dUTP Nucleotidohydrolase Isoform Expression in Normal and Neoplastic Tissues: Association with Survival and Response to 5-Fluorouracil in Colorectal Cancer

Thymidylate metabolism has long been an important target for widely used chemotherapeutic agents (e.g., the fluoropyrimidines) that provide benefit in the treatment of head and neck, breast, and gastrointestinal cancers (1, 2). The major mechanism of action of this class of antineoplastic drugs is the inhibition of enzymes that mediate critical steps in thymidylate metabolism. The de novo biosynthesis of TMP occurs by the reductive methylation of dUMP by the enzyme TS²to yield TMP, which is then converted to TTP for DNA replication (Fig. 1). The methyl donor in this reaction, MTHF, is oxidized to dihydrofolate so that the TS reaction constitutes a significant drain on cellular tetrahydrofolate pools. The levels of MTHF are maintained during TMP synthesis by the combined actions of DHFR and serine hydroxymethyltransferase (Fig. 1,A). Chemotherapeutic agents such as 5-FU and FUdR block TMP biosynthesis by inhibiting TS directly. Inhibitors of DHFR (e.g., methotrexate) indirectly block TMP production by limiting the availability of MTHF (Fig. 1 B).

Studies attempting to elucidate the molecular mechanisms of cell killing mediated by inhibitors of TS and DHFR suggest that cytotoxicity results from a process termed “thymineless death.” Historically,cell death initiated by thymineless conditions was presumed to be the result of DNA synthesis arrest and DNA degradation because of extreme TTP pool depletion (3). However, more recent investigations suggest that multiple factors contribute to the underlying mechanism of thymineless death, particularly imbalance of other deoxynucleotide triphosphate pools. It has now become largely accepted that elevated dUTP pools and misincorporation of uracil into DNA play a significant role in initiating DNA damage and cell death in response to inhibition of de novo thymidylate metabolism(Ref. 4 and references therein). Because of these findings, there is growing interest in the role of dUTP metabolism as a mediator of cytotoxicity and as a determinant of efficacy in the clinical use of antithymidylate chemotherapeutics.

In virtually all known organisms, uracil is not a native component of DNA. However, uracil can arise in DNA either by the spontaneous deamination of cytosine residues or through dUTP utilization by DNA polymerases during replication (5, 6). Because cytosine deamination can lead to G:C to A:T transition mutations, the cell has evolved highly efficient mechanisms to facilitate the exclusion of uracil from DNA (7). When uracil does occur in DNA, UDG initiates the base-excision repair pathway to remove and correct the misincorporated nucleotide. To prevent dUTP utilization during DNA replication, the enzyme dUTPase hydrolyzes dUTP to yield dUMP and PP_i. This reaction effectively eliminates dUTP from the DNA biosynthetic pathway and also provides substrate (dUMP)for the de novo synthesis of thymidylate. Therefore, under normal cellular conditions, the maintenance of uracil-free DNA is achieved through the combined actions of dUTPase and UDG.

Although dUTP is a normal intermediate in thymidylate biosynthesis, its extensive accumulation and misincorporation into DNA is lethal in both prokaryotic and eukaryotic organisms (8, 9). The exact biochemical basis for uracil-DNA-mediated cell death has not been definitively proven; however, there is substantial evidence suggesting that UDG-initiated repair is a central component of this process. For example, inactivation of dUTPase in Escherichia coli results in the dramatic accumulation of dUTP pools leading to extensive uracil misincorporation during replication. Under conditions of elevated dUTP pools, the cell engages in repeated cycles of uracil misincorporation and UDG-mediated repair. This iterative process results in increased recombination, DNA strand breaks, and ultimately cell death(8).

