Enhancement of 5-Fluorouracil Sensitivity by an rTS Signaling Mimic in H630 Colon Cancer Cells

The rTSβ protein has been hypothesized to synthesize signaling molecules that can down-regulate thymidylate synthase. These molecules share biological and chemical properties with acyl-homoserine lactones (AHL), suggesting some AHLs might act as rTS signaling mimics and down-regulate thymidylate synthase. We have determined that the AHL, 3-oxododecanoyl homoserine lactone (3-oxo-C12-(l)-HSL) can down-regulate thymidylate synthase protein at 10 μmol/L and reduce H630 (human colorectal cancer) growth by 50% at 23 μmol/L (IC₅₀) in cell culture. At its IC₅₀ concentration, 3-oxo-C12-(l)-HSL reduces the apparent IC₅₀ of 5-fluorouracil (5-FU) from 1 μmol/L to 80 nmol/L (12-fold) in a colony formation assay. 3-Oxo-C12-(l)-HSL enhances the activity of 5-fluorodeoxyuridine, tomudex, and taxol but not the activity of 5-fluorouridine, methotrexate or Adriamycin. The unexpected interaction with taxol probably results from effects of the AHL on tubulin expression. Differences in taxol sensitivity, tubulin, and cellular morphology between H630 and the thymidylate synthase and rTSβ-overproducing, 5-FU-resistant H630-1 cell line as determined by colony formation assays, Western analysis of one-dimensional and two-dimensional gels, and photomicroscopy confirm that cytoskeletal changes are induced by the AHL or by rTS signaling. Isozyme differences in thymidylate synthase and rTSβ also exist in the two cell lines. Phosphorylation of rTSβ amino acid S121 is shown to occur and is decreased at least 10-fold in the drug-resistant cells. The data presented provide support for further investigations of rTS signaling mimics as enhancers to thymidylate synthase–directed chemotherapy, evidence that the phosphorylation state of rTSβ may be a marker for 5-FU resistance and a previously unrealized relationship between rTS signaling and the cytoskeleton.

The thymidylate synthase (EC 2.1.1.45) prodrug 5-fluorouracil (5-FU) has been in use as an anticancer agent for almost 50 years (1) and continues to be a mainstay of chemotherapeutic approaches for the treatment of solid tumors. Because the elevated expression of thymidylate synthase is a frequent factor in clinical resistance to 5-FU, a great deal of effort has gone in to studying the regulation of thymidylate synthase and attempting to identify predictors of response to thymidylate synthase inhibitors. In addition to the study of thymidylate synthase itself, there continues to be interest in the identification of other contributing factors that may affect tumor response to 5-FU or thymidylate synthase inhibitors (2, 3). Our laboratory has been investigating the function of the rTS (ENOSF1) gene and recently reported that the major protein product of this gene, rTSβ, is likely responsible for the production of signal molecules that can cause the down-regulation of thymidylate synthase under certain conditions in cell culture (4). We reported that these signal molecules were similar to acyl-homoserine lactones (AHL), molecules widely used by microorganisms for the regulation of gene expression (5). Using an extraction protocol similar to that used to purify AHLs from bacterial culture, extracts of spent culture medium prepared from the thymidylate synthase, and rTSβ-overproducing (and 5-FU resistant) H630-1 colon tumor cell line were shown to activate a bacterial AHL bioassay much more than extracts prepared from the H630-spent cell culture medium (4). It was also observed that these compounds and a product formed by rTSβ catalysis were metabolites of methionine and were formed directly from S-adenosylmethionine and another undefined substrate, indicating that the means to synthesize these metabolites is similar to that used by microorganisms to manufacture AHLs (4, 6). These similarities suggested to us that some AHLs might be able to cause the down-modulation of thymidylate synthase in human cells. Because AHLs have been shown to have activity as immunomodulators, vasorelaxants, or growth inhibitors against a variety of mammalian cell types (7), it is possible that thymidylate synthase is an unestablished site of action for some of these compounds. The identification of low molecular weight compounds that could cause the down-regulation of thymidylate synthase would have obvious potential value in improving the response to 5-FU or other thymidylate synthase inhibitors. In this report, we show the identification of an AHL that effectively modulates thymidylate synthase and enhances the effects of three thymidylate synthase prodrugs in a colony formation assay.

