Reprints Available Directly from the Publisher Photocopying Permitted by License Only Lymphoid Tumors of Xenopus Laevis with Different Capacities for Growth in Larvae and Adults

Three new lymphoid tumors offering an assortment of variants in terms of MHC class expressions, MHC class II expression, and Ig gene transcription have been discovered in the amphibian Xenopus. One was developed in an individual of the isogenic LG15 clone (LG15/0), one in a frog of the LG15/40 clone (derived from a small egg recombinant of LG15), and one (if-2) in a male ff sib of the individual in which MAR1, the first lymphoid tumor in Xenopus was found 2 years ago. These tumors developed primarily as thymus outgrowths and were transplantable in histocompatible tadpoles but not in nonhistocom-patible hosts. Whereas LG15/0 and LG15/40 tumor cells also grow in adult LG15 frogs, the if-2 tumor, like the MAR1 cell line, is rejected by adult ff animals. Using flow cytometry with fluorescence-labeled antibodies and immunoprecipitation analysis, we could demonstrate that, like MAR1, these three new tumors express on their cell surface lymphopoietic markers recognized by mAbs FIF6 and RC47, as well as T-cell lineage markers recognized by mAbs AM22 (CD8-1ike) and X21.2, but not by immunologobulin (Ig) nor MHC class II molecules. Another lymphocyte-specific marker AM15 is expressed by 15/0 and 15/40 but not if-2 tumor cells. The if-2 tumor cell expresses MHC class molecule in association with ]2-microglobulin on the surface, 15/40 cells contain cytoplasmic class o chain that is barely detected at the cell surface by fluocytometry, and 15/0 cells do not synthesize class o chain at all. The three new tumors all produce large amounts of IgM mRNA of two different sizes but no Ig protein on the membrane nor in the cytoplasm. All tumor cell types synthesize large amount of Myc mRNA and MHC class I-like transcripts considered to be non classical.


INTRODUCTION
The embryonic and larval development of the anuran amphibian Xenopus is a good system for studying ontogenic aspects of the immune system (review in Flajnik et al., 1987;Du Pasquier et al., 1989). Moreover, the unique developmental transition occurring during metamorphosis allows one to study the regulation of the immune fuction in a way that is not possible in mammals. During this crisis period, the immune system undergoes a major remodeling: MHC class I antigens appear on cell membranes (Du Pasquier et al., 1979;Flajnik et al., 1986;; the tissue distribution of MHC class II changes (Du ; the adfflt antibody repertoire replaces the larval one (review in Du Pasquier et al., *Corresponding author. 1989); and an active tolerance to newly arising, adult-specific antigens is acquired (Du Pasquier and Bernard, 1980). However, the advantages of this model have been counterbalanced by a limited armamentarium of methods for studying the phenomena associated with the metamorphosis of the Xenopus immune system. This area of research would greatly benefit from the development of new models such as cell lines and clones with larval or adult phenotypes and exhibiting stage-specific responses in vitro. Lymphoid cell lines are exceedingly rare in amphibians. In fact only one such line has been described so far; B3B7 is derived from a thymic lymphoid tumor MAR1 and grows in histocompatible ff tadpoles but not in histocompatible ff adults (Du Pasquier and Robert, 1992). Recently, three new, apparently similar tumors developed in our colony, each in a different strain. We report here our characterization of these tumors by in vivo 298 J. ROBERT et al. transplantation experiments, flow cytometry, immunoprecipitation, and Western and Northern blot analysis. In future experiments on the metamorphosis of the immune system, we hope to be able to exploit the differences between adult and larval behavior that we found.

Origin of Lymphoid Tumors
The three new lymphoid tumors were found in our frog colony during August and September 1992.
They developed primarily in the thymus, and the frogs bearing a large outgrowth on one side of the head were easily detected. In each case, samples of the tumors' tissue and cell suspension were frozen, transplanted in other animals, and put into culture medium in an attempt to grow the cell in vitro.

