A soluble LAG-3 protein (eftilagimod alpha) and an anti-PD-L1 antibody (avelumab) tested in a phase I trial: a new combination in immuno-oncology.

BACKGROUND: Eftilagimod alpha (efti) is a major histocompatibility complex class II agonist activating antigen-presenting cells which leads to greater systemic type 1 T helper response and more cytotoxic CD8+ T-cell activation. This phase I trial evaluated the administration of efti, a soluble lymphocyte activation gene-3 (LAG-3) protein, combined with the anti-programmed death-ligand 1 (PD-L1) antibody avelumab in advanced solid tumors. PATIENTS AND METHODS: Patients with heavily pretreated metastatic solid tumors received intravenous avelumab (800 mg) combined with subcutaneously administered efti (6 or 30 mg) for up to 12 cycles, followed by avelumab monotherapy. The primary endpoint was the assessment of the recommended phase II dose (RP2D) of efti in combination with avelumab. RESULTS: Twelve patients with different tumor entities were enrolled (six patients in each cohort). During treatment, no dose-limiting toxicities occurred, and the severity of most adverse events was grade 1 or 2. In total, nine serious adverse events were documented, resulting in a fatal outcome in two cases, but none of them were assessed to be treatment related. Five patients (42%) achieved partial response. The median progression-free survival was 1.96 months and the median overall survival was not reached, with a 12-month survival rate of 75%. CONCLUSION: Subcutaneously administered efti plus avelumab was well tolerated, and efti of 30 mg was determined to be RP2D. The activity is promising and warrants further investigation in future phase II trials.

Antibody-mediated depletion of lymphocyte-activation gene-3 (LAG-3(+) )-activated T lymphocytes prevents delayed-type hypersensitivity in non-human primates.

Lymphocyte-activation gene-3 (LAG-3, CD223) is a marker for recently activated effector T cells. Activated T lymphocytes are of major importance in many autoimmune diseases and organ transplant rejection. Therefore, specifically depleting LAG-3(+) T cells might lead to targeted immunosuppression that would spare resting T cells while eliminating pathogenic activated T cells. We have shown previously that anti-LAG-3 antibodies sharing depleting as well as modulating activities inhibit heart allograft rejection in rats. Here, we have developed and characterized a cytotoxic LAG-3 chimeric antibody (chimeric A9H12), and evaluated its potential as a selective therapeutic depleting agent in a non-human primate model of delayed-type hypersensitivity (DTH). Chimeric A9H12 showed a high affinity to its antigen and depleted both cytomegalovirus (CMV)-activated CD4(+) and CD8(+) human T lymphocytes in vitro. In vivo, a single intravenous injection at either 1 or 0·1 mg/kg was sufficient to deplete LAG-3(+) -activated T cells in lymph nodes and to prevent the T helper type 1 (Th1)-driven skin inflammation in a tuberculin-induced DTH model in baboons. T lymphocyte and macrophage infiltration into the skin was also reduced. The in vivo effect was long-lasting, as several weeks to months were required after injection to restore a positive reaction after antigen challenge. Our data confirm that LAG-3 is a promising therapeutic target for depleting antibodies that might lead to higher therapeutic indexes compared to traditional immunosuppressive agents in autoimmune diseases and transplantation.

A novel subset of human lymphocytes with a T cell receptor-gamma complex.

We have previously characterized a CD3+ T cell receptor (TCR) alpha/beta- human fetal cloned cell line, termed F6C7, which surface-expresses a CD3-associated gamma chain identified by anti-NKFi, an mAb with a restricted clonotypic reactivity. Here, we have produced an additional antibody, anti-Ti-gamma A, which recognizes a public epitope of the gamma molecule defined by anti-NKFi. Ti-gamma A is present on approximately 3% of circulating lymphocytes with a wide range (1-15%) among 30 healthy individuals tested. Two-color immunofluorescence experiments performed with anti-Ti-gamma A and BMA 031 mAb (a reagent specific for the TCR-alpha/beta receptor) showed that surface expression of Ti-alpha/beta and Ti-gamma A is mutually exclusive. Moreover, it was found that most Ti-gamma A+ cells are CD2+, CD3+, CD4-, CD5+, NKH1-, HLA class II-negative. In contrast, the expression of the CD8 molecule on these T lymphocytes appears to be variable from one individual to another. Finally, we found that Ti-gamma A+ cells represent a majority of peripheral lymphocytes that express CD3 proteins but not the TCR-alpha/beta heterodimer. The delineation of this unique lymphocyte subset should help further studies on the biology of cells with a CD3-associated gamma complex.