Alterations of gene expression that are not directly related to thymidylate synthase, occur in some 5-FU-resistant cell lines. These genes are linked to slowed rates of growth and changes in the size and shapes of the resistant cells (3, 8, 9). During the course of our investigations on AHLs, we unexpectedly observed that the effects of 3-oxo-C12-(l)-HSL included an effect on tubulin expression and the enhancement of the growth inhibitory properties of taxol. We also observed that the H630-1 and H630 cell lines differ significantly in their sensitivity to taxol. Because microtubules are the site of taxol inhibition (10), we looked for more evidence of a relationship between microtubule functions and the effects of both 3-oxo-C12-(l)-AHL and rTSβ overexpression. We report that expression of rTSβ is negatively correlated with the two major microtubule and cytoskeleton proteins (α-tubulin and β-actin) and that the relationship of rTS signaling to cytoskeletal proteins likely explains the morphologic differences between the two cell lines.

Two dimensional gel electrophoresis was used to determine whether two proteins with apparent antagonistic expression profiles (thymidylate synthase and rTSβ) can both be elevated in H630-1 cells, as a result of differential isozyme expression. We present evidence that the isozyme patterns for rTSβ, thymidylate synthase, and tubulin are distinct in the two cell lines. We show that one difference in rTSβ isozymes is due to S121 phosphorylation, with the extent of phosphorylation being much lower in H630-1 cells.

Reagents. AHLs were synthesized and purified as described previously (11), and all structures were confirmed by ¹H-NMR. The structures of the compounds used are presented schematically in Fig. 1. An (l) or a (d, l) designation refers to the chiral α-carbon in the homoserine lactone ring. Specific AHL stereoisomers were prepared from their respective homoserine lactone precursors. AHLs are (l) stereoisomers unless otherwise specified. The nomenclature for these compounds incorporates the term CX, where X is the number of carbons in the acyl chain. All reagents used for two-dimensional gel electrophoresis were obtained from Bio-Rad (Hercules, CA). Unless specified otherwise, all reagents used were of the highest quality available.

Cell culture and inhibitor studies. The conditions used to propagate and maintain the H630 and H630-1 cell lines have been described (4, 8). Cells were routinely screened and found negative for Mycoplasma. The thymidylate synthase modulation assay was done as described previously (12). H630-1 cells were grown in the absence of 5-FU for at least three passages before use. Growth inhibition and colony formation assays were done using six-well tissue culture dishes. For growth inhibition studies, wells were plated in triplicate with 10⁵ cells in 2 mL of medium and the compounds to be tested were added after 24 hours as 100- or 1,000-fold concentrated stock solutions. Compounds were dissolved in growth medium, or for the C12 compounds in DMSO (0.1% final concentration). For studies involving the C12 compounds, all wells were adjusted to contain 0.1% DMSO. For the investigation of effects on thymidylate synthase expression, cells were extracted at 18 to 20 hours after the addition of AHLs. For growth inhibition studies, the cells were counted 3 days after the addition of AHLs and the data for treated cells normalized as described below. The medium was aspirated and the cells were carefully rinsed twice with ice-cold PBS (137.9 mmol/L NaCl, 2.7 mmol/L KCl, 1.5 mmol/L KH₂PO₄, and 6.5 mmol/L Na₂HPO₄) and harvested by trypsinization. Cell numbers were determined using a hemacytometer. For the preparation of extracts for Western blotting, the harvested cells were combined with Laemmli sample buffer (13) supplemented to contain 4% SDS, 100 mmol/L freshly added DTT, and 1× protease inhibitor cocktail III (Calbiochem-Novabiochem, La Jolla, CA). The samples were boiled for 10 minutes. Protein amounts were determined using the bicinchoninic acid method and bovine serum albumin as a standard with a kit obtained from Pierce (Rockford, IL).

For colony formation assays, cells (200 and 500 cells per well for H630 and H630-1, respectively) were plated in triplicate. The compounds of interest were immediately added, then the cells were allowed to grow for ∼2 weeks (until colonies were visible by eye). The colonies were stained with crystal violet and counted. Resulting values for both the growth inhibition and colony formation assays are expressed as percent control cell growth or control colony number. For colony formation assays involving two compounds, the AHL was present at the concentration needed for 50% growth inhibition after 3 days (IC₅₀, 23 μmol/L for H630 and 8 μmol/L for H630-1; see Fig. 2B; data not shown). For graphic presentation, the values for cells treated with two compounds were normalized by multiplying these values by a number that gives 100% for cells treated with AHL alone. Data presented for cells treated with one compound in both types of growth assays are normalized to either untreated cells or cells to which solvent was added. All results are expressed ± SD. All experiments were repeated at least twice with similar results.

