Development and Analysis of Various Clonal Alloantigen- Dependent Cytotoxic Cell Lines from Channel Catfish1 2

To determine the phenotypes of cytotoxic cells in channel catfish, clonal alloantigen-dependent leukocyte lines were established from mixed leukocyte cultures. Each clone was analyzed for expression of TCR α and β genes by RT-PCR and for target cell specificity by 51Cr-release assay. Based on the above criteria, the following five different cell types were identified among the 19 clones analyzed: 1) TCR αβ+ allospecific cytotoxic cells, 2) TCR αβ+ nonspecific cytotoxic cells, 3) allospecific TCR αβ+ noncytotoxic cells, 4) TCR αβ− nonspecific cytotoxic cells, and 5) TCR αβ− allospecific cytotoxic cells. The demonstration of cloned, TCR αβ+, allospecific cytotoxic effectors provides the strongest evidence to date for the existence of cytotoxic T cells in fish.

A lthough teleost fish possess subpopulations of lymphoid cells that mediate immune responses analogous to those seen in "higher" animals (reviewed in Ref. 1), relatively little is known about cell-mediated immunity in any fish species. For example, Ag-specific CTL homologues have yet to be unequivocally identified in any teleost. Similarly, the existence of bona fide NK cells in teleosts remains debatable. Major factors contributing to this dearth of information in teleosts have been the lack of physiologically relevant in vitro model systems and the paucity of molecular and cellular markers.
Previous work showed that several fish species possess nonspecific cytotoxic cells (NCC) 4 that lyse xenogeneic targets (reviewed in Ref. 2). However, these effectors, suggested to be homologues of the NK cells of "higher" animals, have not been well characterized at the molecular level. In the best characterized of these systems, Evans and his coworkers (2) identified a 34-kDa cellsurface receptor protein, termed NCCRP-1, on channel catfish NCC that is thought to bind target cells. In addition, a previously generated mAb, 5C6, was shown to bind NCCRP-1 and to provided a facile way to identify NCC. However, mAb 5C6, which reacts with head kidney-derived catfish NCC, fails to react with catfish cytotoxic effectors derived from peripheral blood leuko-cytes (PBL), suggesting that NCC and PBL-derived cytotoxic effectors represent distinct populations (3). In our laboratory, several recent developments within the channel catfish model have greatly facilitated progress in characterizing PBL-derived cytotoxic cells. First, the development of clonal long-term autonomous (i.e., able to proliferate continuously in culture without the need for restimulation or exogenous growth factors) lines of B cells, T cells, and macrophages (4 -7) makes available well-defined homogeneous populations of allotargets, which express both MHC class I and II genes (8,9). Second, the finding of greatly enhanced cytotoxicity after one-way mixed leukocyte culture of catfish PBL with x-irradiated allotargets provides a useful in vitro system for generating large numbers of teleost cytotoxic cells (10). Finally, as reported in this paper, it is now possible to clone mixed leukocyte culture (MLC)-derived effectors and determine their phenotype by monitoring their ability to express the TCR ␣ and ␤ genes and to recognize and lyse specific allogeneic targets. As detailed below, the findings in the current study demonstrate that catfish possess at least four different types of cytotoxic cells that can be quantitatively expanded in vitro by MLC. In addition, a fifth population of noncytotoxic, allospecific TCR ␣␤ ϩ cells that may correspond to mammalian Th cells was also isolated.

Experimental animals and lymphoid cell isolation
Channel catfish (Ictalurus punctatus; 1-2 kg) were obtained and maintained in individual tanks as described previously (11). Blood was drawn from the caudal vein of anesthetized (tricane methansulfonate) fish into heparinized vacutainers, and PBL were isolated by centrifugation over Lymphoprep (Accurate Chemicals, Westbury, NY) as described previously (6).

Alloantigen immunization
Channel catfish #32 was hyperimmunized i.p. with 3B11 cells, a clonal allogeneic B cell line (2-4 ϫ 10 7 cells in 2 ml PBS). The fish was boosted after 14 days and then weekly for an additional 6 wk. Two weeks after the last boost, 10 -15 ml blood was collected using Vacutainer blood collection tubes (Becton Dickinson, Franklin Lakes, NJ). The PBL were used to develop the alloreactive lymphocyte clones described below.

