Dynamics of tumor cell killing by human T lymphocytes armed with an anti-carcinoembryonic antigen chimeric immunoglobulin T-cell receptor.

Chimeric immunoglobulin T-cell receptors (IgTCR) join the antigen-binding portion of an antibody to one of the signaling chains of the TCR. A previous report described the construction and functional testing of an IgTCR gene directed against the carcinoembryonic tumor antigen (CEA). These preclinical studies showed the proper assembly and cell surface expression of anti-CEA IgTCR molecules, specific target antigen binding, and activation of T-cell effector functions. Although IgTCR-modified T cells function well in vitro, therapeutic applications in humans may be complicated by various factors, such as the availability of appropriate T-cell cytokines, high systemic levels of antagonistic soluble CEA, and antigenic diversity in tumor cell populations. The current study analyzes tumor cell killing by IgTCR-modified human T cells under conditions that more closely model those that may be encountered in persons with cancer. This analysis shows that 1) depriving IgTCR-modified T cells of interleukin-2 does not diminish anti-CEA cytotoxic T lymphocyte activity, but does eliminate killing by lymphokine-activated killer cells; 2) high levels of soluble CEA do not significantly inhibit tumor cell killing even when approximately 80% of the chimeric receptors are blocked; and 3) CEA+ tumor cells that can down-regulate cell surface CEA evade immune destruction by IgTCR-modified T cells. These results have important implications for application strategies and protocol design considerations for early clinical testing of IgTCR anti-tumor therapies.

Targeting of T lymphocytes to melanoma cells through chimeric anti-GD3 immunoglobulin T-cell receptors.

Immunoglobulin T-cell receptors (IgTCRs) combine the specificity of antibodies with the potency of cellular killing by grafting antibody recognition domains onto TCR signaling chains. IgTCR-modified T cells are thus redirected to kill tumor cells based on their expression of intact antigen on cell surfaces, bypassing the normal mechanism of activation through TCR-peptide-major histocompatibility complex (MHC) recognition. Melanoma is one of the most immunoresponsive of human cancers and has served as a prototype for the development of a number of immunotherapies. The target antigen for this study is the ganglioside GD3, which is highly expressed on metastatic melanoma with only minor immunologic cross-reaction with normal tissues. To determine an optimal configuration for therapy, four combinations of IgTCRs were prepared and studied: sFv-epsilon, sFv-zeta, Fab-epsilon, Fab-zeta. These were expressed on the surface of human T cells by retroviral transduction. IgTCR successfully redirected T-cell effectors in an MHC-unrestricted manner, in this case against a non-T-dependent antigen, with specific binding, activation, and cytotoxicity against GD3+ melanoma cells. Soluble GD3 in concentrations up to 100 microg/ml did not interfere with recognition and binding of membrane-bound antigen. Based on the outcomes of these structural and functional tests, the sFv-zeta construct was selected for clinical development. These results demonstrate key features that emphasize the potential of anti-GD3 IgTCR-modified autologous T cells for melanoma therapies.

Coupling CD28 co-stimulation to immunoglobulin T-cell receptor molecules: the dynamics of T-cell proliferation and death.

Immunoglobulin T-cell receptor (IgTCR) molecules are potentially potent immune response modifiers because they allow T cells to bypass tolerance. Tolerance to self antigens has been one of the major barriers to the development of effective adoptive immunotherapies for treating cancer. In vitro studies in several laboratories have shown that cross-linking IgTCR molecules with the target antigen leads to cytolytic activity, cytokine release, and T-cell proliferation in model systems. However, many of these studies have used established T-cell lines rather than normal T cells or indirect assays of cytotoxicity, proliferation, and cytokine release. We have sought to establish the validity of these model systems while developing more effective adoptive immunotherapies using normal human T cells. In the present study the activation of T-cell proliferation after IgTCR cross-linking was evaluated. The results show that, in addition to IgTCR signals, CD28 costimulation is required to induce expansions of normal peripheral blood mononuclear cell-derived T cells. Signals from IgTCR alone can induce transient cell division, but they do not induce the prolonged polyclonal expansions that are characteristic of native immune responses. Very strong IgTCR signals could circumvent the CD28 requirement, but only at levels that are unlikely to be physiologically relevant. CD28 costimulation also suppressed the deletion of tumor-reactive subclones by activation-induced cell death. These studies confirm the importance of CD28 costimulation to the proliferation of IgTCR-modified human T cells, a key feature of an effective, reconstructed antitumor response.