A unique V-J-C-rearranged gene encodes a gamma protein expressed on the majority of CD3+ T cell receptor-alpha/beta- circulating lymphocytes.

We have recently described an mAb, anti-Ti gamma A, that recognizes an antigenic determinant carried by a TCR gamma chain. This antibody binds to approximately 3% of human PBLs and delineates a CD2+, CD3+, TCR-alpha/beta-, CD4-, CD8+/-, CD5+, NKH1-, and HLA class II- subset. The present study was designed to identify the gene encoding the Ti gamma A epitope. A first analysis was carried out on a previously characterized TCR gamma + fetal-cloned cell line termed F6C7. It was found that F6C7 cells have one gamma rearrangement on each chromosome: one joins V gamma 3 to J gamma 1, and the second joins V gamma 9 to J gamma P. Because only the latter allele appeared to be transcribed in the F6C7 lymphocytes, these data strongly suggested that anti-Ti gamma A mAb is specific for either a V gamma 9 or a V gamma 9-J gamma P-encoded peptide. To confirm this point, we studied an additional series of 13 randomly selected Ti gamma A+ cloned cells derived from peripheral blood of three distinct adult individuals. Each one of these lymphocytes was shown to both possess and transcribe a V gamma 9-J gamma P-C gamma 1-rearranged gene. It is therefore concluded that a predominant subpopulation of CD3+ TCR-alpha/beta- human circulating T lymphocytes (namely, the subset defined by anti-Ti gamma A mAb) surface expresses a gamma protein with a limited potential of variability from one cell to another.

T cell target 1 (TCT.1): a novel target molecule for human non-major histocompatibility complex-restricted T lymphocytes.

We have studied two gamma/delta T cell clones, E102 and E117, generated in a mixed lymphocyte culture using an allogeneic Epstein-Barr virus-transformed B cell line, E418. These clones were both found to express a molecular form of T cell receptor (TCR) infrequent in human peripheral blood, associating a V1-J1-C delta chain and a V3-JP2-C2 gamma chain. Functionally, they appeared as cytotoxic T lymphocytes (CTL) with non-major histocompatibility complex (MHC) (class I and II) requiring cytotoxicity, able to kill both the immunizing (i.e., E418) and unrelated (e.g., K562, REX, F601, and KAS) target cells. A monoclonal antibody, anti-10H3, able to selectively inhibit the cytotoxic activity of the clones has been produced. This reagent defines a 43-kD molecule, designated TCT.1, with broad distribution in the hematopoietic system, that appears to be distinct from class I MHC gene products. A series of functional experiments using various effector/target cell combinations strongly suggested that TCT.1 may represent a unique TCR ligand involved in the interaction between these particular CTL clones and certain of the target cells tested, while others were likely to be recognized and killed through a TCR-independent natural killer-like pathway. Although further experimentation will be needed to strengthen our interpretation of the present data, this study provides additional evidence that some T lymphocytes, in particular of the gamma/delta type, may interact specifically with target cells in a non-MHC class I/II-requiring fashion.

Further analysis of the T cell receptor gamma/delta+ peripheral lymphocyte subset. The V delta 1 gene segment is expressed with either C alpha or C delta.

In the present study, we have characterized the reactivity of two mAbs that are directed at the human TCR-gamma/delta. These reagents, designated anti-A13 and anti-TiV delta 2, were found to recognize antigenic determinants encoded by the TCR V delta 1 and V delta 2 gene segments, respectively. Immunofluorescence analyses performed with the antibodies confirmed that, in the TCR-gamma/delta+ cell subpopulation, the expression of V delta 2+ delta chains is largely predominant, as compared with the V delta 1+ counterparts. However, these experiments led to an apparently discrepant finding. Indeed, the total number of cells recognized by the anti-A13 plus the anti-TiV delta 2 antibodies was often greater than that detected with anti-TCR-delta 1, a reagent specific for a constant epitope of the human delta chain. Further investigation showed that the presence of a sizeable peripheral lymphocyte subset coexpressing the BMA031 and the A13 epitopes. Because the former antibody is known to recognize an invariant antigenic determinant of the TCR-alpha/beta dimer, these results suggested that the V delta 1 gene segment may be expressed with either C delta or C alpha. This hypothesis was confirmed using T2, an IL-2-dependent BMA031+ A13+ polyclonal cell line developed from peripheral blood of a healthy adult donor. Indeed, T2 cells were found to have productively rearranged the V delta 1 gene. Together, results of Northern blot analysis and cDNA cloning indicated that V delta 1 was expressed in these cells as part of a 1.6-kb full-length message including J alpha-C alpha segments.

