Somatic mutations of the mitochondrial genome in human colorectal tumours.

Alterations of oxidative phosphorylation in tumour cells were originally believed to have a causative role in cancerous growth. More recently, mitochondria have again received attention with regards to neoplasia, largely because of their role in apoptosis and other aspects of tumour biology. The mitochondrial genome is particularly susceptible to mutations because of the high level of reactive oxygen species (ROS) generation in this organelle, coupled with a low level of DNA repair. However, no detailed analysis of mitochondrial DNA in human tumours has yet been reported. In this study, we analysed the complete mtDNA genome of ten human colorectal cancer cell lines by sequencing and found mutations in seven (70%). The majority of mutations were transitions at purines, consistent with an ROS-related derivation. The mutations were somatic, and those evaluated occurred in the primary tumour from which the cell line was derived. Most of the mutations were homoplasmic, indicating that the mutant genome was dominant at the intracellular and intercellular levels. We showed that mitochondria can rapidly become homogeneous in colorectal cancer cells using cell fusions. These findings provide the first examples of homoplasmic mutations in the mtDNA of tumour cells and have potential implications for the abnormal metabolic and apoptotic processes in cancer.

Mismatch repair gene defects in sporadic colorectal cancers with microsatellite instability.

Microsatellite instability has been observed in both sporadic and hereditary forms of colorectal cancer. In the hereditary form, this instability is generally due to germline mutations in mismatch repair (MMR) genes. However, only one in ten patients with sporadic tumours exhibiting microsatellite instability had a detectable germline mutation. Moreover, only three of seven sporadic tumour cell lines with microsatellite instability had mutations in a MMR gene, and these mutations could occur somatically. These results demonstrate that tumours can acquire somatic mutations that presumably do not directly affect cell growth but result only in genetic instability. They also suggest that many sporadic tumours with microsatellite instability have alterations in genes other than the four now known to participate in MMR.

Evaluation of candidate tumour suppressor genes on chromosome 18 in colorectal cancers.

Chromosome deletions are the most common genetic events observed in cancer. These deletions are generally thought to reflect the existence of a tumour suppressor gene within the lost region. However, when the lost region does not precisely coincide with a hereditary cancer locus, identification of the putative tumour suppressor gene (target of the deletion) can be problematic. For example, previous studies have demonstrated that chromosome 18q is lost in over 60% of colorectal as well as in other cancers, but the lost region could not be precisely determined. Here we present a rigorous strategy for mapping and evaluating allelic deletions in sporadic tumours, and apply it to the evaluation of chromosome 18 in colorectal cancers. Using this approach, we define a minimally lost region (MLR) on chromosome 18q21, which contains at least two candidate tumour suppressor genes, DPC4 and DCC. The analysis further suggested genetic heterogeneity, with DPC4 the deletion target in up to a third of the cases and DCC or a neighbouring gene the target in the remaining tumours.

Mad-related genes in the human.

Resistance to the growth inhibitory effects of TGF-beta is common in human cancers. However, the mechanism(s) by which tumour cells become resistant to TGF-beta are generally unknown. We have identified five novel human genes related to a Drosophila gene called Mad which is thought to transduce signals from TGF-beta family members. One of these genes was found to be somatically mutated in two of eighteen colorectal cancers, and three of the other genes were located at chromosomal positions previously suspected to harbor tumour suppressor genes. These data suggest that this gene family may prove to be important in the suppression of neoplasia, imparting the growth inhibitory effects of TGF-beta-like ligands.

TGF-beta 1 is an autocrine-negative growth regulator of human colon carcinoma FET cells in vivo as revealed by transfection of an antisense expression vector.