MLC-derived catfish lymphoid cell lines
MLC-generated cytotoxic effector cells were obtained as described previously (10). Briefly, 5 ϫ 10 6 catfish PBL from a 3B11-immunized fish (fish #32) or nonimmunized fish (fish #10 and #75) were incubated with 2 ϫ 10 6 irradiated 3B11 cells in 1 ml AL-5 medium/well using 24-well tissue culture plates (Corning Glass, Corning, NY) and incubated at 27°C in a humidified atmosphere containing 5% CO 2 /95% air. The cultures were passaged when cell density increased and culture medium became acidified. Typically, 1 ml AL-5 medium/well was added after day 5 of culture, and on day 6 the pooled contents of two wells were transferred to a 25-cm 2 tissue culture flask (Corning) with an additional 4 ml of AL-5 medium. On day 8, 8 ml AL-5 medium was added, and on day 10 the contents (16 ml) of the 25-cm 2 flask were transferred to a 75-cm 2 T-flask (Corning) along with an additional 16 ml of medium. The cells were harvested on or after day 12 and cloned by limiting dilution (see below) or continuously cultured by weekly restimulation with irradiated 3B11 cells. In the later case, 10 6 MLC cells were cultured with 2 ϫ 10 6 irradiated 3B11 cells in 1 ml complete medium containing 10% conditioned medium from 75C T cells (as a presumed source of growth factors). After 2 days, an additional 1 ml of 10% conditioned medium was added to the wells, and after 4 days, the pooled contents of two wells were transferred to a 25-cm 2 flask together with 4 ml of 10% conditioned medium.

Cloning and propagation of MLC cells
Cloning of alloantigen-responsive cytotoxic cells was performed by limiting dilution in 10% conditioned medium using freshly isolated PBL-or MLC-activated leukocytes from both 3B11 immune and nonimmune fish. Briefly, 50-l suspensions of PBL (20 cells/ml) or MLC-activated cells (6 cells/ml) were transferred to wells in 96-well round-bottom tissue culture plates together with 50 l irradiated (4000 -5000 rad) catfish allogeneic B cells (3B11) suspended at 2 ϫ 10 5 cells/ml in conditioned medium. The cells were incubated at 27°C in 5% CO 2 /95% air. After 8 -15 days in culture, proliferating responder cells were restimulated. Briefly, half of the culture supernatant in each well containing proliferating cells was removed and the cells were resuspended and completely transferred to a flat-bottom 96-well plate. Irradiated 3B11 cells (3 ϫ 10 6 cells/ml) were suspended in 10% conditioned medium, and 100-l aliquots were transferred to each well containing responder cells. The clones were cultured for 6 days and then transferred to the wells of 24-well tissue culture plates together with 2 ϫ 10 6 irradiated 3B11 cells in 1.5 ml 10% conditioned medium and cultured 6 more days. From this point on, the cloned cells were passaged in wells of 24-well tissue culture plates at 6-day intervals by transferring 5 ϫ 10 5 responder cells and 2 ϫ 10 6 irradiated 3B11 cells in a final volume of 1 ml 10% conditioned medium per well. Optimal expansion of cloned cells during a 6-day culture period was achieved by adding 1 ml 10% conditioned medium to each well after 4 days. Alternatively, the pooled contents of three wells were transferred to 25-cm 2 tissue culture flasks together with 3 ml 10% conditioned medium on day 4 and cultured in flasks for 2 more days. The specificity of alloantigen stimulation was assayed by culturing 10 5 cells from the various clonal MLC-derived lymphocyte lines with 2 ϫ 10 5 irradiated 3B11 cells (as specific Ag) or 1G8 cells (as nonspecific Ag) in 200 l of AL-5 medium with or without 10% conditioned medium in 96-well round-bottom tissue culture plates. Triplicate cultures were incubated at 27°C in a humidified atmosphere containing 5% CO 2 and 95% air for 4 days. The cultures were pulsed with 0.5 Ci Cytotoxic assays 51 Cr-release assays were performed as previously described (13). Briefly, effector cells in 100 l AL medium were mixed with 5 ϫ 10 4 51 Cr-labeled target cells in 100 l AL medium in round-bottom 96-well tissue culture plates (Corning). The plates were centrifuged for 1 min at 200 ϫ g to induce contact between effector and target cells and were incubated for 4 h at 27°C. The cells were resuspended by pipetting, and the plates were centrifuged again for 3 min at 550 ϫ g. One hundred microliters of cellfree supernatant was removed from each well and cpm were determined in a COBRA II auto gamma-counter (Packard). E:T ratios were as indicated in the text and figure legends. All experiments were done in triplicate. Percent specific release was calculated using the following formula: % specific release ϭ {100 ϫ [cpm (experimental) Ϫ cpm (minimum release)] / [cpm (maximum release) Ϫ cpm (minimum release)]}. Maximum release wells received 100 l 2% Nonidet P-40 (Sigma, St. Louis, MO) instead of effector cells to lyse all target cells. Minimum release wells received 100 l AL instead of effector cells.