Characterization of the lymphocyte activation gene 3-encoded protein. A new ligand for human leukocyte antigen class II antigens.

The lymphocyte activation gene 3 (LAG-3), expressed in human activated T and natural killer (NK) cells, is closely related to CD4 at the gene and protein levels. We report here the initial characterization of the LAG-3-encoded protein. We have generated two monoclonal antibodies after immunization of mice with a 30-amino acid peptide that corresponds to an exposed extra loop region present in the LAG-3 immunoglobulin-like first domain. The reactivity of these reagents is directed against LAG-3 since they recognize both membrane-expressed and soluble recombinant LAG-3 molecules produced in a baculovirus expression system. The two antibodies are likely to react with the same or closely related epitope (termed LAG-3.1) exposed on the LAG-3 first domain extra loop, as assessed in competition experiments on LAG-3-expressing activated lymphocytes. Cellular distribution analysis indicated that the LAG-3.1 epitope is expressed on activated T (both CD4+ and CD8+ subsets) and NK cells, and not on activated B cells or monocytes. In immunoprecipitation experiments performed on activated T and NK cell lysates, a 70-kD protein was detected after SDS-PAGE analysis. 45-kD protein species were also immunoprecipitated. Both the 70- and 45-kD proteins were shown to be N-glycosylated. In Western blot analysis, only the former molecule was recognized by the anti-LAG-3 antibodies, demonstrating that it is LAG-3 encoded. These anti-LAG-3 antibodies were used to investigate whether the LAG-3 protein interacts with the CD4 ligands. By using a high-level expression cellular system based on COS-7 cell transfection with recombinant CDM8 vectors and a quantitative cellular adhesion assay, we demonstrate that rosette formation between LAG-3-transfected COS-7 cells and human leukocyte antigen (HLA) class II-bearing B lymphocytes is specifically dependent on LAG-3/HLA class II interaction. In contrast to CD4, LAG-3 does not bind the human immunodeficiency virus gp120. This initial characterization will guide further studies on the functions of this molecule, which may play an important role in immune responses mediated by T and NK lymphocytes.

Sequences and diversity of human T cell receptor beta chain variable region genes.

The nucleotide sequences of 22 human T cell antigen receptor (TcR) beta chain variable region genes isolated from various T lymphocytes have been analyzed. Of the 19 variable gene segment (V beta)-containing sequences, 17 were unique. The V beta gene segments were grouped into 11 families. Comparisons were made with the data of Concannon et al. to unify the nomenclature. The data is consistent with a total V beta gene segment repertoire with a most probable value of 38 members and an upper bound of 104 members at the 95% confidence level. Southern blot data of germline DNA using selected TcR V beta cDNAs as probes support this estimate. The human repertoire is approximately three to four times greater than that reported for the mouse. Explanations for this discrepancy are proposed.

Surface expression of functional T cell receptor chains formed by interlocus recombination on human T lymphocytes.

Structural diversity of lymphocyte antigen receptors (the immunoglobulin [Ig] of B cells and the alpha/beta or gamma/delta T cell receptor [TCR] of T cells) is generated through somatic rearrangements of V, D, and J gene segments. Classically, these recombination events involve gene segments from the same Ig or TCR locus. However, occurrence of "trans" rearrangements between distinct loci has also been described, although in no instances was the surface expression of the corresponding protein under normal physiological conditions demonstrated. Here we show that hybrid TCR genes generated by trans rearrangement between V gamma and (D) J beta elements are translated into functional antigen receptor chains, paired with TCR alpha chains. Like classical alpha/beta T cells, cells expressing these hybrid TCR chains express either CD4 or CD8 coreceptors and are frequently alloreactive. These results have several implications in terms of T cell repertoire selection and relationships between TCR structure and specificity. First, they suggest that TCR alloreactivity is determined by the repertoire selection processes operating during lymphocyte development rather than by structural features specific to V alpha V beta regions. Second, they suggest the existence of close structural relationships between gamma/delta and alpha/beta TCR and more particularly, between V gamma and V beta regions. Finally, since a significant fraction of PBL (at least 1/10(4)) expressed hybrid TCR chains on their surface, these observations indicate that trans rearrangements significantly contribute to the combinatorial diversification of the peripheral immune repertoire.