Transforming growth factor-beta 1 (TGF-beta 1) has previously been implicated as a potential negative autocrine or paracrine growth regulator of certain cell types (Arteaga, C. L., R. J. Coffey, Jr., T. C. Dugger, C. M. McCutchen, H. L. Moses, and R. M. Lyons. 1990. Cell Growth & Differ. 1:367-374; Hafez, M. M., D. Infante, S. Winawer, and E. Friedman. 1990. Cell Growth & Differ. 1:617-626; Glick, A. B., K. C. Flanders, D. Danielpour, S. H. Yuspa, and M. B. Sporn. 1989. Cell Regulation. 1:87-97). This is based mainly on experiments assessing the effects of exogenous TGF-beta 1 or neutralizing antibodies to TGF-beta 1 on normal or tumor cell proliferation in vitro. However, direct evidence demonstrating such a negative regulation of tumor cell growth in vivo is still lacking. To overcome this problem we have constructed and used an antisense expression vector for TGF-beta 1 as a means of regulating endogenous TGF-beta 1 expression in tumor cells. Antisense-transfected FET human colon carcinoma cells showed a fivefold reduction in TGF-beta 1 mRNA and 15-fold reduction in TGF-beta 1 secretion. Antisense mRNA was detected in transfected cells by an RNase protection assay. Compared to control cells, cultured antisense-transfected cells showed a reduction in lag phase time rather than a change in doubling time. Cloning efficiencies of transfected cells were four times greater than control cells in anchorage-independent assays. Control cells did not form tumors at 5 x 10(5) in athymic nude mice. Antisense-transfected cells formed tumors in 40% of animals injected. At higher inocula (1 x 10(6) cells) antisense-transfected cells formed tumors in 100% of animals injected, but control cells still failed to form tumors. These results show that TGF-beta 1 acts as a negative growth regulator of human colon carcinoma cells in vivo as well as in vitro. Acquisition of partial or full resistance to such inhibitory effects may therefore contribute to tumor development and progression.

Suppression of human colorectal carcinoma cell growth by wild-type p53.

Mutations of the p53 gene occur commonly in colorectal carcinomas and the wild-type p53 allele is often concomitantly deleted. These findings suggest that the wild-type gene may act as a suppressor of colorectal carcinoma cell growth. To test this hypothesis, wild-type or mutant human p53 genes were transfected into human colorectal carcinoma cell lines. Cells transfected with the wild-type gene formed colonies five- to tenfold less efficiently than those transfected with a mutant p53 gene. In those colonies that did form after wild-type gene transfection, the p53 sequences were found to be deleted or rearranged, or both, and no exogenous p53 messenger RNA expression was observed. In contrast, transfection with the wild-type gene had no apparent effect on the growth of epithelial cells derived from a benign colorectal tumor that had only wild-type p53 alleles. Immunocytochemical techniques demonstrated that carcinoma cells expressing the wild-type gene did not progress through the cell cycle, as evidenced by their failure to incorporate thymidine into DNA. These studies show that the wild-type gene can specifically suppress the growth of human colorectal carcinoma cells in vitro and that an in vivo-derived mutation resulting in a single conservative amino acid substitution in the p53 gene product abrogates this suppressive ability.

Mutations of GTBP in genetically unstable cells.

The molecular defects responsible for tumor cell hypermutability in humans have not yet been fully identified. Here the gene encoding a G/T mismatch-binding protein (GTBP) was localized to within 1 megabase of the related hMSH2 gene on chromosome 2 and was found to be inactivated in three hypermutable cell lines. Unlike cells defective in other mismatch repair genes, which display widespread alterations in mononucleotide, dinucleotide, and other simple repeated sequences, the GTBP-deficient cells showed alterations primarily in mononucleotide tracts. These results suggest that GTBP is important for maintaining the integrity of the human genome and document molecular defects accounting for variation in mutator phenotype.

Growth stimulation by coexpression of transforming growth factor-alpha and epidermal growth factor-receptor in normal and adenomatous human colon epithelium.

Autocrine stimulation of the epidermal growth factor receptor (EGF-R), by coexpression of transforming growth factor-alpha (TGF-alpha), causes malignant transformation of some fibroblast cell lines. TGF-alpha and EGF-R are both known to be expressed in colon carcinoma tissue and have been shown coexpressed in colon carcinoma cell lines. TGF-alpha autocrine activation of EGF-R has been suggested as a potential mechanism contributing to abnormal growth control in colon cancer. We now report coexpression of TGF-alpha and EGF-R transcripts in morphologically normal colonic epithelium from five individuals, in colonic adenomas from three individuals, and in a nontumorigenic colon adenoma cell line, VACO-330. Functional studies demonstrate VACO-330 growth is stimulated by exogenous TGF-alpha and is completely abolished by a blocking anti-EGF-R antibody. Autocrine stimulation of EGF-R by TGF-alpha is therefore required for growth of the adenoma cell line. Autocrine stimulation of EGF-R by TGF-alpha does not cause malignant transformation of the colonic epithelial cell. In normal and adenomatous human colon TGF-alpha, via either an autocrine or paracrine mechanism, is likely an important physiologic stimulant of epithelial proliferation.

A benign cultured colon adenoma bears three genetically altered colon cancer oncogenes, but progresses to tumorigenicity and transforming growth factor-beta independence without inactivating the p53 tumor suppressor gene.