RNA isolation, RT-PCR, 5Ј-rapid amplification of cDNA ends (RACE), and sequence analysis
Total RNA was prepared from the cell lines by the method of Chomczynski and Sacchi (14). To detect expression of channel catfish Ig, TCR, and actin, ϳ1 g of total RNA was converted into first-strand cDNA using 50 ng of an oligo(dT) primer with 200 U of reverse transcriptase (SuperScript II; Life Technologies) according to the manufacturer's recommended protocol. One percent of the first-strand reaction was used as template in RT-PCR. The reaction mixture (100 l) contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 3 mM MgCl 2 , 150 mM dNTP, 40 mM tetramethylammonium chloride, and 0.5 g of specific forward and reverse primers. Two units of Taq DNA polymerase (Perkin-Elmer, Branchburg, NJ) were added, and 30 cycles of amplification (94°C for 1 min, 65°C for 2 min, 72°C for 3 min) were performed. Primer pairs for Ig (15), TCR (12), and actin (16) have been described. The TCR V regions of the different cell lines were obtained either by RACE (17) using the 5Ј-RACE System version 2.0 (Life Technologies) or by RT-PCR using specific oligo primers. Primers for the first-strand RACE cDNA synthesis were 5Ј-GCAGGCAAATGAAAGTAGAATT-3Ј for ␣ (designated G-902) and 5Ј-AAAACCCTGTCTCCTAACGATGTA-3Ј for ␤ (designated G-997). The specific reverse and constant region primers for PCR amplification with the 5Ј-RACE abridged anchor primer were 5Ј-GTTTGTAAATTGACGGCT-3Ј for ␣ (designated G-1085) and 5Ј-CACTTCCTTCTCTGAAACA-3Ј for ␤ (designated G-1013). Reaction conditions were as recommended by the manufacturer. The TCR ␣-chains of clones TS.32.1, TS.32.5, TS.32.17, and TS.32.34 were amplified using a TCR ␣ leader region forward primer (5Ј-ATGGCTTCAATAGGTGAC-3Ј) and reverse G-1085. The 5Ј-RACE products and RT-PCR fragments were cloned into pCR2.1 (Invitrogen, Carlsbad, CA) and sequenced using the dideoxynucleoside triphosphate chain-termination method (18). Nucleotide and amino acid sequences were initially aligned using the Megalign program (DNAstar, Madison, WI) by the Clustal V method with the PAM250 residue weight table and then were adjusted by visual inspection to maximize homology.

Clonal alloantigen-dependent cytotoxic cell lines generated from channel catfish PBL
Previous studies demonstrated that high levels of nonspecific cytotoxic activity directed against allogeneic target cells were generated in one-way MLC using catfish PBL as responders and irradiated allogeneic cells as stimulators (10). Subsequently, cytotoxic cultures were found to continuously proliferate after weekly restimulation with alloantigen(s) in the presence of conditioned medium. Each stimulation was characterized by a short proliferative response resulting in a 3-to 4-fold expansion of cell numbers. Conditioned medium was not required for cell expansion but was important for survival, i.e., cells that were no longer proliferating remained viable for several days in various conditioned media, whereas cells suspended in nonconditioned medium died quickly. Because the phenotype(s) of the MLC-expanded catfish cytotoxic cells was not known, cells responsible for killing allogeneic targets were cloned and subsequently characterized using molecular probes for the catfish TCR ␣ and ␤ genes. PBL from two nonimmune and one 3B11-immunized channel catfish were cloned by limiting dilution either with or without prior stimulation and expansion in MLC. The number of wells containing proliferating cells (cloning efficiency) was estimated by microscopy. Expansion in MLC before cloning dramatically increased the cloning efficiency from both sources. For example, only 0.7% of PBL from nonimmune fish #75 could be cloned directly, whereas 47% of the MLC-expanded cells yielded clones. Prior immunization with alloantigen appeared to increase the cloning efficiency as PBL from a 3B11-immunized fish (#32) yielded significantly higher numbers of clones after both primary and repeat stimulation, i.e., 6.3% and 83%, respectively. These results showed the feasibility of cloning catfish cytotoxic cells and demonstrated that both prior immuni-zation and MLC expansion lead to markedly enhanced cloning efficiency.