LAG-3, a novel lymphocyte activation gene closely related to CD4.

We have identified a novel human gene of the Ig superfamily, designated LAG-3. Expression of this gene is undetectable in resting PBL, while it is found (a 2-kb message) in activated T and NK cells. The LAG-3 gene includes eight exons; the corresponding cDNA encodes a 498-amino acid membrane protein with four extracellular IgSF domains. The first one belongs to the V-SET; it is particular since it includes an extra loop in the middle of the domain and an unusual intrachain disulphide bridge. The three other domains belong to the C2-SET. Strong internal homologies are found in the LAG-3 molecule between domains 1 and 3, as well as between domains 2 and 4. It is therefore likely that LAG-3 has evolved by duplication of a pre-existing gene encoding a two IgSF-domain structure. The compared analysis of LAG-3 and CD4, with respect to both their peptidic sequence as well as their exon/intron organization, indicated that the two molecules are closely related. This point is strengthened by the finding that both genes are located on the distal part of the short arm of chromosome 12.

Clonal T cell expansion induced by interleukin 2 therapy in blood and tumors.

In a phase I clinical trial on the effects of preoperative adjuvant IL-2 therapy given to patients undergoing hepatic resection of colorectal adenocarcinoma metastases, we monitored the putative induction of T cell clonal expansion in both tissues and blood. The presence of T cell clonotypes was analyzed with a PCR-based method that determines V-D-J junction size patterns in T cell receptor (TCR) V beta subfamilies in samples before and after a 5-d IL-2 infusion. This high resolution method analyzing CDR3 sizes of TCR transcripts was used in conjunction with FACS analysis of the corresponding T cell subpopulations with TCR V beta-specific mAb. At time of surgery (day 8 after starting IL-2), we found in the three patients analyzed with V beta-C beta primers multiple dominant T cell clonotypes in the tumor and peritumoral tissues which had probably expanded as a result of therapy. In three control patients not treated with IL-2, multiple oligoclonal patterns were not observed with this set of primers. In the fourth control patient a unique V beta 21-C beta CDR3 pattern which corresponds to two dominant clonotypes was found in the tumor. The same dominant clonotypes identified in the tumor after IL-2 were also detectable in the blood and comparison of the profiles obtained before and after IL-2 therapy indicates that they were induced by IL-2. The relative expansion of the corresponding T cell subpopulations was maintained for varying periods of time after surgery (4-7 d and almost 2 yr in one case). Together, these results indicate that IL-2 induces marked expansion of several T cell clones. Systemic IL-2 administration may represent, either alone or as a vaccine adjuvant, an appropriate way of boosting antigen-specific immune responses.

Direct evidence to support the immunosurveillance concept in a human regressive melanoma.

The concept of immunosurveillance against cancer has been an extensively debated question over the last decades. Multiple indirect arguments have supported the view that the immune system may control, at least in certain cases, malignant cell growth while direct demonstration is still lacking in the human. In an attempt to address this issue, we have selected a study model, namely spontaneously regressive melanoma. In previous series of experiments, the variability of T cell receptors (TCRs) in the lymphocytes infiltrating a regressive tumor lesion was investigated. Results demonstrated that clonal T cell populations, precisely defined through their V-D-J junctional sequences, were amplified in situ. One clone was predominant, expressing the V beta 16 variable gene segment. A specific anti-V beta 16 TCR mAb was generated here to purify and functionally characterize the corresponding cells. A tumor-infiltrating lymphocyte-derived V beta 16+ T cell line was developed using this reagent. These in vitro cultured cells were found to express the in vivo predominant TCR sequence exclusively and to display an HLA-B14-restricted cytotoxic activity against the autologous tumor cells. Immunohistochemical experiments, performed with the anti-V beta 16 mAb, showed that the corresponding CTLs are present in the tumor area, some of them being closely opposed to the melanoma cells. Together, these studies demonstrate the existence of a local adaptive immune response clinically associated to tumor regression, thus strongly supporting the validity of the immunosurveillance concept in certain human tumors.

Soluble CD30 and lymphocyte activation gene-3 (CD223), as potential serological markers of T helper-type cytokine response induced by acellular pertussis vaccine.