We describe the spontaneous progression of a colon adenoma cell line to tumorigenicity and growth factor independence. This system allows direct comparison of biologic stages of malignant progression with alterations of colon cancer suppressor genes and oncogenes. VACO-235, a human colon adenoma cell line, is at early passages nontumorigenic in the nude mouse, unable to grow in soft agar, growth stimulated by serum and EGF, and growth inhibited by TGF-beta. VACO-235 daughter passages 93 and higher have in culture spontaneously progressed to being weakly tumorigenic, but retain all other growth characteristics of VACO-235 early passages. A mouse xenograft from late passage VACO-235 was reestablished in culture as the granddaughter cell line, VACO-411. VACO-411 is highly tumorigenic, clones in soft agar, and is unresponsive to serum, EGF, and TGF-beta. Early passage VACO-235 bears a mutant K-ras allele, bears only mutant APC alleles, expresses no DCC transcripts, and expresses only wild type p53 transcripts. VACO-411 retains the identical genotype, still expressing only wild type p53. Colonic cells after ras mutation, APC mutation, and DCC inactivation remain nontumorigenic and growth factor dependent. Malignant progression involves at least two additional steps, and in VACO-411 can proceed by a novel pathway not requiring p53 inactivation.

Regulation of transforming growth factor alpha expression in a growth factor-independent cell line.

Aberrant transcriptional regulation of transforming growth factor alpha (TGF alpha) appears to be an important contributor to the malignant phenotype and the growth factor independence with which malignancy is frequently associated. However, little is known about the molecular mechanisms responsible for dysregulation of TGF alpha expression in the malignant phenotype. In this paper, we report on TGF alpha promoter regulation in the highly malignant growth factor-independent cell line HCT116. The HCT116 cell line expresses TGF alpha and the epidermal growth factor receptor (EGFR) but is not growth inhibited by antibodies to EGFR or TGF alpha. However, constitutive expression of TGF alpha antisense RNA in the HCT116 cell line resulted in the isolation of clones with markedly reduced TGF alpha mRNA and which were dependent on exogenous growth factors for proliferation. We hypothesized that if TGF alpha autocrine activation is the major stimulator of TGF alpha expression in this cell line, TGF alpha promoter activity should be reduced in the antisense TGF alpha clones in the absence of exogenous growth factor. This was the case. Moreover, transcriptional activation of the TGF alpha promoter was restored in an antisense-TGF alpha-mRNA-expressing clone which had reverted to a growth factor-independent phenotype. Using this model system, we were able to identify a 25-bp element within the TGF alpha promoter which conferred TGF alpha autoregulation to the TGF alpha promoter in the HCT116 cell line. In the TGF alpha-antisense-RNA-expressing clones, this element was activated by exogenous EGF. This 25-bp sequence contained no consensus sequences of known transcription factors so that the TGF alpha or EGF regulatory element within this 25-bp sequence represents a unique element. Further characterization of this 25-bp DNA sequence by deletion analysis revealed that regulation of TGF alpha promoter activity by this sequence is complex, as both repressors and activators bind in this region, but the overall expression of the activators is pivotal in determining the level of response to EGF or TGF alpha stimulation. The specific nuclear proteins binding to this region are also regulated in an autocrine-TGF alpha-dependent fashion and by exogenous EGF in EGF-deprived TGF alpha antisense clone 33. This regulation is identical to that seen in the growth factor-dependent cell line FET, which requires exogenous EGF for optimal growth. Moreover, the time response of the stimulation of trans-acting factor binding by EGF suggests that the effect is directly due to growth factor and not mediated by changes in growth state. We conclude that this element appears to represent the major positive regulator of TGF alpha expression in the growth factor-independent HCT116 cell line and may represent the major site of transcriptional dysregulation of TGF alpha promoter activity in the growth factor-independent phenotype.

Induction of nm23 gene expression in human colonic neoplasms and equal expression in colon tumors of high and low metastatic potential.

