Diminished DNA Repair and Elevated Mutagenesis in Mammalian Cells Exposed to Hypoxia and Low pH¹

Genetic instability is a hallmark of malignancy (1). As cancers develop, they often acquire an increasing number of genetic alterations, manifested both at the chromosomal level and at the DNA level (2, 3, 4). Normal cells have multiple mechanisms to recognize and repair DNA damage before mutations arise(5), serving to maintain the integrity of the genome. As a result, the high frequency of mutations found in cancer cells cannot be accounted for by the low spontaneous mutation rate observed among somatic cells. This has led to the hypothesis proposed by Loeb(6) that cancer cells exhibit a mutator phenotype and that the loss of genome stability function(s) early during tumor development could produce an increase in mutation rate in the affected cells. This notion has gained support from work establishing an association of genetic defects that lead to genome instability with cancer predisposition. For example, inherited mutations in genes associated with DNA mismatch repair have been identified as the underlying cause of hereditary nonpolyposis colon carcinoma (7, 8).

In addition to the endogenous genetic factors, growing evidence also suggests that the exogenous environment within solid tumors itself may be mutagenic and constitute a significant source of genetic instability(9). The tumor microenvironment is characterized by regions of fluctuating and chronic hypoxia, low pH, and nutrient deprivation (10, 11). This microenvironmental heterogeneity develops very early in the growth of solid tumors due to inadequate blood supply (10). Both in vivo and in vitro data demonstrate that exposure of cells to these adverse conditions can lead to genome alterations. For example, several groups observed an increased frequency of drug resistance among cells transiently exposed to hypoxia or low pH (12, 13, 14), effects attributed to amplification at loci encoding the drug resistance genes. Besides large chromosomal changes, small deletions and point mutations have also been found to arise in cells treated with hypoxia(15). Furthermore, hypoxia, acidosis, and glucose starvation all appear to enhance the metastatic potential of tumor cells, a process that may be associated with induced genetic changes(14, 16, 17, 18).

In vivo studies have taken advantage of transformed cell lines that are capable of forming tumors in immune-deficient mice. We previously established an animal model in which a mouse cell line carrying a chromosomally based λsupF phage shuttle vector was implanted s.c. into the flanks of nude mice to generate tumors(15). We found that the frequency of supFreporter gene mutations arising in cells within the tumors was 5-fold higher than that in otherwise identical cells grown in culture. In another study also using tumor xenografts, Wilkinson et al.(19) reported that the mutation frequency in the hprt gene in cells explanted from experimental tumors was severalfold higher than that found in cells grown in culture for an equivalent period of time.

However, the mechanism by which the tumor microenvironment induces genetic instability has not been established. We previously proposed that the abnormal physiology accompanying a developing tumor may play a role (15). Indeed, several features of the tumor microenvironment have been shown to cause severe disturbances in cell metabolism and function. The effects of hypoxia have been most extensively characterized: after initiation of hypoxia, cellular ATP levels decrease rapidly (20);G₁-phase cell cycle checkpoints become activated(21); and DNA replication is inhibited (20, 22). In addition, protein synthesis decreases, and protein degradation increases, with relatively enhanced synthesis of oxygen-related proteins (20, 23, 24). Unphysiological pH can alter the structure and function of cellular proteins, including DNA polymerases (25, 26). In addition, it was demonstrated that cells deprived of serum exhibit higher levels of intracellular oxidants compared with the levels in control cells (27). Hence, it is conceivable that such profound perturbations in cell physiology may lead to conditions that either cause increased spontaneous damage to DNA or inhibit DNA repair processes.

To elucidate the mechanisms by which the tumor microenvironment might contribute to genetic instability, we tested whether the adverse conditions found in solid tumors can alter the capacity of the cells to repair DNA damage. Using UV-induced damage as a probe, we examined the effects of hypoxia and low pH on the NER³pathway. We report here that cells exposed to hypoxia at pH 6.5 exhibit a diminished capacity to reactivate UV-damaged plasmids compared with control cells. Consistent with this reduction in repair, we find that cells incubated under such conditions are hypermutable to UV damage.

3340 is a mouse fibroblast cell line carrying in its genome 15 copies of the λsupFG1 shuttle vector DNA (28). The cells were maintained in DMEM supplemented with 10% fetal bovine serum (Life Technologies, Inc., Gaithersburg, MD) at 37°C in a humidified incubator with 5% CO₂.