A similar phenomenon is thought to occur during inhibition of de novo thymidylate metabolism by anticancer agents(10, 11, 12, 13, 14, 15). Inhibition of the TS reaction leads to the accumulation of cellular dUMP pools and, as a result of mono- and diphosphate kinases, induces a dramatic increase in dUTP pools. Once levels of dUTP accumulate beyond a threshold level, overwhelming cellular dUTPase activity, the dUTP:TTP ratio increases. Under these conditions, dUTP is misincorporated into replicating DNA, resulting in uracil-DNA-mediated cytotoxicity (Fig. 1 B). Implicit to this model of cell killing is the central role of dUTPase. As the main regulator of dUTP pools, the expression of dUTPase could have profound effects on the utility of chemotherapeutics that inhibit thymidylate biosynthesis. Normally, dUTPase mediates a protective role by limiting the expansion of dUTP pools and countering the cytotoxic effect of uracil misincorporation. According to this model, elevated levels of dUTPase would prevent the accumulation of dUTP required for cell killing. To test this hypothesis, Canman et al.(16) ectopically overexpressed the E. colidUTPase in a FUdR-sensitive human colorectal tumor cell line (HT29) and measured the response to the TS inhibitor FUdR. The manipulated cell lines (containing dUTPase activity 4–5-fold higher than controls) were protected from FUdR-induced DNA strand breaks and showed an increased viability over control cells (16). This study provided the first evidence in human cells implicating dUTPase enzyme levels as an important factor in determining the efficacy of TS inhibition. Although the contribution of uracil-DNA-mediated cytotoxicity toward overall cell death is likely to vary between different cell lines(17), these data support the role of dUTP pool imbalance and uracil misincorporation as a contributing mechanism of TS inhibitor-based cytotoxicity.

Cancer of the gastrointestinal system is one of the leading causes of cancer death, and the fluoropyrimidines remain the chemotherapeutic agent of choice used to combat these diseases (18). The difficulty in successfully treating patients with this class of drugs is the high incidence of resistance intrinsic to these cancers. Because resistance is a common phenomenon, understanding and predicting response and overall patient outcome is of great importance for the clinical evaluation and management of the cancer patient. Considerable effort has been directed toward identifying useful molecular markers that predict for patient response to chemotherapy and overall survival in gastrointestinal cancers. For example, recent advances have focused on measuring the intratumoral expression of molecular determinants of drug action (i.e., TS and TP) and mechanisms of response to DNA damage (e.g., p53; Refs. 19, 20, 21, 22, 23). However,there has been no clinical evaluation of potential predictive markers involved in dUTP metabolism. Considering the central role of dUTPase in uracil-DNA-mediated cytotoxicity, we conducted a retrospective study to evaluate the ability of dUTPase isoform expression to predict for response to 5-FU-based chemotherapy and overall patient survival in colorectal cancers.

dUTPase-specific MAbs (MAb DUT415) were generated and prepared as described previously (24). The DUT415 MAb is useful for immunoprecipitation experiments and immunohistochemistry but is not effective for immunoblot analysis. For immunoblot analysis,dUTPase-specific polyclonal antibodies were generated against recombinant DUT-N protein (expressed in a baculovirus system) as described previously (25). The dUTPase-specific polyclonal antibodies were purified by immunoaffinity chromatography and used for immunoblot analysis at a dilution of 1:1000. Immunoprecipitation of dUTPase was performed using DUT415. Total HeLa S3 cell extracts (1 mg)were incubated for 2 h with 5 μg of DUT415. Protein A-Sepharose(Sigma) was added (150 μl of a 10% solution) to complex and pellet the antigen-antibody complexes. The immunoprecipitations were washed five times in RIPA buffer (150 mm NaCl, 1.0% NP40,0.5% deoxycholic acid, 0.1% SDS, and 50 mm Tris, pH 8.0)and subsequently prepared for SDS-PAGE analysis as described previously(25). For Western/immunoblot analysis, protein was fractionated by 15% SDS-PAGE and transferred to nitrocellulose. The membrane was blocked in TBST [10 mm Tris (pH 8.0), 150 mm NaCl, and 0.05% Tween 20] containing 5% powdered skim milk. Antibody incubations were performed as described previously(25). Visualization of the protein bands was performed using the ECL chemiluminescent Western blotting detection system(Amersham Corp).