Photomicroscopy was done using an Olympus CK2 microscope, using a 20× objective and a 3.3× eyepiece in the laboratory of Dr. M. Ip. Photographs were taken with a Nikon Fx-35A camera of colonies stained with methylene blue.

Western blot analysis. Cells were extracted as described except for studies involving protein phosphorylation and those using two-dimensional PAGE. Western blotting was done by standard methods using blots blocked with 5% nonfat dried milk dissolved in 10 mmol/L Tris-HCl (pH 8.0), 150 mmol/L sodium chloride, and 0.05% Tween 20. For studies involving rTSβ phosphorylation, the extraction buffer and blocking buffers also contained 0.2 mmol/L sodium metavanadate and 50 mmol/L sodium fluoride. Proteins were separated by SDS-PAGE (10% polyacrylamide) and electroblotted to polyvinylidene difluoride membranes. The blots were probed using either monoclonal (4D5E11) or polyclonal antibodies to thymidylate synthase (gifts of Dr. Edward Chu, Yale University) and probed for α-tubulin (monoclonal antibody B-5-1-2, Sigma Biochemicals, St. Louis, MO) to confirm equivalent protein loading. The same blots were successively probed with anti-rTSβ or anti–thymidylate synthase and finally anti-tubulin. After probing for each protein, the blots were stripped by incubation in 8 mol/L guanidine HCl. Detection of rTS protein involved either the use of mouse monoclonal antibody D3 prepared against recombinant rTSβ protein or a rabbit polyclonal antibody prepared to a phosphorylated peptide. Preparation of the monoclonal D3 antibody will be described elsewhere. The anti-pS-peptide antibody was produced using the phosphorylated peptide DPRTLVpSCIDFR as a hapten, where S is amino acid 121 of rTSβ. Synthesis and purification of the phosphorylated peptide, coupling to a carrier and antibody production were conducted by Open Biosystems (Huntsville, AL) using their in-house protocols. To prove that the anti-pS peptide antibody only detects phosphorylated rTSβ, blots (either new or previously probed and stripped) were incubated in the presence of calf intestinal alkaline phosphatase (100 units/mL at 37°C, 24 hours) before probing or reprobing with the anti-pS peptide antibody and then checking for the loss of signal (data not shown). The blots were loaded with two alternate forms of recombinant rTSβ (not phosphorylated) containing either M118 or T118. These recombinant proteins were prepared from inclusion bodies as described previously (4). All Western blots were detected using the horseradish peroxidase–based West Pico Dura chemiluminescent substrate (Pierce) and horseradish peroxidase–coupled F(ab′)₂ fragments (Jackson ImmunoResearch Laboratories, West Grove, PA) as secondary antibodies.

RNA analysis. Thymidylate synthase expression was evaluated in 3-oxo-C12-HSL treated cells by reverse transcription-PCR (RT-PCR) as previously described (12). Microarray analysis of H630 RNA was done by the RPCI Gene Expression facility using the Affymetrix (Santa Clara, CA) HGU133A GeneChip. Microarray data was analyzed using Genetraffic software (Stratagene, La Jolla, CA) with the Robust Multichip Analysis method (14) for measuring probe intensity and scaling. Results for treated and untreated cells were analyzed for significant changes within each experiment (P < 0.05) as well as for changes in pooled data from three separate experiments. RNA for RT-PCR or microarray analysis was prepared using an RNAeasy kit from Qiagen (Valencia, CA). Each analysis made use of RNA from control cells and cells treated 20 hours with 23 μmol/L 3-oxo-C12-HSL prepared on three separate occasions from H630 cells.