Culture medium was acidified by supplementing the regular medium with 25 mm HEPES and 25 mm4-morpholinepropanesulfonic acid (Sigma, St. Louis, MO). The acidity of the medium was adjusted to a final pH of 6.5 with 1 n NaOH.

Hypoxic culture conditions were established as described previously(15), using a continuous flow of a mixture of 95%N₂ and 5% CO₂ gas certified to less than 10 ppm O₂ (Airgas Northeast, Cheshire, CT).

Plasmid pGL3-luciferase (Promega, Madison, WI), which encodes a luciferase gene driven by the SV40 promoter, was damaged in vitro by exposure to 5000 J/m² of UVC irradiation. A β-galactosidase-expressing plasmid, pSV-βGal(Promega), was used as an internal control to normalize for transfection efficiency.

3340 mouse fibroblasts were plated in triplicate in 60-mm dishes at a density of 3 × 10⁵ cells/dish. The next day, they were transfected with 2 μg of either intact or damaged pGL3-luciferase along with 0.5 μg of pSV-βGal using FuGENE 6 transfection reagent, as directed by the manufacturer (Boehringer Mannheim, Indianapolis, IN). Three h later, medium containing the transfection mixture was removed. Half of the dishes were replenished with fresh medium and incubated under standard conditions. The other half received low pH medium (pH 6.5) and were placed in a hypoxic incubator. Approximately 24 h after transfection, cells in each dish were lysed with 600 μl of Reporter Lysis Buffer (Promega). Transient expression of luciferase and βgalactosidase was determined by mixing 20 μl of cell extract with 100 μl of Luciferase Assay Reagent (Promega) or 300 μl of Galacto-Star Reaction Buffer (Tropix Inc., Bedford, MA) and reading the light emission on a luminometer.

Values of luciferase expression were normalized to theβ-galactosidase control and averaged over the triplicates. The repair efficiency (i.e., reactivation of the damaged plasmids by the host cells) for each condition was determined as the percentage of the luciferase activity expressed from the damaged plasmid relative to that from the undamaged plasmid. SD was calculated according to the Taylor series expansion formula.

3340 cells (8 × 10⁵ cells/100-mm dish) were exposed to UV light at a dose of 3 J/m² using a 254 nm germicidal lamp. Afterward,cells were either incubated under standard culture conditions (normoxia and pH 7.4) or exposed to hypoxia at pH 6.5 for 24 h. Both sets of cells were then maintained in standard culture conditions, and high molecular weight DNA was isolated from the cells 1 week later. Lambda shuttle vector rescue from the chromosomal DNA of the mouse 3340 cells and detection and characterization of mutations in the supFG1 reporter gene were carried out as described previously (29).

To test the hypothesis that the conditions of the tumor microenvironment may compromise cellular DNA repair processes, we examined the ability of cells to reactivate a UV-damaged plasmid when exposed to hypoxia and low-pH conditions. This type of assay, termed HCR, has been used in a variety of studies to assess repair activities in vivo (30, 31). The advantage of this experimental strategy is that the physiology of the cell itself is not perturbed by the DNA-damaging agent. DNA substrates containing preformed damage are introduced into cells, and the ability of the cells to repair the damage in the transfected DNA is probed under various conditions. In our experiments, plasmid pGL3-luciferase, which encodes a luciferase gene, was damaged by UV irradiation at a dose of 5000 J/m². This damaged plasmid DNA, or an equal amount of undamaged luciferase plasmid DNA, was transfected into mouse fibroblast cells (cell line 3340) that were subsequently placed under either normal culture conditions or hypoxic and acidic conditions to be tested. Repair and removal of the transcription-blocking UV lesions allow the subsequent expression of the luciferase reporter gene, which is assayed as a measure of repair. To control for transfection efficiency in each sample, an undamaged β-galactosidase-expressing plasmid, pSV-βGal, was also included in all of the transfections. Twenty-four h after transfection, cells were lysed, and luciferase andβ-galactosidase activities were measured.

Under standard conditions (normoxia and pH 7.4), luciferase activity expressed from the damaged plasmid was about 60% (68% in experiment 1 and 55% in experiment 2) of that expressed from the undamaged plasmid. This indicates that a substantial amount of the UV-generated damage present on the plasmid DNA was repaired by the host cells by the time of assay. However, when the cells were placed under hypoxic and low-pH conditions after transfection, the level of luciferase expression from the damaged plasmid relative to the undamaged control was only in the range of 30% (26% in experiment 1 and 35% in experiment 2),significantly reduced compared with the normal conditions (Fig. 1). We also performed the HCR assay on a human cell line, RCneo, and observed a similar decrease in the capacity of the cells to reactivate damaged plasmid under hypoxia and low-pH conditions (data not shown). These data suggest that cellular repair functions, at least with respect to the NER pathway, are less efficient under the suboptimal hypoxic and acidic culture conditions.