T cell responses are involved in vaccine-induced immunity to pertussis but no easy-to-monitor, serological markers are available to assess these responses. The lymphocyte activation gene-3 (CD223) molecule is present on, and released by, activated T helper (Th) 1 cells, whereas CD30 molecules have been associated with Th2 immune responses. Starting from the recent knowledge of the cytokine profile induced by pertussis vaccination, we examined the levels of soluble (s)CD223 and sCD30 proteins in child recipients of acellular pertussis (aP) and diphtheria-tetanus (DT) vaccines and in children receiving DT vaccine only, as control. The correlation of the two proteins with specific antibody and T cell responses was assessed. The main findings are: i) sCD223 and sCD30 levels are inversely related, suggesting that the two markers are the expression of different and counter-regulated T-cell responses; ii) sCD30 level correlated with induction of T cell proliferation to pertussis vaccine antigens and antibody response to pertussis toxin. Overall, sCD30 and sCD223 levels seem to be promising candidate markers to assess the induction of Th-type responses in vaccine recipients.

Evidence for in situ amplification of cytotoxic T-lymphocytes with antitumor activity in a human regressive melanoma.

We have derived from lymphocytes infiltrating a human regressive melanoma lesion a series of T-cell receptor alpha/beta-dependent, HLA-B14-restricted cytotoxic T-lymphocyte clones reactive against the autologous tumor. Analysis of the T-cell receptor gene expression revealed that all the clones represented a unique cell expressing a V beta 13.1/J beta 1.1 gene segment. T-cell receptor transcripts expressed in the cloned cells were compared to those present in the uncultured tumor tissue. This analysis demonstrated that the specific cytotoxic T-lymphocyte clones characterized in vitro was actually selected and amplified in vivo at the lesion site. These results provide strong evidence that effector T-cells have contributed to tumor regression.

T-cell receptor repertoire in neuroblastoma patients.

Spontaneous regression of widespread lesions is a characteristic feature of neuroblastoma. One may postulate that the immune response contributes to these clinical regressions. Accordingly, we studied the T-cell receptor (TCR) repertoire of tumor-infiltrating lymphocytes in eight neuroblastoma tumors. The expression of 29 V alpha and 24 V beta gene segment subfamily specificities was analyzed by PCR and compared by computerized densitometry of Southern blots to values obtained in the blood. Overall, the TCR repertoire of these eight patients was diverse, with virtually all V alpha and V beta specificities expressed. Nonetheless, four of these patients showed V beta 2 gene segment subfamily overexpression in the tumor corresponding to local expansion of polyclonal T-cell subpopulations. In one patient, this expansion could be due to local secretion of superantigenic activity, as suggested by the specific stimulation of murine T cells expressing a human V beta 2 chain by supernatant of the corresponding neuroblastoma cell line. In addition, high-resolution analysis of the TCR beta transcript complementarity-determining region 3 sizes identified three patients (of six studied) with marked clonal T-cell expansion in the tumor not seen in the blood. The specific expression of several dominant clono-types in the tumor may be related to the recognition of neuroblastoma-specific antigens in these patients. Together, these results on the TCR repertoire expressed in vivo may lead to the characterization of putative immune response mechanisms (i.e., antigen- or superantigen-driven stimulation) which participate in tumor regression.

Analysis of T-cell receptor alpha/beta variability in lymphocytes infiltrating a melanoma metastasis.

Multiple experimental and clinical studies have suggested that the immune system may, to some extent, control the development of melanomas. The presence of tumor-infiltrating lymphocytes could reflect an in situ immune reaction directed to the malignant cells. The characterization of T-cell receptor (TCR) expressed by tumor-infiltrating lymphocytes is one way to precisely analyze these local T-cell responses. In this study, we have assessed the TCR alpha/beta variability in tumor-infiltrating lymphocytes from a subcutaneous metastasis of a melanoma patient. Using the anchored-polymerase chain reaction 268 TCR alpha and 266 TCR beta chain transcripts have been cloned and sequenced. Their analysis shows that the T-cell infiltrate is extremely diverse, with no preferential TCR gene segment usage.

Evidence for T-cell clonal expansion in a patient with squamous cell carcinoma of the head and neck.