All cell lines were maintained in a humidified atmosphere of 5.5%CO₂ at 37°C. HeLa S3 cells (CCL 2.2) were purchased from American Type Culture Collection and maintained in DMEM supplemented with 5% FCS purchased from Life Technologies, Inc. Human PBLs were prepared from venous whole blood with LeucoPREP cell separation tubes (Becton Dickinson) using the manufacturer’s recommendations. PBLs were cultured in RPMI 1640 supplemented with 10%fetal bovine serum. PBLs were stimulated by the addition of PHA to a final concentration of 15 μg/ml (Sigma). The onset of DNA replication was monitored in stimulated PBLs by[³H]thymidine incorporation. At 12-h intervals after PHA stimulation, cells were labeled for 30 min with 10 μl of[³H]thymidine (1 mCi/ml). The remainder of the procedures was performed as described previously (26).

Cultured cells were harvested at the appropriate times, washed in PBS,counted, and resuspended in PBS at 3 × 10⁵ cells/ml. Positively charged slides were placed in cytocarriers (adapter 1024) for the IEC-Centra-8R centrifuge(International Equipment Co.). One-ml aliquots of the cell suspension were centrifuged at 400 × g for 5 min. Cells were fixed in 4% paraformaldehyde in PBS (pH 7.2) for 30 min at room temperature while the slides were still attached to the slide carrier. Slides were removed from the carrier and washed five times for 2 min in PBS (pH 7.0). Cells were then lightly digested with Proteinase K Dako cat# 53020 diluted 1:500 with PBS for 15 min at room temperature. Cells were washed again five times in PBS, and immunocytochemistry was performed using the ABC-alkaline phosphatase protocols. dUTPase-specific MAb (DUT415) was used at a concentration of 2 μg/ml and incubated for 30 min at room temperature.

Formalin-fixed, paraffin-embedded human tissue specimens were sectioned at 4–5 μm and placed on Capillary Gap charged and precleaned glass microscope slides (Ventana Medical Systems, Inc.). The tissue sections were deparaffinized in xylene and rehydrated through graded alcohol solutions. Prior to immunohistochemical labeling, epitope recovery was performed by steam pretreatment (Ref. 27 and references therein). Capillary Gap slide pairs were placed in a slide holder in 10 mm sodium citrate (pH 6.0). Water was brought to a boil in a steamer (Handy Steamer Plus; Black and Decker), and the entire slide holder was placed in the upper chamber of the steamer with the lid on and steamed for 20 min. After steam pretreatment, the slides were cooled for 10 min, rinsed in PBS for 5 min, and incubated in a 3%hydrogen peroxide solution for 5 min to quench any endogenous peroxidase activity. The tissues were rinsed again in PBS and placed in normal goat serum (5% in PBS) for 5 min to reduce nonspecific staining. The tissues were subsequently incubated with primary antibody(e.g., DUT415) for 30 min at room temperature. After incubation, the tissues were rinsed in PBS and incubated in biotinylated goat-anti-mouse secondary antibody for 30 min (Vector Laboratories). The slides were washed again in PBS to reduce nonspecific staining of secondary antibody and incubated with ABC-peroxidase for 30 min (Vector Elite). After a final rinse in PBS,the tissues were incubated with DAB substrate (0.02% DAB, 0.005%hydrogen peroxidase) for 15 min to develop the colorimetric reaction,then counterstained in hematoxylin 100% (GILL 11 formula), and mounted with glass coverslips using Permount.

All antibodies were individually optimized for working dilutions. All tissue samples were pretreated by steaming as described above for antigen recovery. The DUT415 MAb was generated in our laboratory and used at 2 μg/ml. Ki67 (MIB-1 clone) was purchased from Immunotech and used at a dilution of 1:60. The mitochondrial marker 113-1 (Mito-M) was purchased from BioGenex and used at a concentration of 1:100.

Specimens from 20 patients with metastatic colon cancer were available for analysis of dUTPase expression. These are a subset of 36 patients who had received protracted infusion of 5-FU with leucovorin for treatment of the metastatic disease and who had consented to have biopsies for analysis of molecular markers. Tumor specimens from these patients had been analyzed previously for TS gene expression by reverse transcription-PCR (28).