To determine whether the diminished repair capacity would confer hypermutability on the cells, we examined the effect of hypoxia and low pH on UV mutagenesis. Because the mouse fibroblast cell line 3340 used in the HCR assay carries a chromosomally integrated lambda shuttle vector containing the supFG1 mutation reporter gene, this cell line was used in the mutagenesis experiments. Cells were exposed to UV at a dose of 3 J/m². After irradiation, the cells were immediately placed under either standard culture conditions or hypoxia at pH 6.5 for 24 h. Cells were then returned to the standard conditions for 1 week of growth in culture, and mutations occurring in the supFG1 reporter gene were assayed by shuttle vector rescue from the cell DNA. The background mutation frequency in the unirradiated 3340 cells under the standard culture conditions was 11.8 × 10⁻⁵ (Fig. 2). Exposure to UV at 3 J/m² and growth under standard conditions produced mutations in the reporter gene at a frequency of 28.0 × 10⁻⁵, a value 16.2 × 10⁻⁵ above the level in unirradiated cells under the same standard conditions. However, when the cells were subjected to hypoxia and pH 6.5 immediately after UV treatment, the mutation frequency was found to be 53.2 × 10⁻⁵, a frequency 30.4 × 10⁻⁵ greater than that seen in cells treated with transient hypoxia and pH 6.5 alone(22.8 × 10⁻⁵). These results indicate that the extent of UV mutagenesis is greater in cells exposed to hypoxia and low pH in the immediate postirradiation period (when repair would be expected to occur) than in cells grown under standard conditions after irradiation.

The supFG1 mutations were examined by PCR amplification and DNA sequence analysis (Fig. 3 and Table 1). All were found to consist of point mutations; no large deletions or rearrangements were detected. The types of mutations from both sets of UV-irradiated cells (grown under either hypoxia and low pH or standard conditions) were similar, with mostly C:G to T:A transitions. Both spectra are characteristic of UV-induced mutagenesis in general and are similar to UV mutation patterns generated in the supFreporter gene in a number of other studies carried out under normoxic conditions (32). Hence, exposure of irradiated cells to hypoxia and low pH promotes hypermutability but does not cause a qualitative change in the types of induced mutations as compared with those generated under the standard conditions.

In this study, we have investigated whether the hypoxic and acidic conditions typically found in the tumor microenvironment can influence cellular DNA repair processes. We tested the ability of cells exposed to hypoxia and low pH to reactivate a UV-damaged luciferase reporter construct as a measure of repair. Cells incubated under such conditions were found to express lower levels of luciferase than did cells under normal conditions. We interpret this difference to reflect a decrease in the DNA repair capacity of the cells under the hypoxic and acidic culture conditions.

As a corollary to this observation, we examined the effect of hypoxia and low pH on mutagenesis. We found that cells exposed to hypoxia and low pH for 24 h in the immediate postirradiation period showed a 2-fold increase in mutation frequency compared with unirradiated cells under the same growth conditions. Because the hypoxia and low pH treatment elevated the mutation frequency even in the unirradiated cells [a result confirming our previous work (15)], the fold difference between the irradiated and unirradiated samples is about the same as that seen in cells maintained under standard conditions. However, the absolute increase in mutations due to UV is more substantial in the hypoxia/pH 6.5-treated samples. By subtracting the mutation frequencies seen in the unirradiated cells under the same growth conditions, it was found that the amount of mutagenesis attributable to UV in cells placed under hypoxia at pH 6.5 was 30.4 × 10⁻⁵, almost double the mutagenic effect of UV on cells maintained under standard conditions(16.2 × 10⁻⁵). This difference constitutes a state of hypermutability in cells that are in hypoxic and acidic conditions. Taken together, the above results provide a new mechanism by which the conditions of the tumor microenvironment may promote genetic instability: diminished DNA repair and hypermutability to DNA damage.