Less repair of pyrimidine dimers and single-strand breaks in genes by scid cells.

Severe combined immunodeficient (Scid) mice have a mutation in the catalytic subunit of the DNA binding protein kinase that is involved in repair of double-strand breaks in DNA. To determine if the protein also influences repair of single-strand breaks, we examined the ability of Scid cells to repair lesions introduced by ultraviolet light and gamma-ray irradiation. DNA repair was measured both in total genomic DNA and in specific genes from murine Scid and wildtype fibroblast cell lines. The removal of pyrimidine dimers and repair of strand breaks in genes was measured using quantitative Southern blot analyses. After ultraviolet irradiation, there was no significant difference in the repair of photoproducts in bulk DNA between Scid and wildtype cells, as measured by cellular survival and unscheduled DNA synthesis. However, deficient repair was evident in genes, where Scid cells had 25-50% less repair in the c-myc and dihydrofolate reductase genes. After gamma-irradiation, Scid fibroblasts had 20-35% less repair of DNA breaks in immunoglobulin kappa and heavy constant genes than wildtype cells. The data suggest that intact DNA-PK enzyme is needed for the efficient operation of cellular repair of pyrimidine dimers and single-strand breaks in genes, as well as in its established role in rejoining double-strand breaks.

Bypassing immunization: optimized design of "designer T cells" against carcinoembryonic antigen (CEA)-expressing tumors, and lack of suppression by soluble CEA.

Tumor-associated antigens are typically nonimmunogenic in cancer patients, "immune surveillance" having manifestly failed. The fact that most tumor antigens are normal human proteins presents significant obstacles to current cancer immunization approaches that researchers are presently striving to overcome. An alternative strategy bypasses immunization altogether by direct genetic alteration of autologous patient T cells, to create "designer T cells" specific to a particular antigen. Chimeric immunoglobulin-T cell receptors (IgTCR) with a specificity for carcinoembryonic antigen (CEA) were created to evaluate the optimal IgTCR structure for cancer therapy. Antigen-binding domains of a humanized antibody were combined with TCR signaling chains to yield four different chimeric IgTCR: single chain Fv fragment (sFv)-zeta, fragment antigen-binding (Fab)-zeta, sFv-epsilon, and Fab-epsilon. All of the IgTCR were well expressed on T cells, and all showed specific binding and activation, as demonstrated by IL-2 production on contact with immobilized or cellular CEA, excepting sFv-epsilon alone which was inert solely against cellular targets for steric reasons unique to this construct. In contrast to prior studies of isolated TCR chains that related increased tyrosine-based activation motifs in zeta as a reason for superior signaling potency, these tests are the first to show that epsilon and zeta are indistinguishable for T cell signaling when assayed in the context of the intact TCR complex. Further, Fab was equivalent to sFv as an IgTCR component for expression and antigen binding, establishing an important alternative for IgTCR antigen recognition because sFvs may often lose antigen affinity. When IgTCR was expressed on normal human T cells, cytotoxic potency was demonstrated at low E:T ratios, with T cell recycling and progressive tumor cell destruction. Contrary to recent speculations, these observations prove that high affinity TCR interactions are not an impediment to serial target engagement and disengagement by cytotoxic T cells. The multivalent intercellular interactions of target cell binding, activation, and cytotoxicity were resistant to inhibition by soluble CEA. These studies establish a potentially important new immunotherapeutic modality for the treatment of CEA-expressing tumors.

Decoupling of DNA excision repair and RNA transcription in translocation breaksite regions of plasmacytoma-susceptible BALB/cAnPt mice.