Patients who experienced a complete response or partial response(>50% reduction of measurable tumor) were classified as responders. Survival was calculated as the time from start of treatment until death of any cause or until last date known to be alive. Time to tumor progression (>25% increase in tumor mass) was determined from patients that had stable disease or responded to chemotherapy. Response and survival information was available for all 20 patients included in this study. See Table 1 for a summary of the demographic characteristics of these patients.

Contingency tables and Fisher’s exact test were used to summarize the association between dUTPase expression in the nucleus and in the cytoplasm with other molecular variables and with tumor response to chemotherapy. The t test after logarithm transformation was used to compare TS levels according to dUTPase status. Kaplan-Meier plots (29) and the log-rank test (30) were used to compare survival of patients according to dUTPase expression. Median survival was calculated based on the Kaplan-Meier estimator. All Ps are two-sided.

The overall goal of this study is to assess the utility of measuring intratumoral dUTPase isoform expression as predictive markers of tumor response to 5-FU-based chemotherapy and overall patient survival in colon cancer. To begin this analysis, it is important to understand the molecular and biochemical details of the dUTPase isoforms, the specificity of the monoclonal antibody used to detect these proteins in patient tumor specimens, and how these proteins are expressed in normal tissues.

In previous studies, we identified and characterized distinct mitochondrial and nuclear isoforms of dUTPase in humans (DUT-M and DUT-N, respectively) that have identical kinetic affinities for dUTP(25). Detailed mass spectrometry-based analysis of native dUTPase proteins revealed that the isoforms are largely identical,differing only in a short region of their NH₂termini (25, 31). Fig. 2 A summarizes these data in schematic format. The protein sequences shown indicate native NH₂-terminal sequence and illustrate where the isoforms overlap in their primary sequences. The identity of the amino acid sequence at the protein level reflects the fact that the same nuclear gene encodes both isoforms,with each isoform arising through the use of alternative 5′ exons(32). Additionally, the DUT-N isoform is phosphorylated on Ser-11 in vivo (31).

To characterize dUTPase expression in human tissues, we developed an immunohistochemical assay using a dUTPase-specific MAb, DUT415. To account for the expression of both isoforms during immunohistochemistry analysis, it is essential to demonstrate that the antibody used in these experiments is able to recognize both protein species. To demonstrate cross-reactivity, dUTPase protein was first immunoprecipitated from total HeLa cell extract with DUT415 MAb and fractionated by 15% SDS-PAGE. The protein isoforms were then visualized by immunoblot analysis using dUTPase-specific polyclonal antisera. We used the polyclonal antibody in this experiment because DUT415 is not useful for immunoblot analysis. The Western blot shown in Fig. 2 B demonstrates that DUT415 is able to immunoprecipitate both nuclear and mitochondrial isoforms of dUTPase. The processed DUT-M protein (M_r23,000) is readily resolved from the DUT-N protein, which has an apparent molecular weight of M_r22,000. These data demonstrate the cross-reactivity of DUT415, which is likely attributable to the large region of homology common to these isoproteins.

Prior to this study, our knowledge of the expression of dUTPase isoforms in human cells relied primarily on cell culture models. Using normal human fibroblasts (34Lu cells), our laboratory has investigated dUTPase isoform expression as a function of cellular proliferation status. Data from these experiments demonstrate that DUT-N is regulated in a growth-dependent manner, with expression correlating with the onset of DNA replication at both the protein and mRNA levels. In contrast, DUT-M is expressed in a constitutive fashion, mimicking the pattern of mtDNA replication that occurs independently of nuclear DNA replication (32, 33). To establish the distribution of dUTPase protein expression in human tissues, we have developed methodologies to detect dUTPase by immunocyto- and immunohistochemistry using DUT415. Initially, PBLs were examined as a primary cell culture model. To provide a basis for the interpretation of isoform expression in human tissues, immunoblot analysis and immunohistochemistry were used in tandem to investigate the differential expression of dUTPase isoforms after mitogenic stimulation of quiescent PBLs.