To determine their phenotype, the resulting clones were cultured for several weeks after which they were screened for expression of TCR ␣, TCR ␤, and Ig genes by RT-PCR and for allospecific cytotoxic activity in 51 Cr-release assays. Based on reactivity patterns, five groups of clones were identified. One group (I) consisting of 10 clones, each derived from 3B11-immune fish #32,  contained cells that were TCR ␣␤ ϩ (Table I). This group (represented by clone TS.32.5 in Fig. 1) lysed 3B11 cells but did not lyse any of three other allogeneic targets, each derived from individual fish different from the one that gave rise to the 3B11 cell line. A second group (II) contained three clones that were also TCR ␣␤ ϩ but did not display strict specificity in target cell lysis. Clone TS.32.32, which is representative of group II, lysed 3B11 as well as 1G8 target cells with high efficiency (Table I & Fig. 1). However, this clone failed to lyse two other allogeneic targets (Fig. 1).
The other two clones in group II were also capable of lysing several different allogeneic targets but not others (data not shown). Two other groups of clones (IV and V), isolated from both immune and nonimmune fish, were TCR ␣␤ Ϫ but differed in cytotoxic activity. The three clones in group IV lacked allospecificity, i.e., they lysed 1G8 target cells as well as 3B11 cells (Table I). A representative member of this group, TS.75.2, lysed three allogeneic target cells including 3B11 cells with no apparent specificity (Fig. 1). However, TS.75.2 did not lyse autologous target cells  (19), and the V, CDR3, and FR4 regions are marked by "ϾϽ." The TCR ␣ sequences of 32.5, 32.34, and 32.17 are not full-length and were amplified using a forward leader primer and a reverse C␣ primer. The TCR ␣ and ␤ sequences of 32.44 obtained by 5Ј-RACE were shorter than expected, which may be due to the quality of the RNA isolated from this cell line. Only a single ␤ transcript was amplified from each of the clones. However, an additional aberrant ␣VJC rearrangement was amplified for 32.5 and clones 32.1, 32.34, 32.17, and 32.43; all had short transcripts (beginning in FR2) identical with the V␣ rearrangement found in clone 32.15. No full-length transcripts of the 32.15V␣ sequence could be amplified from these clones, even by using specific 32.15V␣ primers. Sequence data were not obtained for TCR ␣␤ ϩ clones 32.13, 32.39, and 32.42; however, RT-PCR analysis yielded the correct size ␣ and ␤ fragments.
(75C) derived from the same fish (Fig. 1). The second group (V) of TCR ␣␤ Ϫ cytotoxic cells consisting of 3 clones displayed allospecificity in target cell recognition, i.e., 3B11 cells were lysed more efficiently than 1G8 allogeneic target cells (Table I). A representative clone from this group, TS.10.1, lysed four different allogeneic target cells including 3B11 cells. However, although 3B11 cells were lysed at high efficiency, lysis of three other allogeneic target cells occurred at relatively low efficiencies (Fig. 1). Finally, a single clone, TS.32.4, was TCR ␣␤ ϩ but was unable to lyse 3B11 or 1G8 target cells in 4-h 51 Cr-release assays. However, TS.32.4 cells proliferated specifically to stimulation with irradiated 3B11 cells but not to stimulation with irradiated 1G8 cells (data not shown).