Transplantation Studies
The if-2 tumor appeared in a male if, sib of the frog in which the first lymphoid tumor, MAR1, was detected in December 1991 (Du Pasquier and Robert, 1992). The animal carrying this tumor was kept alive as long as possible, and at the time of sacrifice, two months later, there were multiple metastases in liver, kidney, spleen, and blood. The contralateral thymic lobe was not affected by the tumor. Cells of this tumor are transplantable in partially inbred MHC-homozygous tadpoles but not in unrelated larval hosts (Table 1). Ascites containing about 107 tumor cells are usually obtained 2-3 weeks after intraperitoneal injection of a 5 x 104-1 x 105 single tumor cell suspension. Like MAR1 cells, if-2 cells are rejected by postmetamorphic young or fully grown adults of the same family.
Dorsal subcutaneous injections of 1 x 106 cells, which in our experience is the most sensitive route of injection in adults, failed to produce any tumor.
Rejection occurs as soon as 1 month after the completion of metamorphosis. Both in normal and sodium perchlorate blocked tadpoles (that do not metamorphose), metastases developed secondarily in spleen, liver, and at the site of injection. In one case, metastases were also observed in kidney.
The other two tumors appeared in two LG15 lines--one in the original LG15/40, a clone derived from a LG 15 small egg recombined clone. Both 15/0 and 15/40 tumor cells are transplantable in isogenic Other features discriminate among the tumors.
The class I-specific TB17 mAb positively stains if-2 cells ( Fig. 1); because all live cells are stained, this is not the result of contamination from host cells. In  Hsu et al., 19914. Flajnik et al., 19915. Flajnik et al., 1990 6. Du Plasquier and  7. Nagata, 19858. Flajnik et al., 1988 and relate to the relative brightness of the whole tumor cell population. ND not done.

Characterization of the Cell Surface by Immunoprecipitation
In order to confirm the analysis by flow cytometry, we used immunoprecipitation after cell-surface iodination (Fig. 2). From lysates of all tumor cell types, including the B3B7 cell line, a unique band, which corresponds to a band of 36 kd in normal splenocyte lysates, was detected with mAb AM22 on SDS-PAGE gels under reducing conditions (Fig. 2B). A second band of 38 kd, usually present in normal spleen lysates ( Fig. 2A), could not be found even when the cells were preincubated with the mAb before lysis (data not shown); this result suggests that these tumor cells express only one chain of the putative. CD8 molecule.
With the other T-cell-specific marker recognized by mAb X21.2, the situation is more complex (Fig.  2B). Whereas three bands are immunoprecipitated from the B3B7 cell line, only two bands are present in lysates of if-2, and no signal at all could be FIGURE 2. Comparative immunoprecipitation analysis under reducing conditions of cell-surface markers expressed by normal lymphocytes and the various tumors. (A and B) Surface iodinated lysate from normal outbred lymphocytes and each tumor cell type were immunoprecipitated by the SPIT technique and separated on 10% SDS-PAGE gels. Lane 1: Negative control with nonspecific mAb supernatants, Lane 2:TB17 anti-MHC class I, Lane 3:AM22 anti-CD8-1ike, Lane 4:X21.1 anti-T cell. Only one band is present in 15/0 lysate. (C) Normal lymphocytes and if-2 tumor cells biosynthetically labeled with 35S-methionine were preincubated with (1) normal nonimmune Xenopus serum or (2) specific anti-ff alloserum, lysed, immunoprecipitated by protein A, and loaded on a 7.5-15% SDS-PAGE gel gradient. detected with lysates of 15/40 and 15/0 tumor cells, even when more gentle washing conditions are used, and despite the fact that a signal was recorded in immunofluorescence. Normal spleen lymphocytes express one major band, but two faint bands were also detected after a long exposure ( Fig. 2A).
Using the class-I-specific mAb TB17, we precipitated a strong band of 45 kd corresponding to a class I c chain from cell-surface iodinated if-2 but not the other tumors. As previously reported, ]/2-microglobulin (the MHC class light chain) could not be detected using this mAb . When we used a specific anti-ff alloserum preadsorbed before lysis and immunoprecipitation on metabolically labeled if-2 tumor cells or normal splenocytes, a band of 13 kd corresponding to ,82m was clearly seen; this band was not present in controls with nonimmune Xenopus serum (Fig. 2C). A higher background is allways obtained after immunoprecipitation of tumor cell lysates with protein A, even after preincubation with normal Xenopus serum and protein A.
No Ig molecules were detected from iodinated cell-surface protein of any tumor cell (data not shown).

