Acidic tumor microenvironment downregulates hMLH1 but does not diminish 5-fluorouracil chemosensitivity

Abstract

Human DNA mismatch repair (MMR) recognizes and binds 5-fluorouracil (5FU) incorporated into DNA and triggers a MMR-dependent cell death. Absence of MMR in a patient's colorectal tumor abrogates 5FU's beneficial effects on survival. Changes in the tumor microenvironment like low extracellular pH (pHe) may diminish DNA repair, increasing genomic instability. Here, we explored if low pHe modifies MMR recognition of 5FU, as 5FU can exist in ionized and non-ionized forms depending on pH. We demonstrate that MMR-proficient cells at low pHe show downregulation of hMLH1, whereas expression of TDG and MBD4, known DNA glycosylases for base excision repair (BER) that can remove 5FU from DNA, were unchanged. We show in vitro that 5FU within DNA pairs with adenine (A) at high and low pH (in absence of MMR and BER). Surprisingly, 5FdU:G was repaired to C:G in hMLH1-deficient cells cultured at both low and normal pHe, similar to MMR-proficient cells. Moreover, both hMSH6 and hMSH3, components of hMutSα and hMutSβ, respectively, bound 5FU within DNA (hMSH6>hMSH3) but is influenced by hMLH1. We conclude that an acidic tumor microenvironment triggers downregulation of hMLH1, potentially removing the execution component of MMR for 5FU cytotoxicity, whereas other mechanisms remain stable to implement overall 5FU sensitivity.

Fig. 1

Acidic tumor microenvironment downregulates hMLH1 alone but does not diminish 5FU chemosensitivity. (A) SW480 and HCT116 were cultured at both low extracellular pH (pH_e) (pH_e 6.2) and normal pH_e (pH_e 7.4) for 72 h. After extraction of protein, Western blotting was utilized to compare expression of hMSH2, hMSH3, hMSH6, hMLH1, TDG, MBD4 between both low and normal pH_e. Representative blots from three independent experiments are shown. (B) Clonogenic assay of SW480 and HCT116 cells in response to 5FU. Cells were plated in media at both at low pH_e (pH 6.5) and normal pH_e (pH 7.4) containing 0, 2.5, 5, 7.5 μM 5FU and allowed to form colonies for 14 days. The plates were then fixed with methanol and stained with 3% Giemsa, and viable colonies were counted.

Fig. 2

5FU incorporated into DNA is preferentially paired with adenine, followed by guanine. (A) Flowchart of procedure of in vitro analysis of binding partner of 5FU within DNA. (B-E) Frequencies of paired base with 5FU incorporated into DNA. One microgram of 5FdU:C containing dsDNA template PCR were utilized for both pH 8.2 or pH5.7 as a model of intracellular pH (pH_i) of tumor cells or normal cells in acidic tumor microenvironment with dNTP (B), only dCTP and dGTP (C), only dCTP, dGTP, and dATP (D), or only dCTP, dGTP, and dTTP (E). After TA cloning and total colony PCR, paired bases were analyzed by direct sequencing. The frequency of paired bases was calculated as the number of each base where 5FdU was inserted by total number of colonies (N ≥ 24) with an informative sequence at the 5FdU site. 5FU within DNA is predominantly paired with adenine when both dGTP and dATP are available (B, D), whereas guanine is preferentially paired with 5FU when dATP is absent (C, E).

Fig. 3

5FU incorporated into DNA is repaired in hMLH1 -downregulated cells in an acidic tumor microenvironment. (A) Flowchart of procedure of 5FU repair analysis in genomic DNA from colorectal cancer cells. SW480 and HCT116 were co-transfected with 5FdU:G-containing dsDNA, T:G mismatched dsDNA, or C:G matched dsDNA, and puromycin selection plasmid. Selection with 0.75 mg/mL puromycin-containing culture medium adjusted to pH_e 7.4 or pH_e 6.5 commenced 12haftertransfection. Ten days after transfection, genomic DNA was isolated. (B) Isolated genomic DNA was used for PCR, showed polyclonal PCR products in both MMR-proficient and hMLH1 -deficient cells cultured at both low and normal pH_e. (C) Concatamer formation after genomic DNA integration. After TA cloning of the purified PCR products followed by colony PCR, we confirmed by direct sequencing that the integrated dsDNA into genomic DNA formed concatamers (see Fig. S1). (D) Repair frequency of 5FU within DNA in MMR-proficient cells. The integrated C:G site of dsDNA was unchanged under both low and normal pH_e (left panel). Integrated T:G heteroduplex sites in MMR-proficient cells repaired to either T:A or C:G at both normal and low pH_e (middle panel). When cells were transfected with 5FdU:G containing dsDNA, 5FdU:G sites were mainly repaired to C:G in MMR-proficient cells at both normal and low pH_e (right panel). Repair frequency between normal and low pH_e did not show any significance by Chi-square test. (E) Repair frequency of 5FU within DNA in hMLH1-deficient cells. The integrated C:G site of dsDNA was unchanged under both low and normal pH_e (left panel). New equivalent T:G to C:G orT:G toT:G changes were seen in these hMLH1 -deficient cells (middle panel). Interestingly 5FdU:G was also predominantly repaired to C:G in hMLH1 -deficient cells as like MMR-proficient cells at both normal and low pH_e (right panel). Repair frequency between normal and low pH_e did not show any significance by Chi-square test.

Fig. 4

Affinity of hMLH1 for 5FU within DNA is lower than that of hMSH3 or hMSH6. (A, B, D, E) Immobilized biotin-labeled dsDNA probes with or without unlabeled DNA probe (competitor) were added in incubation buffer followed by mixture with 25 μg of nuclear lysate isolated from SW480 (MMR-proficient) (A), HCT116 + ch5 (hMLH1^−/− but hMSH3 restored) (B), HCT116 + ch3 (hMSH3^−/− but hMLH1 restored) (D) and HCT116 (hMLH1^−/− and hMSH3^−/−)(E), and incubated using a rotator for 1 hat room temperature. After washing, precipitated proteins were mixed with protein sample buffer and boiled for 3 min, and the supernatant was collected for Western blowing. (C, F) Unlabeled 5FdU:G containing dsDNA probes (5FdU:G competitor) were mixed with biotin-labeled 5FdU:G containing dsDNA probes and the competing off rate was compared which reflects the affinity level. In MMR-proficient cells, we found that the affinity level of 5GdU:G competitor for hMLH1 was markedly lower than that of hMSH3 and hMSH6 with hMSH6>hMSH3 ≫ ≫ hMLH1 (P < 0.01). Furthermore, the competing off rates for hMSH6 and hMSH3 were reduced when hMLH1 was absent. Each experiment was performed in triplicate.