Two-dimensional gel electrophoresis. For comparing H630 and H630-1 cells, the two cell lines were grown and processed simultaneously. Cells (50-70% confluent in 100-mm plates, ∼1.4 × 10⁷ cells per plate) were trypsinized, collected, and washed twice with an ice-cold isotonic solution [40 mmol/L Tris-HCL (pH 7.2), 250 mmol/L sucrose, and 5 mmol/L magnesium acetate]. The cells were lysed using 1 mL of extraction solution [5 mol/L urea, 2 mol/L thiourea, 2% CHAPS, 2% caprylyl sulfobetaine (SB3-10), 40 mmol/L Tris, 0.2% Bio-Lyte 3/10 ampholyte, 5 mmol/L magnesium acetate, 50 μg/mL DNase I, 10 μg/mL RNase A, 10 mmol/L NaF, 1 mmol/L sodium metavanadate, 1× protease inhibitor cocktail 3 (Calbiochem), and 2 mmol/L tributylphosphine] per plate of cells. After addition of the extraction solution, the samples were gently shaken at room temperature for 5 minutes. After centrifugation to remove insoluble materials (18, 000 × g for 5 minutes at room temperature), the samples were aliquoted and stored at −80°C or processed immediately for two-dimensional gel electrophoresis. Frozen samples were thawed on ice before electrophoresis. Samples were passively rehydrated at 20°C for 90 minutes (60 minutes after a mineral oil overlay) followed by active rehydration (50 V) for 16 hours. Approximately 100 μg of protein (based upon a value of 500 pg protein/cell) was used for isoelectric focusing in 7 cm ReadyStrips (pH 5-8) using a Bio-Rad Protean IEF cell. Isoelectric focusing in the first dimension was conducted for 10,000 V hour (20°C) using a rapid ramp protocol. After isoelectric focusing, the proteins were equilibrated in buffers as recommended by Bio-Rad and separated in the second dimension by 10% SDS-PAGE. The gels were electroblotted as described above. All denaturing gels included prestained molecular weight markers to allow evaluation of the apparent molecular weights of the identified proteins.

Two-dimensional microarray analyses. The two-dimensional evaluation of genes (i.e., mRNA) whose expression correlates with rTSβ gene expression was done using serial analysis of gene expression and the oligo microarray expression data for the NCI 60 cell line panel (i.e., NCI60 Novartis) by application of the two-dimensional display function of the Cancer Genome Anatomy Project (15, 16). Results for expression of the 10 genes whose expression best correlates positively and negatively with rTSβ, along with the correlation coefficients (R) and the level of significance (<P) for the NCI 60 panel of cancer cell lines can be accessed online by entering HSRTSBETA (Genbank accession no. X67098) for gene name at web site http://cgap.nci.nih.gov/SAGE/AnatomicViewer, and following the Gene Info link to the two-dimensional array displays link, and evoking the NCI60_Novartis (X67098) data.

Anti-phospho-rTSβ antibody. The design of the peptide for eliciting anti-phosphopeptide antibody to rTSβ was based upon several considerations. Analysis of the rTSβ primary amino acid sequence using NetPhos 2.0 (17) suggested a high probability of phosphorylation at seven serines (S121, S253, S307, S352, S378, S388, and S395; all with scores >0.94). We chose S121 as an interesting site for phosphorylation because of its proximity to amino acid 118, the site of the only known single nucleotide polymorphism (SNP) in the protein coding region of rTSβ (18). This SNP codes for either M118 or T118 depending upon whether the gene is polymorphic for T (M118) or C (T118). Considerations for the likelihood of peptide antigenicity and uniqueness were also taken into account in consultation with the Open Biosystems technical staff.

3-Oxo-C12-HSL can down-regulate thymidylate synthase. A variety of AHLs with acyl chains containing 4 to 12 carbons (C4-C12) and 3-hydroxy or 3-keto substituents were prepared and analyzed for their effect on thymidylate synthase protein expression (Fig. 1). Although several of the shorter chain AHLs (C4-C6) can cause loss of thymidylate synthase, the required concentrations are in the millimolar range. Note that the (l)-isomer of C6-HSL is active whereas the racemic mix is less active, suggesting that C6-(d)-HSL is inactive for down-regulation of thymidylate synthase. In contrast to the short chain AHLs, 3-oxo-C12-HSL can cause down-regulation of thymidylate synthase at concentrations as low as 10 μmol/L (Fig. 1B). In contrast, C12-HSL, which lacks the 3-oxo group, is inactive at concentrations as high as 100 μmol/L. The observed difference in activities when a 3-oxo group is present (3-oxo-C12-HSL versus C12-HSL) suggests that the down-regulation of thymidylate synthase by 3-oxo-C12-HSL is receptor mediated and that small changes in AHL structure can have pronounced effects on AHL effectiveness. Previously, it had been shown that the length of the acyl-side chain in AHLs increases stability in bacterial cell culture media (7), and this may be true in mammalian cell culture media as well. The results therefore are not necessarily an indication of the affinity of the AHLs for a human AHL receptor. Tubulin, a protein frequently employed as an internal standard for protein loading also seems to decrease with increasing concentrations of 3-oxo-C12-HSL but not with increasing concentrations of C12-HSL (Fig. 1B). This latter observation suggests 3-oxo-C12-HSL has effects on gene expression aside from the down-regulation of thymidylate synthase. The effect of 23 μmol/L 3-oxo-C12-HSL on thymidylate synthase mRNA in H630 cells (20 hours of exposure) was evaluated by RT-PCR. As shown in Fig. 1C, thymidylate synthase mRNA levels are unaffected by the compound. Microarray analysis also showed that thymidylate synthase mRNA levels were unaffected and also failed to identify a single gene that is significantly altered in response to treatment with the compound (data available upon request). These results indicate the effects of 3-oxo-C12-HSL on gene expression 20 hours after treatment of cells are due to post-translational events.