In our experiments, we used UV as a model mutagen to introduce damage both on a plasmid and on cellular genomic DNA. Although UV irradiation is not expected to be a physiological challenge for cells within a solid tumor, it is useful as a tool to probe repair and mutagenesis under selected conditions. UV-induced lesions, primarily thymine dimers and 6-4 photoproducts, are processed via the NER pathway, and thus our findings suggesting diminished DNA repair bear directly on that repair pathway. By extrapolation, however, it is likely that other repair pathways, such as base excision repair, double-strand break repair, and mismatch repair, may also be altered by the suboptimal cellular conditions imposed by hypoxia and low pH, although this remains to be determined.

In the NER pathway, there are at least 16 polypeptides involved in damage recognition, 3′ and 5′ dual incision, and various other aspects of the repair process (33). Our finding that cells exposed to hypoxia at pH 6.5 exhibit a decreased capacity to repair lesions normally subject to NER suggests that the hypoxic and acidic environment could impair some or all of the proteins involved in the cellular NER pathway.

It remains unclear exactly how the repair proteins might be affected by these conditions. One possibility is that the levels of certain proteins involved in the NER may be reduced. Whereas hypoxia can induce the expression of some oxygen-related proteins (20), it generally has a negative impact on protein metabolism. It has been found to both decrease protein synthesis and accelerate protein degradation (23, 24). However, immunoblot analyses did not reveal changes in the levels of selected NER proteins after 24 h of hypoxia at pH 6.5 (XPA and XPD; data not shown).

Another possibility is that the unfavorable environment may functionally inactivate or impair the activity of many proteins,including those critical for NER. It has been reported that the level of ATP drops rapidly after initiation of hypoxia (20) and,as a result, may reduce the activity of repair enzymes. An unphysiological pH is also likely to disturb proper protein conformation and folding and disrupt protein-protein interactions,further compromising the ability of cells to perform repair functions when challenged with DNA damage.

Our experiment was carried out transiently in culture with a pH of 6.5 and an oxygen tension of ≤10 ppm. It is important to note that such in vitro conditions do not exactly mimic the complex and dynamic microenvironment in a developing tumor. Cells growing in a solid tumor can be transiently or chronically hypoxic (34)and can be deprived of critical nutrients (35). Oxygen tension (15) as well as acidity (36) also varies spatially and temporally. In this regard, it is interesting to note that we saw only a small decrease in repair in cells treated with hypoxia alone (data not shown). However, when cells were exposed to a combination of hypoxia and low pH, larger differences were observed. This is in line with several studies reporting a synergistic effect of low oxygen and low pH on cellular energy metabolism and cell survival(37, 38). Hence, it is not unreasonable to hypothesize that in solid tumors in vivo, where various environmental factors may interact with one another, the influence of hypoxia and low pH on the NER pathway may be even more important than either factor alone.

Consistent with a diminished repair capacity, we found that exposure of cells to hypoxia at pH 6.5 caused an elevation in UV-induced mutagenesis. Sequencing analysis revealed that the majority of the mutations were consistent with the typical pattern of UV-induced mutagenesis in mammalian cells, with mostly C-to-T transitions. Hence,whereas we detected a quantitative increase in UV mutagenesis due to hypoxia and low pH, qualitatively, the types of induced mutations arising in hypoxic and acidic cells were found to be similar to those seen in cells under standard conditions.

Most UV-induced mutations are thought to arise from trans-lesion bypass synthesis across unrepaired damage(39). Hence, the increase in UV mutation frequency could theoretically arise as a result of either diminished repair or increased error-prone trans-lesion synthesis. We favor the former possibility, in which the observed hypermutability is due to a reduced capacity of cells to remove DNA damage under hypoxia and low pH, because that is consistent with our plasmid reactivation assay data. At this point, however, we cannot completely rule out some contribution of altered DNA polymerase activity, except to note that the pattern of mutations does not point to any unusual or novel bypass polymerase activity.

The work presented here supports the concept that the microenvironment within a solid tumor may be an important source of genetic instability(9, 15). Specifically, our results implicate diminished DNA repair as a possible mechanism underlying this instability. In addition, the concept that the conditions of the tumor microenvironment can inhibit DNA repair and consequently promote genetic instability provides a basis for understanding the observation that very hypoxic tumors follow a more aggressive clinical course (17, 40, 41, 42).

We thank K. Vasquez, L. Cabral, S. Peretz, X. S. Xu, M. Macris, J. Mendes, B. Casey, T. J. Li, R. Franklin, and S. J. Baserga for their help.