PBLs were isolated from venous whole blood and either maintained in culture in a resting state or were stimulated by the addition of PHA. Fig. 3 A illustrates a time-course experiment where protein extracts from equivalent numbers of PBLs were fractionated by SDS-PAGE and dUTPase was detected by Western blot analysis. Time points were taken at 24 and 48 h after PHA stimulation, including negative controls. Untreated PBLs express the DUT-M isoform at both 24- and 48-h time points, whereas the DUT-N isoform is undetectable in these quiescent cells. However, after PHA addition, DUT-N undergoes a dramatic induction as PBLs are stimulated to proliferate. Analysis of DNA replication by [³H] thymidine incorporation indicates that PBLs begin DNA synthesis ∼24 h after stimulation,suggesting that DUT-N is up-regulated just prior to the onset of DNA replication in these cells (data not shown). These data are in agreement with the 34Lu fibroblast model, where DUT-M is constitutively expressed and DUT-N expression is growth dependent (32).

To determine whether we could detect dUTPase directly within these cells, the DUT415 antibody was used in immunocytochemistry experiments on either quiescent or stimulated PBLs. Fig. 3 B illustrates dUTPase expression in PBLs that were either left untreated or mitogenically stimulated with PHA for 48 h. The quiescent PBLs exhibit no detectable staining using this methodology, suggesting that the level of DUT-M protein in these cells is below the sensitivity of this assay. In contrast, the PHA-stimulated cells demonstrate strong nuclear expression of dUTPase, as indicated by the presence of red chromogen. Preincubating the DUT415 antibody with a 5-fold excess of recombinant antigen abolishes staining in the PHA treated cells (data not shown). These observations are in agreement with the immunoblot analysis, where DUT-N is dramatically up-regulated in stimulated cells. These data also demonstrate the ability of the DUT415 antibody to detect dUTPase protein in normal human cells by immunocytochemical techniques.

To determine the expression of dUTPase in normal tissues, we developed methodology to detect dUTPase in formalin-fixed and paraffin-embedded human tissues by immunohistochemistry. In addition to dUTPase, the proliferation marker Ki67 (MIB-1 clone) and the mitochondrial marker[113-1 clone (Mito-M)] were used to stain serial sections of normal tissue. These markers were used as controls in these experiments to identify proliferating cells (Ki67) and cells that possess a rich population of mitochondria (Mito-M; Refs. 34 and35). To establish the pattern of dUTPase expression throughout the human body, we surveyed a panel of different normal human tissues with the DUT415 antibody. Results from these experiments reveal both nuclear and cytoplasmic staining by DUT415 (Fig. 4). An ABC-peroxidase technique was used in these experiments, and protein expression is indicated by the presence of brown chromogen. Nuclear dUTPase staining parallels Ki67 in highly proliferative cell types in all cases throughout the human body. Examples include proliferative cells of the immune system and replicating cells of epithelia. Fig. 4,A illustrates nuclear staining by DUT415 in palantine tonsil. The proliferating B cells of the germinal center and stratum basale layer of the associated stratified squamous epithelium stain in the nucleus with the DUT415 antibody. The proliferation marker Ki67 demonstrates a similar pattern of expression,as shown in Fig. 4 B. The nuclear staining of dUTPase in proliferating cells of normal human tissues is consistent with the studies of DUT-N expression in both 34Lu cells and PBLs, where DUT-N protein is expressed in response to proliferation status. In addition,DUT415 staining was consistent within and among individual tissue samples. For example, reproducible staining patterns were observed in 10 different tonsil specimens, and repeated staining of the same tissues yielded similar results.

Cytoplasmic expression of dUTPase was observed throughout the human body in metabolically active tissues such as the kidney, heart,thyroid, and adrenal glands. An example of cytoplasmic dUTPase expression is shown in Fig. 4,C. In this image, the DUT415 MAb was used to stain the nonproliferative cells of the myocardium. Although the nuclei are readily visible and lack detectable nuclear dUTPase staining, these cells stain strongly in the cytoplasm with DUT415. Staining of a serial section of myocardium with the Mito-M polyclonal antibody reveals a similar staining pattern as DUT415 (Fig. 4 D). In all cases tested, cytoplasmic expression of dUTPase is correlated with a high mitochondrial content, as determined by Mito-M immunoreactivity. Specific examples include the proximal convoluted tubules of the kidney and thyroid follicular cells. The cytoplasmic expression of dUTPase does not correlate with proliferation status and is found in both replicating and nonreplicating cell types. This observation is in agreement with dUTPase expression observed in 34Lu cells and PBLs in which expression of the DUT-M isoform occurs independent of proliferation status. Together these data suggest that the observed cytoplasmic staining in normal tissues represents the mitochondrial isoform of dUTPase.

