Glutathione transferase activity and isoenzyme composition in primary human breast cancers.

The human glutathione transferases (GSTs) are a multigene family of detoxication enzymes with patterns of expression that are both tissue specific and genetically determined. Changes in the levels of one or more GST isoenzymes have been associated with the development of anticancer drug resistance in cultured cell lines. In this study, total GST activity and GST isoenzyme composition have been determined for 45 primary human breast carcinomas using a 1-chloro-2,4-dinitrobenzene substrate assay and Western blotting, respectively. The GST activity ranged from 5-208 mU/mg protein with a mean of 67 mU/mg protein (+/- 44 SD). GST-pi) isoenzyme protein was detectable on Western blots in 44 of 45 samples. Mu Class GST protein was detected in 18 of 38 samples and undetectable in 20 of the 38 samples tested. By polymerase chain reaction analysis of genomic DNA, the absence of mu class GST in breast tumors was determined to be due to the deletion of the gene for GST-mu in the DNA of those tumors. None of the 43 primary human breast cancer samples tested contained detectable alpha class GST protein. Neither the total GST activity of tumor samples, the quantity of GST-pi protein, nor the presence or absence of mu class GST correlated with other factors known to be of prognostic significance including tumor size, nodal status, estrogen receptor protein positivity, or progesterone receptor protein positivity. Substantial differences exist among primary breast carcinomas in both the amount of GST activity and GST isoenzyme composition. However, these are not tightly linked either to tumor stage or to hormone receptor status. Whether the levels of these enzymes are independent predictors of either risk of recurrence or response to anticancer therapy has yet to be tested directly.

Killing Epstein-Barr virus-positive B lymphocytes by gene therapy: comparing the efficacy of cytosine deaminase and herpes simplex virus thymidine kinase.

Epstein-Barr virus (EBV)-positive lymphomas are frequent among immunosuppressed patients. We have examined the feasibility of killing EBV-immortalized B lymphocytes by gene transfer involving the use of "suicide" genes whose expression in target cells renders them susceptible to killing by a prodrug. We examined two gene/prodrug pairs: the Escherichia coli cytosine deaminase (CD) gene with the prodrug 5-fluorocytosine (5-FC), and the herpes simplex virus thymidine kinase (HSV-TK) gene with the prodrug ganciclovir. Retroviral vectors and drug selection were used to obtain CD or HSV-TK expression in cells. Both the CD/5-FC and the HSV-TK/ganciclovir combinations yielded substantial killing of EBV-immortalized B lymphocytes in vitro, although the CD/5-FC regimen had a significantly greater therapeutic margin than the HSV-TK/ganciclovir combination. The CD/5-FC pair, but not the HSV-TK/ganciclovir pair, was shown to have a "bystander killing effect" in vitro. When only 30% of the cells expressed the suicide gene, scid mouse tumors regressed in both the CD/5-FC regimen and the HSV-TK/ganciclovir regimen, documenting an in vivo bystander effect with both regimens. However, a greater percentage of tumors completely regressed with the CD/5-FC regimen. Overall, the sum of our data indicates that the CD/5-FC combination is the more promising regimen for treatment of EBV-associated lymphomas in vivo.

Multiple integrations of human foamy virus in persistently infected human erythroleukemia cells.

Foamy viruses are complex retroviruses whose replication strategy resembles that of conventional retroviruses. However, foamy virus replication also resembles that of hepadnaviruses in many respects. Because hepadnaviruses replicate in an integrase-independent manner, we were interested in investigating the characteristics of human foamy virus (HFV) integration. We have shown that HFV requires a functional integrase protein for infectivity. Our analyses have revealed that in single-cell clones derived from HFV-infected erythroleukemia-derived cells (H92), there were up to 20 proviral copies per host cell genome as determined by Southern blot and fluorescent in situ hybridization analysis. Use of specific probes has also shown that a majority of the proviruses contain the complete tas gene, which encodes the viral transactivator, and are not derived from Deltatas cDNAs, which have been shown to arise rapidly in infected cells. To demonstrate that the multiple proviral sequences are due to integration instead of recombination, we have sequenced the junctions between the proviral sequences and the host genome and found that the proviruses have authentic long terminal repeat ends and that each integration is at a different chromosomal site. A virus lacking the Gag nuclear localization signal accumulates fewer proviruses, suggesting that nuclear translocation is important for high proviral load. Since persistently infected H92 clones are not resistant to superinfection, the relative importance of an intracellular versus extracellular mechanism in proviral acquisition has yet to be determined.

Isolation and analysis of the gene and cDNA for a human Mu class glutathione S-transferase, GSTM4.

We have isolated the human gene and cDNA that encode the fourth member of the Mu class glutathione S-transferases (GSTs). The complete gene sequence (GenBank M96233) and the cDNA sequence (M96234) are reported and named GSTM4. The gene, located on chromosome 1, is comprised of eight exons with an organization similar to that of other Mu class genes. GSTM4 differs from a partial gene sequence, termed GSTmu2 (Taylor, J. B., Oliver, J., Sherrington, R., and Pemble, S. E. (1991) Biochem. J. 274, 587-593), in exons 3 through 5 by only 11 nucleotides within introns. The cDNA for GSTM4 was isolated from a library derived from the cervical carcinoma cell line, HeLa. Northern blots demonstrate the presence of GSTM4 mRNA in human heart, placenta, lung, brain, liver, skeletal muscle, pancreas, testis, cerebral cortex, uterus, ovary, a lymphoblastoid cell line, and four carcinoma cell lines. The deduced amino acid sequence of GSTM4 is 87% (GSTM1), 83% (GSTM2), and 70% (GSTM3) identical to the previously described human Mu class GSTs. The polypeptide encoded by the GSTM4 cDNA, produced in a bacterial expression system, conjugates 1-chloro-2,4-dinitrobenzene to GSH with a specific activity of 1.39 +/- 0.21 mumol/min/mg.

A comparison of the enzymatic and physicochemical properties of human glutathione transferase M4-4 and three other human Mu class enzymes.

The multigene family of cytosolic glutathione S-transferases (GSTs) consists of four classes (Alpha, Mu, Pi, and Theta), all involved in the detoxication of reactive electrophiles. The human Mu class GSTs consist of at least four expressed isozyme subunits, GST M1, GST M2, GST M3, and GST M4, which have 70-90% amino acid sequence identity. The gene and cDNA sequences for GST M4 have been determined recently (K. E. Comstock, K. J. Johnson, D. Rifenbery, and W. D. Henner, J. Biol. Chem. (1993) 268, 16958-16965). Cloning of GST M4 cDNA into an Escherichia coli expression system permitted the production of the corresponding protein. The enzyme was purified and shown to have a relatively low specific activity with the standard GST substrate 1-chloro-2,4-dinitrobenzene (1.4 +/- 0.2 mumol min-1 mg-1 protein), but an activity equivalent to other Mu class enzymes with other tested substrates. The protein forms functional dimers composed of subunits with a M(r) of approximately 26,400. A detailed comparison of the activity with various substrates and inhibitors was performed between GST M4-4 and other human Mu class GSTs, GST M1a-1a, GST M2-2, and GST M3-3, produced in bacterial expression systems. Despite the high level of amino acid sequence identity, the enzymatic properties of these enzymes were quite different. Comparisons with the crystallographic structure of a homologous rat GST, GST 3-3, indicate that a number of the nonconserved amino acid residues can be assigned to the putative active site of GST M4-4. This suggests that diversification in the evolution of these genes has occurred primarily in the substrate binding regions to cope with an increasing variety of foreign compounds.