The thymidylate synthase modulation assay was developed to identify compounds that affected the translational autoregulation of thymidylate synthase mRNA and detect compounds that can cause the removal of thymidylate synthase from its mRNA binding elements (12). If thymidylate synthase is the major site of action of AHLs that can down-regulate thymidylate synthase, we would expect to see activity in the assay and a good correlation of thymidylate synthase down-regulation with a compound's growth inhibition. In contrast, inhibitors of cell growth such as actinomycin D, azacytidine, and thioguanine do not affect thymidylate synthase protein levels at concentrations that inhibit growth under similar conditions (12). We compared a long-chain AHL (3-oxo-C12-HSL) and a short-chain AHL (3-oxo-C6-HSL) for their ability to modulate thymidylate synthase in the thymidylate synthase modulation assay (12). Up-regulation of a luciferase reporter in this assay can result from any process that causes detachment of thymidylate synthase from its own mRNA. These factors could be thymidylate synthase down-regulation, thymidylate synthase inhibition, or increased concentrations of thymidylate synthase ligands that can cause dissociation of thymidylate synthase from its mRNA and allow the synthesis of the luciferase reporter (19). H630 cells were treated with IC₅₀ concentrations of each compound and results are shown in Fig. 2A. The shorter-chain compound 3-oxo-C6-HSL is a potent modulator of thymidylate synthase, much more so than the more classic thymidylate synthase inhibitors studied previously using this assay (12). In contrast 3-oxo-C12-HSL, although a much more effective down-modulator of thymidylate synthase (Fig. 1), has no detectable thymidylate synthase modulatory activity in this assay. Despite the lack of activity of 3-oxo-C12-HSL in the thymidylate synthase modulation assay, the concentration dependence of H630 growth inhibition by this compound closely mimics the down-regulation of thymidylate synthase (compare Fig. 1B with Fig. 2B). Unlike 3-oxo-C6-HSL, the growth inhibition curve for 3-oxo-C12-HSL seems biphasic, suggesting there may be a second site of action. The inability of 3-oxo-C12-HSL to elicit a strong response in the modulation assay indicates the compound likely interacts with other sites besides thymidylate synthase that affect expression of the reporter. Although we currently do not know what these sites are, we speculate that effects on mRNA translation or subcellular trafficking that prevent translation of mRNA from which thymidylate synthase has been detached could be involved. For comparison, Fig. 2B also shows growth inhibition data for the 3-oxo-(d, l)-C12-HSL and C12-HSL. It is evident that like C6-HSL (Fig. 1), the (l)-stereoisomer of 3-oxo-C12-HSL is more active than the racemic mixture. Figure 2B also shows that C12-HSL (Fig. 1) is a relatively poor growth inhibitor (IC₅₀ = 1 mM), consistent with its lack of effect on thymidylate synthase. Thus, it seems thymidylate synthase is an important but not the only target for the growth inhibitory effects of 3-oxo-C12-HSL.