Preferential repair of pyrimidine dimers in rodent cells is thought to be directly coupled to the RNA transcription machinery. The most compelling evidence for this notion is the finding that excision repair occurs more rapidly in the template strand of DNA of transcribed genes than in the non-template strand. A thorough test of this coupling concept by careful comparison of the rate of repair to the rate of transcription of a gene and its regulatory region has not been reported. In the present study, we used nuclear run-on as a measure of transcription in the c-myc and Pvt1 genes in normal B-lymphoblasts from plasmacytoma-susceptible (BALB/cAnPt) and plasmacytoma-resistant (DBA/2N) strains of mice. Previous studies have shown that these loci, but not c-abl or Dhfr are repaired differently in mouse strains: poorly in BALB/cAnPt but efficiently in DBA/2N. The results presented here indicate that in DBA/2N cells, run-on transcription from both DNA strands can be readily detected in the regions of c-myc and Pvt1 that were efficiently repaired. Unexpectedly, however, in BALB/cAnPt lymphoblasts, transcription was equivalent to that of DBA/2N, despite a dramatic reduction in efficiency of excision repair. This finding indicates that, in BALB/cAnPt lymphoblasts, DNA repair 5' to c-myc and in Pvt1 is decoupled from the RNA transcription machinery. We postulate that this dissociation of repair and transcription represents a BALB/cAnPt-specific defect in a component of the transcription/repair complex that specifically compromises repair activity but not transcription. This defect may be responsible for the inability of normal BALB/cAnPt lymphoblasts to repair DNA sequences in the c-myc 5' flank and the Pvt1 gene, inducing gene-specific instability that predisposes these loci to genetic accidents, including chromosomal translocation, retroviral integration and other mutations.

DNA repair in the endogenous and episomal amplified c-myc oncogene loci in human tumor cells.

We have studied the repair of u.v.-induced cyclobutane pyrimidine dimers (CPDs) in amplified c-myc oncogene loci in human colon cancer cells to better understand the relationship between chromatin structure, transcription and DNA repair. To assess the variation in DNA repair in the same gene whether located in a chromosomal site or in a extra-chromosomal site, we have quantitated the efficiency of excision repair after u.v. exposure in the endogenous and episomal c-myc genes isolated from COLO320HSR and DM cells. In the HSR cells, c-myc is localized in a homogeneously staining region (HSR), and in the DM cells, the gene is localized in double minute chromosomes (DM). Our results indicate that the repair is less efficient in c-myc amplicons organized as double minute chromosomes than in the endogenous c-myc amplicons. The episomal gene is not repaired with the same efficiency as when it is intrachromosomal. This may reflect differences in chromatin structure. An advantage of this biological system is that the cells possess two different alleles of the c-myc gene, one that is active and another which is inactive. We have studied the relationship between DNA repair and transcriptional activity in the c-myc locus by measuring the efficiency of excision repair after u.v. exposure in the normal and rearranged alleles of the c-myc gene. Surprisingly, the c-myc gene is repaired with similar efficiency in the highly transcribed allele as in the poorly expressed allele. However, u.v. damage is selectively removed from the transcribed strand of the active c-myc allele, but DNA repair is not strand specific in the non-expressed c-myc allele.

DNA repair defects associated with chromosomal translocation breaksite regions.

Using an assay that measures the removal of UV-induced pyrimidine dimers in specific DNA sequences, we have found that the Pvt-1, immunoglobulin H-C alpha (IgH-C alpha), and IgL-kappa loci are poorly repaired in normal B lymphoblasts from plasmacytoma-susceptible BALB/cAnPt mice. Breaksites in these genes are associated with the chromosomal translocations that are found in > 95% of BALB/cAnPt plasmacytomas. In contrast to those from BALB/cAnPt mice, B lymphoblasts from plasmacytoma-resistant DBA/2N mice rapidly repair Pvt-1, IgH-C alpha, and IgL-kappa. Further, (BALB/cAnPt x DBA/2N)F1 hybrids, which are resistant to plasmacytoma development, carry an efficient (DBA/2N-like) repair phenotype. Analysis of allele-specific repair in the IgH-C alpha locus indicates that efficient repair is controlled by dominant, trans-acting factors. In the F1 heterozygotes, these factors promote efficient repair of BALB/cAnPt IgH-C alpha gene sequences. The same sequences are poorly repaired in the BALB/cAnPt parental strain. Analysis of the strand specificity of repair indicates that both strand-selective and nonselective forms of repair determine repair efficiency at the gene level in nonimmortalized murine B lymphoblasts.

