5. No detectable serum hCG was noted in controls up to 7 weeks after sham operation. Absolute serum-hCG values were not found to be a reliable indicator of tumour volume when comparing different animals, possibly reflecting variable proportions of viable and necrotic tissue. Typical gonadotrophic responses to circulating hCG can be seen in host mice, in Distribution of thiocyanate-treated nonimmune rabbit y-globulin and anti-hCG rabbit y-globulin in normal male nude mice Normal rabbit y-globulin (Nordic, rabbit IgG, Batch 27-479) (1 mg/ml) was treated with 2-5M ammonium thiocyanate in phosphate buffer (0-05M, pH 7.5) for 1.5 h at 4°C, and dialysed thoroughly against phosphate buffer at 4°C. The resulting y-globulin preparation was labelled to a sp. act. of 1.5 ,uCi/,tg with 1251 (IMS 30, Amersham) in ice, by a modification of the chloramine-T method (McConahey & Dixon, 1969). The combined peak protein-containing fractions from fractionation on Sephadex G-200 (column dimensions 1-5 x 30 cm) were diluted in saline, and 100 pl injected by tail vein into 6 normal nude mice, so that each received 3 HCi 1251/2 ,tg y-globulin.
After 3 days the mice were killed and the organs were excised and the weighed tissues were counted for 1311 and 1251 (LKB-Wallac 8000, 60 sec). The "total injected" was also counted on the day of excision by means of reference aliquots. Surface blood was removed from tissue samples by careful blotting. The results are given in Table I. The immunological activity of the 1311-labelled antibody was checked relative to the starting material by an antiserum dilution curve with 1251-labelled hCG. 60-70% of the binding activity was consistently maintained.
Comparison of the distribution of nonimm,une rabbit y-globulin and anti-hCG rabbit y-globulin in nude mice bearing human choriocarcinoma xenografts The paired antibodies which had been shown in the previous experiment to be distributed equally in normal nude mice were labelled and injected as before (1251-normal y-globulin, 3 MCi/2 ,ug; 1311-anti-hCG y-globulin 2 ,Ci/2 jug per animal) into 5 male nude mice bearing CC3 xenografts of the 6th passage. The mice were killed after 5 days and the tumours and organs excised and counted as for the normal animals. The results are Tumour ct/min/g as % of total/lung ct/min/g as % total, for specific antibody Tumour ct/min/g as % of total/lung ct/min/g as % total, for nonspecific antibody are given in Table III. DISCUSSION It can be seen by comparing Figs 1 and 2 with 3 and 4 that the consistent histological appearance of the human choriocarcinoma cells (CC3) has been maintained during their passage in nude mice. Thus the first essential requirement of our xenograft model has been fulfilled. That the functional integrity of the cells is sustained is evidenced by the steady release of hCG into the mouse serum (Fig.  5). The high levels of hCG in the serum simulate the clinical situation. Under these circumstances, circulating complexes between hCG and xenogeneic antibodies have been indicated (Begent et al., 1980). We have established ( Table I) that, if the antibodies are paired carefully, both the specific and nonspecific Y-globulins are distributed equally in the tissues of normal nude mice. For 6 normal animals, the mean+ s.d. of the counts per gram as a percentage of the total injected were calculated for each tissue listed, and the linear regression plotted (n=7, r=0*998 and 0-925, respectively, P< 0.001) for the specific and nonspecific antibodies.
Treatment of the control y-globulin with ammonium thiocyanate (as in the elution of affinity-purified specific yglobulins) appeared to be an important 1 -   factor in matching the biological half-life paired for molecular size by G-200 of the antibodies. The evidence for the chromatography. The isotope labels could effect of thiocyanate on the distribution of be reversed without altering the distribuy-globulins in normal nude mice will be tion in normal animals. It can therefore presented separately (Lewis & Keep, in be presumed that any difference in distripreparation). The 2 antibodies were bution encountered in tumour-bearing  animals is determined by the antigenic specificity of the immune antibody, whether by direct binding of the free antibody or as a function of complex formation with antigen in body fluids. The expression of the results must attempt to take into account 2 factors: circulating complexes may change the overall pattern of distribution of the isotope associated with the specific antibody; and genuine preferential retention may occur in the tumour as a result of antigen-antibody binding in situ.
We have obtained specificity indices based on the ratio of the counts retained in the tumour to those in the lung, liver and muscle (Table III). In each xenografted animal the lung was the normal organ with the highest associated counts. Since this tissue contains a relatively higher level of blood contamination (J. C. M. Lewis, unpublished) and of macrophages, it seems reasonable to suggest that if specificity indices are demonstrated to be greater than unity when compared to the lung control, there is preferential retention in the tumour. Specificity indices of 1*36 to 2.0 were found relative to the lung (Table III). Despite the fact that antibodies were paired so that they behaved identically in normal nude mice, the overall clearance rate of the nonspecific antibodies in the tumour-bearing mice appeared to be diminished. This is most clearly seen in the histograms (Fig. 6) which are a graphical representation of results for 2 mice from Tables I and II. Further work, with controls deliberately injected with varying levels of circulating hCG, may clarify whether it is the presence of circulating immune complexes, or of the tumour itself, which is responsible for the depressed clearance rate of the nonspecific antibody.
It is evident that the exact tumour localization ratio can vary depending upon the time at which tissues are excised. It seems probable that an observed differential distribution between specific and nonspecific antibodies is determined largely by the time taken for the specific antibody or its complex to be leached away from the neighbourhood of the malignant tissue, i.e. retention rather than uptake is the dominant factor. The diffusion of specific antibodies towards the tumour may be severely hampered by complex formation with circulating antigen. It is interesting that, as in the example quoted in Table II,   antibody. This would be consistent with the observation (Selby et al., 1979) of high hCG levels in fluids from the centre of the xenografted tumouur HX36. Trapping of immune complexes may be occurring in our model. The relative mobility of antibodies into and from cystic spaces with poor surrounding vasculature may contribute to apparent localization, even though the specific antibody or complex is not in intimate contact with the malignant-cell surface. There are obviously a number of questions still to be answered regarding the fate of the antibodies. We hope to determine by autoradiography what proportion of the antibody is internalized by the malignant cell.
During the radioimmunodetection of hCG-producing cancers in humans, activity-concentration ratios of iodine-labelled anti-hCG varying between 1 and 2.87 have been found (Goldenberg et al., 1980). It appears reasonable to suggest that the CC3 choriocarcinoma xenograft, with specificity indices of 1-4 to 2.0, is a realistic working model for considering further the specificity and efficacy of drug-linked antibodies.