Western Blotting
When cell lysates were analyzed by Western blotting, we could identify the MHC class I c chain protein in lysates of the 15/40 tumor (Fig. 3). To avoid contamination with host class I, lysates were prepared from tumor cells grown in vitro. A band with a slightly slower mobility than that of if-2 tumor cells was clearly present, and no signal was detected in lysates of 15/0 tumor cells or B3B7. This difference in gel mobility is probably due to allelic differences between the LG15 and ff strains of Xenopus . The integrity of cell lysates was controlled with mAb AM22, which stained a band of the expected apparent molecular weight in all tumor samples, but not in spleen. We interpret these findings to mean that MHC class c

Northern Blotting
Although cells of the three new tumors do not express Ig molecules on the cell surface or in the cytoplasm, they all produce large amounts of / heavy-chain mRNA detectable on Northern blots; no signal at all could be detected in the B3B7 cell line (Fig. 4). In all three new tumors, the C/ probe detects, in addition to the usual 1.8-kb / mRNA previously described (Schwager et al., 1988), another transcript of 1.35 kb that is barely detectable in normal lymphocytes. The shorter / band was observed in several experiments, and it does not seem to result from a general RNA degradation, as judged by hybridization of other probes on the same membrane. The large amount of specific/ messenger cannot be attributed to contamination from host RNA alone. Moreover, RNA from 15/40 cells grown in vitro contains comparably large amount of / transcript.
To analyze MHC class molecules the membrane was first hybridized with a probe specific for the classical MHC class I transcript (A12 c3 domain), which corresponds to the antigen recognized by mAb TB17 (Schum et al., 1993). The membrane was then hybridized with an 3 CL4 probe that had been shown to hybridize with at least 10 to 20 different nonclassical MHC class I-like sequences of yet unknown function . By using the A12 probe, two transcripts of 2.4 and 1.4 kd were detected in normal thymus and in if-2 and 15/40 tumor cells (either from in vivo transfer or from in vitro culture). In contrast, B3B7 and 15/0 yielded no signals. But after hybridizing with the CL4 probe under stringent conditions, two transcripts were detected in all RNA samples (Fig. 4), including the 15/0 tumor and B3B7 cell line. The sizes of the transcripts identified by the A12 and CL4 probes are similar.
Hybridization with a Xenopus myc-specific probe detected a large amount of Myc mRNA, especially in the B3B7 cell line and the 15/0 tumor, which we estimate to contain more Myc RNA than do normal thymocytes (Fig. 4) or the A6 kidney cell line (data not shown). Two distinct copies of Xenopus c-myc gene have been isolated, but their discrimination is only possible by using oligonucleotide specific of the 5' untranslated region (Principaud and Spohr, 1991;Vriz et al., 1989). The elevated or deregulated expression of c-myc oncogene has been reported for a large number of tumor cells in higher vertebrates and its involvement in neoplastic transformation has been postulated (for a review, see DePinho et al., 1991). A similar implication of c-myc is likely in these Xenopus lymphoid tumors.