Repair of mitochondrial DNA after various types of DNA damage in Chinese hamster ovary cells.

Using methodology recently developed to assess gene-specific DNA repair, we have demonstrated that it is possible not only to study mitochondrial DNA repair, but also directly to compare mitochondrial and nuclear DNA repair in the same biological sample. Complex enzymatic mechanisms recognize and repair nuclear DNA damage, but it has long been thought that there was no DNA repair in mitochondria. Therefore, in an attempt to delineate more clearly which DNA repair mechanisms, if any, are functioning in mitochondria, we have investigated the repair of several specific DNA lesions in mitochondrial DNA. They include cyclobutane dimers, cisplatin intrastrand adducts, cisplatin interstrand crosslinks and alkali-labile sites. We find that pyrimidine dimers and complex alkylation damage are not repaired in mitochondrial DNA, and that there is minimal repair of cisplatin intrastrand crosslinks. In contrast, there is efficient repair of cisplatin interstrand crosslinks as evidenced by approximately 70% of the lesions being removed by 24 h. Additionally, there is efficient repair of N-methylpurines following exposure to methylnitrosourea with approximately 70% of the lesions being removed by 24 h. The results of these studies reveal that repair capacity of mitochondrial DNA damage depends upon the type of lesion produced by the damaging agent. We speculate that a process similar to the base excision mechanism for nuclear DNA exists for mitochondrial DNA but that there is no nucleotide excision repair mechanism to remove more bulky lesions in this organelle.

DNA repair in the c-myc proto-oncogene locus: possible involvement in susceptibility or resistance to plasmacytoma induction in BALB/c mice.

This report describes an unexpected difference in the efficiency of removal of UV-induced DNA damage in the c-myc locus in splenic B lymphoblasts from two inbred strains of mice. In cells from plasmacytoma-resistant DBA/2N mice, 35% of UV-induced damage in the regulatory and 5' flank of c-myc is removed by 12 h. However, in cells from plasmacytoma-susceptible BALB/cAn mice, damage is not removed from this region. In the protein-encoding region and 3' flank of c-myc as well as in two dihydrofolate reductase gene fragments, UV damage is repaired with similar efficiency in B lymphoblasts from both strains of mice. Furthermore, in the protein-encoding portion and 3' flank of c-myc, damage is selectively removed from only the transcribed strand. No repair is detected in the nontranscribed strand. In contrast, DNA repair in the 5' flank of c-myc is not strand specific; in DNA from DBA/2N cells, UV damage is rapidly removed from both the transcribed and nontranscribed strands. In BALB/cAn cells no repair was detected in either strand in the 5'flank, consistent with the results with double-stranded, nick-translated probes to this region of c-myc. In addition to the repair studies, we have detected post-UV-damage formation: in most of the genes studied, we find that additional T4 endonuclease-sensitive sites are formed in the DNA 2 h after irradiation. Our findings provide new insights into the details of gene-specific and strand-specific DNA repair and suggest that there may be close links between DNA repair and B-cell neoplastic development.

DNA damage induced by phorbol ester-stimulated neutrophils is augmented by extracellular cofactors. Role of histidine and metals.