Squamous cell carcinomas of the head and neck (SCCHN) often contain a large mononuclear cell infiltrate, composed mainly of T-lymphocytes which could reflect an in situ immune reaction against the malignant SCCHN cells. We analyzed the molecular structure of the T-cell receptor (TCR) alpha and beta chains expressed by lymphocytes in the tumor, peripheral blood (PBMC), and in the peritumoral tissue in six cases of localized SCCHN. We first determined V alpha and V beta gene segment subfamily usage by polymerase chain reaction using a panel of V subfamily-specific oligonucleotide primers (V alpha 1-w29 and V beta 1-w24). An apparently unrestricted usage of V alpha or V beta gene segment subfamilies was observed for all three samples from the six cases examined. No major difference was observed in T-cell receptor repertoire expression of PBMC between SCCHN and healthy donors, making unlikely the expression of putative tumor-related superantigen(s) in these patients. Intersample comparisons for a given V alpha or V beta T-cell receptor specificity revealed some differences in V gene segment usage in tumor-infiltrating lymphocytes versus PBMC. A detailed analysis of these V segment subfamily specificities (cloning followed by sequencing and CDR3 size distribution analysis) in one patient revealed the presence of recurrent T-cell receptor transcripts (i.e., identical V-N-J sequences) in the tumor (e.g., 40%) and in PBMC (e.g., 75%). These results show that unique T-cell subpopulations are clonally amplified in SCCHN patients, possibly as a consequence of antigen-driven selection.

In vivo local expansion of clonal T cell subpopulations in renal cell carcinoma.

Renal cell carcinoma (RCC) is one human tumor to which the immune response may control the growth of tumor cells. These tumors are infiltrated by a large mononuclear infiltrate mainly composed of T lymphocytes. To characterize the lymphocytes infiltrating RCC, we analyzed the molecular structure of the T cell receptor (TCR) alpha and beta chains in tumor and paired peripheral blood lymphocytes from a series of 6 untreated patients. We first determined V alpha and V beta gene segment usage by PCR using a panel of V specific oligonucleotide primers (V alpha 1-w29 and V beta 1-w24). A highly diverse usage of TCR V alpha and V beta gene usage was observed in 5 of 6 tumors. In addition, the few tumor overexpressed V beta specificities detected by reverse transcription-PCR were shown to contain minor T cell expansions. Strikingly, 1 of the 6 tumor studied displayed a skewed TCR repertoire with V beta 4 transcript representing 25% of the TCR signals. Clonality of the tumor overexpressed V beta transcripts was analyzed by CDR3 size distribution analysis. In the particular tumor displaying a biased repertoire large expansions of T cell subpopulations were detected (particularly in V beta 4) specifically at the tumor site. Such T cells may be expanded locally in response to tumor antigens.

TCR gene segments from at least one third of V alpha subfamilies rearrange at the delta locus.

Using PCR and an experimentally validated V alpha subfamily-specific oligonucleotide panel (V alpha 1-w29), we have investigated whether the TCR delta chain may increase its combinatorial diversity by using V genes considered as alpha chain-specific. We show that at least 10 distinct human V alpha segments rearrange at the J delta locus, leading to scrambling of the two V gene repertoires. Fifty-five per cent of the V alpha/J delta transcripts characterized here were in frame. The 17 V alpha/C delta chains analysed included an extended CDR3 region with up to 18 aa encoded by the junctional region. In addition, a new J delta segment (J delta 4) has been characterized. Together, these findings demonstrate that combinatorial diversity in the human delta locus is larger than previously thought.

Alternatively spliced T cell receptor transcripts expressed in human T lymphocytes.

We used the anchored-polymerase chain reaction (A-PCR) procedure to study human TCR transcripts derived from a variety of polyclonal T cell populations. In this series of experiments, 31 'unusual' cDNAs, which do not include exclusively V-J-C, J-C or 5'C genomic sequences, were identified. Ten of these were found to represent distinct types of alternatively spliced TCR alpha transcripts whose structure is derived from unusual splicing of one, two or even three intervening intronic sequences. The splicing events led to either conservation of a novel exon in the mRNA structure (designated aE1 alpha-aE5 alpha) between the V-J and C segments or to deletion of the 3' V region-J segment. In three cases, the alternatively spliced exons (aE1 alpha-aE3 alpha) interrupt the open translational reading frame of the corresponding V-J alpha segment. Nineteen and two cDNA represent sterile C beta or C delta transcripts, respectively. Their structures are derived from the conservation of a non-translatable exon, aE1 beta or aE1 delta, which is precisely spliced at the 5' end of the corresponding C exon sequences. Interestingly, the 3' region of the aE1 beta sequence is homologous to the murine C beta 0 exon. Together, these results led to the characterization of nine novel exons in the TCR alpha, beta and delta loci.