Levels of expression of the murine nm23 gene inversely correlate with metastatic potential in several rodent tumor model systems. Expression of the human nm23 homologue also is lower in human breast cancers of high metastatic potential than in breast cancers of low metastatic potential. In the present study, we examined changes in nm23 expression during colon carcinogenesis as found in 18 matched pairs of normal and neoplastic human colon tissues. We found that a 0.8-kilo-base nm23 transcript was expressed in all samples of morphologically normal colon mucosa. In 16 of 18 colon neoplasms, nm23 expression was further increased in the neoplastic, compared with the morphologically normal, colon mucosa from the same individual. Expression of nm23 was elevated over normal mucosa in 3 of 3 polyps, in 2 of 3 nonmetastatic cancers, and in 11 of the 12 cancers that were metastatic at the initial presentation. The levels of nm23 expressed were similar in the 12 metastatic colon neoplasms and in the 6 colon neoplasms of lower clinical stage. In addition, nm23 expression was maintained in culture in each of 12 cell lines initiated from human colon neoplasms and did not differ between lines established from neoplasms of high or low metastatic capability. We concluded that nm23 was expressed in normal colon mucosa. Expression of nm23 increased during early stages of colon carcinogenesis and remained increased in metastatic colon cancer. Therefore, in the colon, tissue-specific events dissociate nm23 expression from loss of tumor metastatic competence.

Effects of mitomycin on human colon carcinoma cells.

Subpopulations of malignant cells from primary cultures of human colon carcinoma were characterized with respect to their response to mitomycin (MMC). Growth inhibition assays indicated values of 2.06, 0.93, and 0.33 microM for the concentration of drug giving 50% inhibition of growth for sublines HCT 116b, HCT 116, and HCT 116a, respectively. Alkaline elution of filter-bound DNA from cells exposed to MMC in vitro showed a positive correlation between the amount of DNA cross-linking and growth inhibition as a function of drug concentration. Comparable DNA cross-linking was obtained at MMC concentrations of 10 microM for HCT 116b and 5 microM and HCT 116. The cross-linking of DNA from HCT 116a cells at 5 microM MMC was approximately equal to that from HCT 116 cells at doses between 10 and 20 microM MMC. Cross-link removal as a function of time after drug removal of MMC-treated cells was also measured. There was little difference in the rates of alkaline DNA elution after drug removal between HCT 116b and HCT 116a, suggesting that the ability to repair cross-links was not responsible for the differential sensitivities of the cells to MMC. The relative sensitivities of the subpopulations to MMC were reflected in vivo by MMC treatment of nude BALB/c mice bearing xenografts of the cultured sublines.

Synergistic antitumor effects of tumor necrosis factor and gamma-interferon on human colon carcinoma cell lines.

We assessed the antiproliferative effects of tumor necrosis factor (TNF-alpha) and gamma-interferon (IFN-gamma) alone and in combination, on nine human colon carcinoma cell lines. All were resistant (less than 30% inhibition) to TNF-alpha alone. Four cell lines were resistant to IFN-gamma alone, two exhibited a minimal degree of sensitivity (30-50% inhibition), one was moderately sensitive, and two were inhibited 70% or greater. A synergistic antiproliferative effect occurred in eight of the nine cell lines treated with a combination of TNF-alpha and IFN-gamma. In seven of these eight, the combination of cytokines resulted in 30-40% more growth inhibition than predicted had an additive interaction occurred (P less than 0.005). In two cell lines with an induced resistance to mitomycin C, an increase in resistance to combined TNF-alpha and IFN-gamma treatment correlated with an increasing resistance to mitomycin C. The data were further analyzed to determine if combination treatment altered the sensitivity of the cells to one or both agents in addition to synergistically potentiating growth inhibitory effects. Combinations of TNF-alpha/IFN-gamma enhanced the dose response activity of TNF-alpha in three cell lines (P less than or equal to 0.09) and decreased the dose response activity of IFN-gamma in another three (P less than or equal to 0.02). Colony forming experiments on HCT 116 cells demonstrated a reduction in the number of 250-micron colonies in the IFN-gamma/TNF-alpha treatment groups when compared to controls, indicating that combined treatment had a cytotoxic effect. We conclude that combination TNF-alpha/IFN-gamma treatment has a synergistic cytotoxic effect on human colon carcinoma cells. IFN-gamma may enhance the effectiveness of TNF-alpha in some cell lines, but not conversely. These results may have therapeutic implications.

Both transforming growth factor-beta and substrate release are inducers of apoptosis in a human colon adenoma cell line.