Interactions of 3-oxo-C12-HSL with 5-fluorouracil and other compounds. To determine whether the down-regulation of thymidylate synthase could have therapeutic use, we investigated the ability of 3-oxo-C12-HSL to enhance the growth inhibitory effects of 5-FU in a colony formation assay. H630 and H630-1 cells were incubated with varying concentrations of 5-FU and the IC₅₀ amount of the AHL for each cell line. There is about a 12-fold enhancement (80 versus 1,000 nmol/L) in the apparent IC₅₀ of 5-FU in H630 cells (Fig. 3) in the presence of 23 μmol/L 3-oxo-C12-HSL. No enhancement is observed with H630-1 cells, suggesting either that the elevated expression of thymidylate synthase or other genetic or epigenetic changes in the 5-FU-resistant cells preclude an effect. For the H630 cells, there is a shift of the inhibition curve to the left in the presence of 3-oxo-C12-HSL.

Investigations of drug interactions of 3-oxo-C12-HSL were expanded to include other prodrugs that predominantly target thymidylate synthase [5-fluoro-2′-deoxyuridine (FdUdr), tomudex], a compound that is metabolized to both a thymidylate synthase inhibitor and to an RNA precursor (FUrd), a drug that targets dihydrofolate reductase (methotrexate) and agents generally considered unrelated to thymidylate synthase (taxol and Adriamycin). The results with these compounds for both the H630 and H630-1 cell lines (Fig. 4A-F) are consistent with thymidylate synthase being responsible for the effect of 3-oxo-C12-HSL. The curves for FdUdr and tomudex with H630 cells (Fig. 4A and B) both show enhancement but to a lesser extent than with 5-FU (Fig. 3). In contrast, no enhancement of FUrd occurs (Fig. 4C). The lack of effect with FUrd may be due to its primary site of action being incorporation into RNA in this system, although we have not pursued this hypothesis.

Enhancement of the effect of 5-FU, FdUdR, and tomudex but the absence of interaction of the AHL with either FUrd, methotrexate, or with Adriamycin clearly indicates that the effects of 3-oxo-C12-HSL upon thymidylate synthase are a major site of its growth inhibitory activity. It was therefore surprising to see some enhancement of taxol's effect (Fig. 4E). Unlike the interaction of 3-oxo-C12-HSL with the thymidylate synthase inhibitors, some enhancement of taxol's activity is seen with both cell lines. The data in Fig. 4E also show that the H630-1 cells are less sensitive to taxol (IC₅₀ = 20 nmol/L) than the H630 (IC₅₀ = 6 nmol/L) cell line. H630-1 cells were selected for 5-FU resistance and have been shown to overexpress thymidylate synthase as a result of gene amplification, as well as rTSβ and the c-yes oncogene which are so closely linked they are likely located within the same amplicon (3, 20, 21). Taxol is a microtubule inhibitor (10, 22). In an effort to find a reason for alterations in the response of H630-1 cells to a microtubule inhibitor, we looked for indicators of alterations of microtubule-related proteins and microtubule function in the H630-1 cell line. The Western blot data for AHL effects on H630 cells indicated that α-tubulin expression is affected by 3-oxo-C12-HSL (Fig. 1B). A two-dimensional microarray analysis of the NCI 60 cell line panel for genes whose expression correlates with rTSβ indicates that both tubulin (R = −0.57, P < 1.78 × 10⁻⁶) and β-actin (R = −0.61, P < 3.38 × 10⁻⁷) are among the 10 genes whose expression is most closely negatively correlated with rTSβ expression. Both tubulin and β-actin are associated with microtubule function and the cytoskeleton, which is responsible for maintenance of overall cell shape and which plays an important role in cell division and motility (23). Because H630-1 cells overexpress rTSβ 10- to 40-fold (Fig. 6B; ref. 9) and 3-oxo-C12-HSL but not C12-HSL seems to have an effect on tubulin expression (Fig. 1B), overproduction of rTS signaling molecules may have effects on microtubule function and the cytoskeleton. A previous report from our laboratory indicated that the two cell lines are morphologically distinct and that methionine restriction caused H630 cells to more closely resemble H630-1 cells (8). This difference is consistent with changes in the cytoskeletal makeup between the cells and is evident by examination of a photomicrograph comparing H630 and H630-1 cells (Fig. 5A and B). It is evident that H630-1 cells are more squamous, have pseudopodia, and some cells with multiple nucleoli, all evidence of an altered cytoskeletal makeup.