To determine whether the TCR ␣␤ ϩ clones derived from fish #32 involved the same or different precursors, TCR ␣ and ␤ cD-NAs were sequenced for each clone. All of the clones expressed members of the V␣ 1a family (12). Two clones, TS.32.12 and TS.32.44, exhibited identical VJ rearrangements and two others, TS.32.15 and TS.32.49, rearranged identical V␣ to different J␣ ( Fig. 2A). Each V␤ rearrangement was unique, with clones TS.32.5, TS.32.12, and TS.32.44 expressing different members from the same V␤ family (Fig. 2B). These findings indicate that each of these cell lines are clonal and likely derived from independent precursors.

Channel catfish alloantigen-dependent TCR ␣␤ ϩ cytotoxic cells proliferate in response to stimulation with specific alloantigen
Irradiated stimulator cells were crucial for propagation of both TCR ␣␤ ϩ and TCR ␣␤ Ϫ cells during the process of cloning. To test whether or not the various cytotoxic clones required specific allostimulation for in vitro expansion (or if any allogeneic stimu-lator would suffice) an allospecific TCR ␣␤ ϩ clone (TS.32.5) and an allospecific TCR ␣␤ Ϫ clone (TS.10.1) were stimulated with specific alloantigen (3B11 cells) or with a different alloantigen (1G8 cells) in the presence or absence of conditioned medium. As shown in Fig. 3, TS.32.5 responded only to 3B11 cells, whereas clone TS.10.1 proliferated in response to both 3B11 and 1G8 cells. Furthermore, TS.32.5 cells proliferated equally well after stimulation with 3B11 cells whether or not conditioned medium was included in the culture. In contrast, a mixture of conditioned medium and irradiated allogeneic cells appeared to synergistically enhance proliferation of TS.10.1 cells (Fig. 3).

Discussion
Both in vivo and in vitro studies indicate that teleosts possess allospecific cell-mediated immunity similar to that seen in higher vertebrates. For example, primary and secondary rejection of allografted skin/scales in vivo has been demonstrated in representatives of nearly all major groups of fish (reviewed in Ref. 20). Moreover, leukocytes isolated from carp immunized with allogeneic erythrocytes lysed homologous erythrocytes, demonstrating allospecific cytotoxic responses in an in vitro system (21). In another study, trinitrophenyl-modified Con A blasts from carp were lysed by autologous but not allogeneic immune leukocyte effectors, suggesting genetically restricted target cell recognition (22). Although the above studies are indicative of CTL-like activity, defined populations of teleost cytotoxic cells previously had not been isolated. Consequently, it had not been possible to determine the function and phenotype of putative subpopulations of fish cytotoxic effector cells. In the work reported here, four types of cytotoxic cells were cloned from channel catfish PBL. This was accomplished using a culture technique similar to that employed for cloning mammalian alloreactive CTL and NK cells, i.e., periodic stimulation with allogeneic cells in the presence of conditioned medium (23). Crucial for propagation of clonal catfish cytotoxic cells was the availability of allogeneic long-term leukocyte lines as continuous sources of defined MLC stimulator cells (5,6,12). Four different groups of cloned catfish cytotoxic cells were identified based on allospecificity and expression of TCR ␣␤ genes. One group of effectors (group IV) was nonspecific, TCR ␣␤ Ϫ , and was considered to be equivalent to mammalian NK cells, whereas two other groups of effectors (groups I and V) demonstrated allospecific cytotoxic activity. Clones in group I expressed messages for both catfish TCR ␣ and ␤ and may be considered bona fide CTL. The second group of allospecific effector clones (group V) lacked TCR ␣␤ expression. Therefore, it is not clear whether these cells are CTL or a second type of NK-like cell. Although mammalian NK cells typically lyse target cells in a nonspecific fashion, it has been demonstrated that human NK cells can cause specific lysis of allogeneic PHA-induced lymphoblasts after stimulation in vitro with homologous lymphoblasts (24). By analogy, a similar mechanism for target cell recognition may exist for catfish allospecific NK-like lysis. It should be noted that although the TCR ␣␤ Ϫ allospecific cytotoxic effector cells described here are termed NKlike, the possibility of TCR ␥␦ expression in one or both of these groups of clones was not ruled out. Putative channel catfish TCR ␥␦ genes have not yet been identified, although TCR ␥␦ (as well as TCR ␣␤) genes have been isolated in cartilaginous fish. Thus, equivalent genes are possibly present in teleosts because it seems likely that TCR ␥␦ genes originated in an ancestor common to all jawed vertebrates (25). Catfish cytolytic cells that were TCR ␣␤ ϩ but nonspecific in target cell recognition (group II) were difficult to categorize because they did not match the phenotype and function of known mammalian cytotoxic cells. It is possible that these clones may recognize a target structure that is similar on the two target cells. Although possibly not applicable to the clones in question, it is interesting to note that mammalian CTL clones may convert from allospecific to nonspecific target cell recognition after prolonged culture or after culture in high concentrations of IL-2 (26,27). It is also possible that these particular fish cells may represent a form of NK-like T cells (28). An additional TCR ␣␤ ϩ clone (group III) proliferated specifically in response to irradiated 3B11 cells but did not lyse 3B11 cells or other allogeneic targets. This clone may be a candidate for a catfish homologue of mammalian T helper cells.