VACO-330, a nontransformed cell line established from a human colon adenoma, undergoes spontaneous apoptosis and shedding of cells into the culture medium. Shed cells were shown to be apoptotic, both by nuclear morphology and by generation of a typical "laddered" pattern of degraded DNA. Quantitation of DNA released into the medium, compared with the amount retained on the plate, demonstrated that 6.2 +/- 1.1% of the total cell mass underwent apoptotic death daily. The addition of transforming growth factor beta (20 ng/ml) accelerated this spontaneous apoptotic rate 3.2-fold. Moreover, apoptosis could be rapidly induced in up to 45% of the VACO-330 cells by using brief exposure to a calcium chelating medium to release the cells from the substratum. We suggest that transforming growth factor beta is a likely physiological regulator of apoptosis during maturation of the colonic epithelial cells. We additionally suggest the existence of an alternate pathway, which at the time of shedding from the crypt induces apoptosis in colonic epithelial cells that have escaped earlier apoptotic signals.

Enhancement of the sensitivity of human colon cancer cells to growth inhibition by acivicin achieved through inhibition of nucleic acid precursor salvage by dipyridamole.

This study was undertaken to determine if salvage of nucleic acid precursors might constitute a mechanism of resistance to acivicin in human colon cancer cells and, if so, to establish whether dipyridamole, an inhibitor of nucleoside and nucleobase transport, can block the salvage process and restore sensitivity to acivicin. Acivicin inhibited the replication of human colon cancer cells (VACO 5) in vitro in a dose- and time-dependent fashion. In addition, marked cell lysis was evident after a 24-hr exposure to acivicin at concentrations greater than 1 microgram/ml. The primary metabolic effect of acivicin was depletion of the cytidine triphosphate and guanosine triphosphate pools. Adenosine triphosphate levels were also reduced, but apparently as a consequence of the guanosine triphosphate depletion. VACO 5 cells exposed to acivicin (3 micrograms/ml) efficiently salvaged low levels (1 micron) of cytidine, guanosine, and guanine and could, therefore, restore the depleted nucleotide pools. The combination of cytidine and guanosine, but not either nucleoside alone, provided significant protection against the growth-inhibitory properties of acivicin. Dipyridamole, at a noncytotoxic concentration (5 microM), blocked repletion of the cytidine triphosphate and guanosine triphosphate pools in cells exposed to acivicin and the nucleic acid precursors. As a result, the growth-inhibitory effects of acivicin were maintained. The salvage of cytidine was particularly sensitive to inhibition by dipyridamole, and no restoration of cytidine triphosphate pools was evident. The cellular uptake of a variety of nucleic acid precursors was differentially sensitive to inhibition by dipyridamole. The 50% inhibitory dose values ranged from 0.01 to 2.5 microM for cytidine and uridine, respectively. The results of this study indicate that, although the replication of VACO 5 cells was inhibited by acivicin, low levels of nucleosides and nucleobases can circumvent the cytotoxicity. Dipyridamole effectively blocked the salvage pathways and restored the sensitivity of the cancer cells to the antiproliferative actions of acivicin.

Biochemical assessment of the effects of acivicin and dipyridamole given as a continuous 72-hour intravenous infusion.

Since this Phase I trial was based on a strategy of biochemical modulation, namely, the inhibition of nucleoside uptake by dipyridamole, a biochemical assessment of the actions of acivicin and dipyridamole was undertaken in order to aid our interpretation of the clinical findings. The primary biochemical objectives of this trial were: (a) to determine whether plasma levels of dipyridamole sufficient to inhibit nucleoside uptake could be achieved with a 72-h continuous i.v. infusion; (b) to monitor the effects of acivicin on two key enzymatic targets, CTP synthetase and GMP synthetase; and (c) to evaluate changes in cellular ribonucleoside triphosphate pools during therapy. Since peripheral blood mononuclear cells have relevant biochemical targets and can be serially obtained during the course of therapy, the biochemical effects of acivicin and dipyridamole were determined in these cells. At the maximally tolerated dose of dipyridamole (23.1 mg/kg/72 h), the steady-state concentrations of total and free dipyridamole averaged 11.9 microM and 27.8 nM, respectively. These levels were sufficient to inhibit cytidine (1 microM) uptake by greater than 50% in the lymphocytes of five of six patients so treated. Using lymphocytes obtained from 14 normal volunteers the concentration of free dipyridamole needed to inhibit the uptake of 1 microM cytidine by 50% averaged 13.8 +/- 1.1 nM. The plasma levels of alpha 1-acid glycoprotein, which tightly binds dipyridamole, ranged from 60 to 300 mg/dl in the patients in this study. As a consequence there were wide variations in the percentage of dipyridamole present as the unbound, pharmacologically active form and in the rates of dipyridamole clearance. The decreased rate of dipyridamole clearance seen in patients with high levels of alpha 1-acid glycoprotein resulted in higher plasma concentrations of total dipyridamole and compensated for the reduced fraction of free drug. Therefore, the plasma concentration of free dipyridamole varied much less than the total drug concentration in these patients. CTP synthetase and GMP synthetase activities were measured in patients' peripheral mononuclear cells prior to and at various times during therapy. CTP synthetase activity was inhibited in a time-dependent fashion by greater than 75% in seven of 13 evaluable courses; GMP synthetase was similarly inhibited in only three of ten cases. Ribonucleoside triphosphate pools were also measured in the patient's lymphocytes. CTP pool reductions of 30 to 50% were seen in nine of 19 courses, but in only four cases was the inhibition greater than 50%.(ABSTRACT TRUNCATED AT 400 WORDS)