Discussion
We regularly examine our frogs for evidence of tumors and found no lymphoid tumors in the first 20 years or so of its existence. Then, at the beginning of 1992, the first Xenopus lymphoid tumor MAR1 was discovered in our colony (Du Pasquier and Robert, 1992), and 8 months later, three more appeared. We are not yet sure of the factors responsible for these neoplasms. It would not seem to be genetics, for three different clonal lines of Xenopus are involved (if, LG15, and LG15/40). Moreover, infectious fungi or bacteria are definitively not involved. The new tumors, like MAR1, can only be propagated in isogenic or highly homozygous hosts; FIGURE 4. Northern blot analysis of the various tumors and normal thymocytes (Thy) with Xenopus C, c-myc, and classical and nonclassical MHC class probes. Cytoplasmic RNA (10/g) was electrophoresed in a 1% formaldehyde-agarose gel, transfered to a nylon membrane and hybridized with 32p-labeled probes. The sizes of mRNAs were calibrated using RNA markers from BRL. The integrity of each RNA sample was controlled by reprobing the membrane with a X. laevis Ef-l probe. 15/40 tumor cells can also grow in vitro. Such criteria strongly support the interpretation that we are dealing with real lymphoid tumors and not with granulomas (Asfari and Thi6baud, 1988). Each tumor seems to be the result of an independent oncogenic event. That is, they appeared in animals of different genetic background, they behave differently after transplantation, and they display distinct phenotypes, as shown by flow cytometry and Northern blotting. All attempts to demonstrate the presence of a virus have been unsuccessful. Like the B3B7 cell line, which was derived from the MAR1 tumor, all of the new tumors have mixed T-B phenotypes (several surface markers of T cells, but Ig RNA is transcribed). Aside from the possibility of enabling us to learn more about neoplasm biology in amphibians, these tumors provide a panel of new experimental models to study lymphocyte differentiation and the ontogeny of the immune system in amphibians. whereas MARl-derived B3B7 cells of the same strain (if) do not. In contrast to mammals, which express class I in the embryonic stages, the classical MHC class molecules of Xenopus (present on the membrance of almost all cells in the adult) are not detected in young tadpoles and appear only during metamorphosis (Flajnik et al., 1986;. In vivo transplantation into young tadpoles (stages 55-56), or into tadpoles that have been inhibited from metamorphosing, does not modify class I expressing if-2 tumor cells as judged 304 J. ROBERT et al. by fluocytometry or immunoprecipitation. Nevertheless, the larval immune system does not seem to be able to reject these class I bearing cells, although there is no obvious reason why it should be tolerant to them.

Growth Capacities of the Transplanted Tumor Cells in Larval and Adult Histocompatible Hosts
The fate of tumor cells in the adult does not seem to depend on the expression of MHC class I molecules on the cell surface, because both class I / if-2 and class I-MAR1 tumor cells do not grow, or rarely grow, in adults older than 2 months after fertilization. Several hypothesis can be advanced to explain this peculiarity. The preferential growth in tadpoles, independence of MHC class I expression, might be due to a better capacity-of the adult immune system to detect and kill tumor cells. The rejection of MAR1 tumor in irradiated adults (with exception of one case) as well as in thymectomized allogenic toads (Du Pasquier and Robert, 1992) suggests that additional, nonimmunological factors are involved. A detailed analysis of tumor growth at various developmental stages is in progress; we will determine whether tumor rejection coincides with the host acquiring full immunological maturity.
Other possible mechanisms of the developmental variation in tumor rejection should be considered: (1) absence of tumor-specific recognition by the larval immune system, (2) developmental activation in the host of a resistance gene during metamorphosis, and (3)  There exist at least 20 different genes encoding class I-like sequences, most of which are not MHCqinked . Their function and their specific expression are as yet unknown, but it is possible that at least one of these MHC I-like proteins is involved in the developmentally regulated tumor rejection. Within the ff family, skin grafts are often accepted for more than 200 days, but these grafts are eventually rejected after 300 days; thus, this family is not isogenic at all histocompatibility loci (Du Pasquier and Chardonnens, 1975). New family studies are necessary to determine whether or not minor histocompatibility antigens still segregating in our ff pool are responsible for the rejection of tumor transplants by adults. In any case, we do not understand why the tumor cells should not be accepted for at least some months. In addition to classical immunological parameters, endocrine factors, homing prob-lems, and NK cells could interfere with the growth of tumor cells in the adult more than in the tadpole. Following the fate of transplanted tumor cells through metamorphosis and for a few months thereafter might enable us to determine the period of rejection more precisely and correlate it with, for example, the histogenesis of the thymus that takes place 1 month after metamorphosis.
In contrast to if-2 and the previously described MAR1 tumor cells, 15/0 and 15/40 cells can grow in adults of the LG15 genotype. The LG15/0 tumor developed in an animal that was genetically identical to the other LG15 individuals (Kobel and Du Pasquier, 1975 Because specific transcripts of the expected size (Schwager et al., 1988)  labeling before immunoprecipitation. Anchored or single-sided PCR with a C/ primer may also be considered in order to isolate and sequence mRNAs of different length (Troutt et al. 1992).
Despite the rearrangement of Ig genes in all tumors, and the expression of Ig messengers in three of them, none of the tumor seems to synthesize Ig protein chains. Perhaps all these messengers are simply out of frame. But another possibility would be that Ig production is selected against, and Ig-negative cells outgrow, the Ig-positive cells in incipient tumors. There might also be some incompatibilities between the expression of the various T-cell markers and the synthesis of Ig. In any case, sequence analysis of corresponding Ig gene rearrangements that are in progress would clarify these point.