Although most of the cytotoxic cells derived in this study were cloned after stimulation in MLC, some were cloned directly from freshly isolated PBL. It is noteworthy that the cloning efficiency of PBL from the nonimmune catfish tested ranged from 0.5 to 1.0%, indicating that the numbers of NK-like cytotoxic effectors and/or precursor cells in PBL were relatively low. This is consistent with previous observations of spontaneous cytotoxic responses in channel catfish, namely that E:T ratios ranging from 15:1 to 40:1 were usually required for 50% lysis of allogeneic target cells by freshly isolated PBL in 4-h 51 Cr-release assays (3,13,29). These observations imply that the spontaneous effector cell population(s) involved in lysis of any given allogeneic target likely ranges from 1 to 4% of the total PBL in the catfish employed in these studies.
This study did not formally demonstrate that the cells designated as catfish CTL employed TCR molecules to recognize target cells, nor were MHC molecules proven to be the structures recognized on target cells (30). However, this seems likely considering the strict specificity in target cell lysis displayed by the catfish CTL clones combined with the fact that these effector cells express TCR ␣␤ genes and the allogeneic target cells express both class I and II MHC genes (Refs. 8 and 9 and our unpublished data). Preliminary observations that blocking of target cell killing by some but not all alloantisera suggest that catfish CTL clones most likely recognize an alloantigen(s) on target cells (data not shown); whether this alloantigen is MHC class I, class II, or another molecule is not clear. Immunoprecipitation studies with such alloantisera have been equivocal and have not yet resolved this issue (31). Consequently, identification of the molecules on target and effector cells involved in target cell recognition by catfish CTL clones awaits the development of immune reagents specific for catfish TCR and MHC molecules.
The target cell recognition mechanism(s) employed by catfish NK-like effector clones was not determined in this study. However, one of the NK-like clones did not lyse autologous target cells but efficiently lysed three different allogeneic targets. It is possible that catfish NK-like cells express killer inhibitory receptors analogous in function to those found on human and rodent NK cells (32); engagement of self-MHC molecules on autologous/syngeneic target cells by killer inhibitory receptors on mammalian NK cells protects these cells from lysis by effectors (33)(34)(35). The availability of autologous and allogeneic clones of effector and target cells from catfish provides a promising model with which to study target cell recognition by fish NK-like cytotoxic cells at both functional and molecular levels.
Channel catfish alloreactive cytotoxic cells were shown to be distinct from NCC by several criteria. First, an NCC-specific mAb, 5C6, failed to block lysis of allogeneic target cells by PBL effectors (3). Second, channel catfish PBL stimulated in primary MLC lysed xenogeneic target cells poorly at E:T ratios that gave maximum lysis of allogeneic target cells (10). Finally, all of the NKlike as well as CTL clones tested did not react with mAb 5C6 (data not shown). Taken together, these results indicate that the various cytotoxic cells described herein are distinct from NCC.
The results described above provide the strongest evidence to date that channel catfish (and possibly other fish species) contain a variety of CTL and NK-like effector cells with different target cell preferences. The ability to clone various types of catfish cytotoxic cells along with the continued development of molecular probes and immunological reagents will likely facilitate study of both CTL and NK-like cells in fish as it has in mammals.