Effects of BMY 25282, a mitomycin C analogue, in mitomycin C-resistant human colon cancer cells.

BMY 25282, a newly designed analogue of mitomycin C (MMC), was assessed for its non-cross-resistant cytotoxic and biochemical action against MMC-resistant human colon carcinoma cells. The analogue has an amidine substituted at position 7 of MMC and has a more efficient intracellular activation to its active species than MMC. In this study we demonstrated that BMY 25282 can overcome MMC resistance in a series of previously described human colon carcinoma cells resistant to MMC (Cancer Res., 44: 5880, 1984). The non-cross-resistance of the analogue in the model was confirmed in vivo by treating tumor xenograft-bearing athymic mice with equitoxic doses of MMC or BMY 25282. We further investigated the formation of interstrand DNA cross-link (IDC) formation by BMY 25282 and MMC. MMC-sensitive cells contained 3 to 8 times as many IDCs as resistant colon carcinoma cells, while no significant differences in IDCs were found between the MMC-sensitive or -resistant cells incubated with BMY 25282. When MMC-sensitive or -resistant cells were exposed to the 70% inhibition concentration of either MMC or BMY 25282, no differences were seen with respect to IDC formation. These studies demonstrate that BMY 25282 is able to overcome MMC resistance in a series of human colon carcinoma cells and that IDC formation in the MMC-sensitive or -resistant cells parallels cytotoxicity for both MMC and the analogue.

Paradoxical inhibition of solid tumor cell growth by bcl2.

The BCL2 gene product has been demonstrated to prevent apoptosis and provide a selective growth advantage to many cell types. We report an unexpected effect of bcl2 expression on the in vitro growth of several solid tumor cell lines. Expression of bcl2 in these cell lines resulted in growth inhibition similar to that seen with p53. In contrast, a COOH-terminal deletion mutant of bcl2 was unable to suppress growth. Thus, the bcl2 protein may exert distinct biological effects in different cell types.

Sensitive enzymatic cycling assay for glutathione: measurements of glutathione content and its modulation by buthionine sulfoximine in vivo and in vitro in human colon cancer.

We have measured glutathione content in small tissue samples derived from biopsies of primary and metastatic human colon tumors and from colon cancer cell lines in tissue culture and xenografts in athymic mice. Measurements were performed using an enzymatic cycling assay designed to quantitate extremely low levels of glutathione (GSH) (down to 10(-14) mol) from perchlorate extracts of tissue samples weighing less than 1 mg wet weight. Glutathione was stable in these acid extracts for at least 6 months when stored at -80 degrees C. A survey of normal tissues in mice, rats, and some human tissues showed considerable variation in GSH content of different tissues but generally similar levels were identifiable for the same tissues from different species. The highest GSH level was 56.9 nmol/mg protein in rat liver and the lowest was 1.8 nmol/mg protein in rat skeletal muscle. High GSH levels were also determined in mouse and human liver, while low GSH levels were detected in mouse muscle. Human colon cancer cell lines showed slightly higher GSH levels than did colon cancer tumor samples obtained from biopsies. These studies revealed a marked inter-individual difference in tumor GSH content, as well as a difference in GSH content between tumor deposits at different metastatic sites in the same individual. These results indicate the importance of direct tumor measurements of GSH content in clinical trials designed to modulate tumor glutathione content to try to increase sensitivity to chemotherapy or radiation therapy. Buthionine sulfoximine, an inhibitor of gamma-glutamyl cysteine synthetase, was shown to produce almost complete depletion of GSH in four different human colon cancer cell lines in 24 h. Buthionine sulfoximine was also shown to be capable of producing drastic depletion of GSH in human colon cancer grown as xenografts in athymic animals.