Activated neutrophils cause extensive DNA damage in neighboring nonphagocytic cells. To determine whether compounds in the extracellular milieu participate in the DNA damage process, murine neutrophils were cocultivated with target tumor cells in media of varying composition. Using the alkaline elution assay, it was found that the level of strand breaks induced was significantly higher (2.8-fold) in complex cell culture media than in minimal phosphate-buffered saline. Addition of amino acids in general and of histidine in particular increased the level of damage nearly to that observed in complete media (2.7- and 2.1-fold, respectively). The histidine stimulation was concentration-dependent and reached a maximum at 100-400 microM. The mechanism whereby this occurred is not proven but probably derived from chelation of metals and participation in a site-specific Fenton reaction. Addition of the cell-impermeable chelator EDTA dramatically inhibited induction of strand breaks by neutrophils in complete media and prevented the enhancement of damage induced by histidine in phosphate-buffered saline. None of the effects on neutrophil-induced damage could be attributed to modulation of the oxidative burst activity of the cells (O2- and H2O2 production). Histidine also enhanced induction of strand breaks by reagent H2O2. However, EDTA had no effect or actually increased the level of damage induced by both a bolus of H2O2 and a flux of H2O2 generated by glucose oxidase. The cell-permeable chelator o-phenanthroline inhibited both neutrophil- and H2O2-induced damage. The results indicate that secondary reactions involving extracellular amino acids and metals contribute significantly to neutrophil-induced DNA damage to neighboring cells. Moreover, the data show that the mechanism whereby neutrophils induce this damage cannot be attributed solely to secretion of H2O2.

DNA repair in the c-myc locus.

We have studied DNA repair after UV damage in the murine c-myc locus. It appears that a region in B cells upstream of the murine c-myc gene is repaired with a different efficiency in plasmacytoma-resistant DBA/2N mice than in plasmacytoma-susceptible BALB/cAn mice. The region just upstream of c-myc is inefficiently repaired in B lymphoblasts derived from BALB/cAn mice. In contrast, this same region of c-myc is efficiently repaired in B lymphoblasts derived from DBA/2N mice. DNA fragments located in the coding region of c-myc and in another gene, dihydrofolate reductase (DHFR), are repaired with equal efficiency in cells from these two strains of mice. It is possible that repair efficiency of the 5' flank of c-myc may be involved in tumor susceptibility of the mouse strain.

Activated neutrophils induce prolonged DNA damage in neighboring cells.

We have measured the capacity of highly-purified, paraffin oil-elicited neutrophils to induce DNA single-strand breaks in a newly established plasmacytoma cell line, RIMPC 2304, which was induced by a retrovirus containing the c-myc and V-Ha-ras oncogenes. This cell line effectively repairs DNA damage induced by gamma-irradiation. DNA damage induced by neutrophils was correlated with the oxidative burst of the neutrophils. The levels of superoxide anion, H2O2, and HOCl produced after stimulation of the neutrophils (6 X 10(5)/cm3) with the tumor promoter phorbol myristate acetate (100 nM) were 33.8 microM, 12.8 microM and 1.7 microM respectively in 15 min, and 98 microM, 20 microM and 8.7 microM respectively in 90 min. The results of alkaline elution experiments revealed that when the same concentration of neutrophils was co-incubated for 15 min in serum-free medium with an equal number of radioactively labeled RIMPC 2304 cells, the latter incurred a level of damage that approximated that caused by 300 rad equivalents of gamma-irradiation or by a 1-min treatment with 20 microM H2O2 at 37 degrees C. Damage from neutrophils was coincident with the oxidative burst; it was induced rapidly (within 5 min) but remained high for more than 90 min. The level of damage achieved was dependent upon the ratio of neutrophils: target cells and was clearly detectable at ratios as low as 0.25:1. Induction of single-strand breaks was completely inhibited by catalase and partially inhibited by superoxide dismutase, mannitol, and reduced glutathione but not by Na azide. Addition of the non-steroidal anti-inflammatory drug indomethacin either enhanced (at 50 microM) or had no effect (at 2 microM) on the damage detected. Finally, repair of strand breaks induced by neutrophils was significantly slower (half-time approximately 10 min) than that observed for repair of similar levels of damage induced by H2O2 or gamma-irradiation (half-times approximately 3 min, each). The results indicate that neutrophils cause prolonged DNA damage in neighboring cells. Moreover, they indicate that although H2O2 produced in the oxidative burst is an essential mediator of the damage observed, additional reactive oxygen intermediates including the superoxide anion are also implicated. The data are discussed in relation to the possible role of neutrophils in chronic inflammation and in pristane-induced plasmacytoma formation in mice.