H630 and H630-1 cells contain different spectra of thymidylate synthase, rTSβ, and tubulin isozymes. Although rTSβ expression is associated with down-regulation of thymidylate synthase under certain conditions of cell growth, there are instances (e.g., in H630-1 cells) where both rTSβ and thymidylate synthase are expressed at high levels. One explanation for this could be the existence of isozymes for either protein that could affect the production of signaling molecules and/or sensitivity of thymidylate synthase. To address this issue, proteins simultaneously extracted from mid-log cells (50-70% confluent) of both cell lines were analyzed by Western analysis after two-dimensional gel electrophoresis. The results (Fig. 6A) show the presence of multiple thymidylate synthase and rTSβ species that differ in both isoelectric point (pI) and molecular weight between the two cell lines. There seem many more thymidylate synthase and rTS species present in H630-1 cells compared with H630 cells. For thymidylate synthase, the H630 cells show only one species of the predicted MW for thymidylate synthase (36 kDa) and one of the same pI at ∼75 kDa. This higher MW species may be an aggregate that has previously been found formed by thymidylate synthase in 2 to 5 mol/L urea (24). However, this thymidylate synthase species is not found in the H630-1 extracts, suggesting modification of the region responsible for its formation in the 5-FU-resistant cells. The rTS protein also displays a set of higher MW forms at ∼75 kDa in the H630-1 cell line. We have investigated the nature of this higher MW form of rTS protein and found that it is sumoylated.³

Because of the morphologic differences between H630 and H630-1 cells (Fig. 5), the two-dimensional blots were probed for the major cytoskeletal protein, α-tubulin. As evident in Fig. 6A, the pattern of tubulin isoforms is also altered between the two cell lines.

Examination of the amino acid sequence of rTSβ for potential post-translational modifications pointed to phosphorylation being one likely type of modification (Materials and Methods). A peptide-specific antibody was developed to a region of rTSβ predicted to contain a phosphoserine residue. Figure 6B shows the results of a Western blot with protein obtained from H630 and H630-1 cells, probed for S121 phosphorylated rTSβ (pS-rTSβ). Figure 6B shows that the pS-rTSβ antibody does not bind to the nonphosphorylated rTSβ isozymes (M118 and T118) but does detect protein in both cell lines. Despite having more rTSβ protein, the H630-1 rTSβ gives a weaker signal than H630 rTSβ when probed for pS-rTSβ. Based upon the amount of expression of rTSβ (D3 antibody), we estimate that <10% of rTSβ in the H630-1 cells is phosphorylated. Because there may be multiple phosphorylation sites on rTSβ, we are unable to relate S121 phosphorylation changes to the two-dimensional pattern for rTSβ shown in Fig. 6A.

Although rTSβ expression is elevated in H630-1 cells, in the dual drug resistant HCT-8/DF2 cell line (FdUdR and tomudex) and in the methotrexate-resistant K562 B1A cell line (9, 25, 26), elevated rTSβ expression seems not a widespread mechanism of 5-FU resistance compared with the expression of other genes (3). This is not surprising, considering the products of the rTSβ catalyzed reaction can cause down-regulation of thymidylate synthase and possibly inhibition of cell growth. Of the three cell lines that have been shown to overexpress rTSβ, only the K562 B1A cell line displays lower thymidylate synthase enzyme activity (9, 27). In thymidylate synthase gene-amplified cells, coamplification of the linked rTS gene is likely to result in overproduction of thymidylate synthase down-regulatory signal molecules. To allow for cell growth the cell must make some adjustment to allow for thymidylate synthase production. The existence of multiple forms of the rTSβ and thymidylate synthase proteins (Fig. 6A and B) in the H630-1 cells suggests that the functions of both proteins have been altered. These functions could include activity in situ, subcellular localization, and the production of and sensitivity to rTS signaling molecules. The reasons for the different response of H630 and H630-1 are currently under investigation. The presence of multiple ¹⁴C-[COOH]-methionine metabolites secreted by H630-1 cells suggests the existence of multiple signaling molecules (4) and it may well be that these could be attributed to the different isozymes of rTSβ. With the ability to detect and differentiate these isozymes, it will be interesting to determine if any correlations exist between the isoenzyme patterns and drug responsiveness in other cell lines and in clinical materials. The demonstration that 3-oxo-C12-HSL can enhance the effectiveness of 5-FU suggests the rTSβ signaling pathway may offer a new approach to increase the antitumor effects of this drug.

Grant support: Grants CA091114 (B.J. Dolnick and J.R. Sufrin) and EB002116 (B.J. Dolnick) and Roswell Park Cancer Institute core grant CA16054.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.