Use of Tumor Cells as a Source of Proteins of Immunological Interest
Large amounts of proteins of immunological interest can be purified from tumor tissue and characterized without contamination by other cell types. One such protein is ,82 microglobulin, for which no specific antibodies are available and for which the gene has not been cloned and sequenced in Xenopus. Its association with the class I chain at the surface of if-2 cells means that anti-MHC I specific allosera can be used to immunoprecipitate and isolate the molecule for peptide sequencing. The tumor offers a better starting material than red cells, the lysate of which are always rich in the globin 13 kd monomer that interferes with ,82-m isolation.
Another protein of interest is the putative CD8 molecule recognized by mAb AM22. The presumption of homology of this molecule to mammalian CD8 is based on the similarity of apparent molecular weight and tissue distribution; there are no protein or DNA sequences. The high level of expression by the various tumors of at least one CD8-1ike subunit should enable us to immunoprecipitate enough material to sequence.
The 15/40 tumor cells are now proliferating in vitro, and new, cloned cell lines will soon be established. The same strategy of successive passages from in vivo transplantation to in vitro culture is being followed with the two other tumor types.

Toads
The Xenopus ff inbred family (Du Pasquier and Chardonnens, 1975) is highly homozygous. As these animals retain skin grafts for more than 200 days, they are considered to be histocompatible.
LG15 isogenic clones have been described in detail elsewhere (Kobel and Du Pasquier, 1975). The LG15/40 line arose from the development of a small egg that did not undergo genomic duplication and was therefore recombinant (Kobel and Du Pasquier, 1977;Wilson et al., 1992). Development was blocked before metamorphosis at stage 48 (Nieuwkoop and Faber, 1967) by adding 0.1% potassium perchlorate (KC104) to the water (Buscaglia, 1977).
Cell Culture A cell line was derived from the LG15/40 tumor in the same way as line B3B7 was derived from MAR1 (Du Pasquier and Robert, 1992) by 6 months of repeated in vivo and in vitro passage. Culture media and culture conditions have been described (Du Pasquier and Robert, 1992).

Flow Cytometry
Samples of 105 cells stained with fluorescencelabeled hybridoma supernatants were analyzed by flow cytometry on a FACScan apparatus, as described by Du Pasquier and Flajnik (1990). Immunoprecipitation Cell-surface iodination and lysis in NP-40 were done as described (Kaufman et al., 1985;. A solid-phase immunoisolation technique (SPIT) was used (Tamura et al., 1984;. A 96-well plate was coated overnight at 4 C with 100 1 of goat anti-mouse Ig at 100 g/ml (Kierkegaard), washed and blocked with 1% BSA in PBS, and incubated at room temperature 2 x 2 hr with 100 1 of specific hybridoma supernatants. After washing several times, radiolabeled cell-surface lysates (10 s cpm/ml corresponding to ca. 106 cells were incubated overnight at 4 C. After washing twice with cold 50 mM Tris (pH 8), 500 mM NaC1, 1 mg/ml ovalbumin, and twice with 50 mM Tris, 100 mM NaC1, the samples were resuspended in Laemlli sample buffer with 5 % 2-mercaptoethanol and separated on 10% SDSpolyacrylamide gels. Cells labelled in vitro with 35S-methionine were immunoprecipitated with specific alloantisera, as described (Kaufman et al., 1985;.