Fig. 4 E illustrates dUTPase expression in normal colonic mucosa. Immunohistochemical staining of colon tumors will be described in the next section; therefore, representative staining of DUT415 in normal colon is included here as a reference. Nuclear staining is observed in the replicating cells of the bottom half of the crypts of Lieberkühn. There is also significant dUTPase staining in the cytoplasm of these cells. Staining of colonic mucosa with Mito-M antiserum demonstrates that these cells are rich in mitochondria,suggesting that the cytoplasmic staining of dUTPase in these cells represents DUT-M (data not shown).

To demonstrate the specificity of the DUT415 antibody in the immunohistochemistry assay, we performed antigen competition analysis on a serial section of normal colon tissue. DUT415 was preincubated with 5-fold molar excess of purified recombinant dUTPase protein, and immunohistochemistry was performed as described. Preincubation with antigen completely abrogates staining with DUT415 in both the nucleus and cytoplasm as illustrated in Fig. 4 F. Competition experiments were also performed with several other tissues including,tonsil, kidney, breast cancer, colon cancer, and melanoma. Identical to normal colon tissue, staining with DUT415 was effectively competed with recombinant antigen (data not shown).

The overall analysis of dUTPase expression in both cell culture and normal human tissues suggests that the pattern of intracellular distribution of dUTPase protein reflects the physiological status of the cell. Nuclear staining (DUT-N) is observed in proliferating cells,corresponding to an up-regulation of S phase-dependent gene products. In contrast, cytoplasmic staining (DUT-M) appears to reflect rich mitochondrial content within the cell that is indicative of a high metabolic status.

In an effort to establish the distribution and intracellular localization of dUTPase expression in human cancers, we performed immunohistochemistry on colon and gastric tumor samples. This analysis was performed on formalin-fixed and paraffin-embedded tumor tissue using the DUT415 antibody. Results from these experiments reveal that dUTPase expression varies dramatically in different tumor specimens,not only in the quantity of expression but also in the intracellular localization. Our data show varied phenotypes of expression ranging from exclusive cytoplasmic expression to exclusive nuclear expression. In addition, there is evidence suggesting that dUTPase is aberrantly expressed in certain tumors. Photomicrographs of dUTPase and Ki67 staining in two colon tumors are presented to illustrate examples of the observed expression patterns.

In Fig. 5,A, dUTPase expression is highly expressed in both the cytoplasm and nucleus within the tumor cells, while the surrounding stromal cells exhibit very little dUTPase expression. A serial section of the same tumor, stained with the proliferation marker Ki67,demonstrates a characteristic nuclear staining pattern in the tumor cells indicating a high proliferative status within this cancer (Fig. 5 B).

In contrast, a different pattern of dUTPase expression is exhibited in Fig. 5,C. This photomicrograph depicts the margin of the tumor where both neoplastic tissue (left) and normal colonic mucosa (right) can be observed. In this case, dUTPase expression within the tumor cells is predominantly cytoplasmic (Fig. 5,C). The colonic mucosa on the right side of this image demonstrates characteristic nuclear dUTPase staining in the replicating cells in the bottom half of the crypts of Lieberkühn,providing an internal positive control for nuclear dUTPase staining (see Fig. 4,E for comparison). Interestingly, a serial section of the same tissue stained for Ki67 demonstrates strong expression in the nucleus of both the tumor cells and the replicating cells of the crypts (Fig. 5 D). These data demonstrate that nuclear dUTPase staining and Ki67 do not always correlate within tumor tissues, suggesting that dUTPase isoforms may be aberrantly expressed or incorrectly localized in certain tumors.

To further illustrate the variable intracellular distribution of dUTPase in tumors, photomicrographs are presented in Fig. 6 of colon cancers stained with the DUT415 MAb. This panel of images illustrates exclusive nuclear expression (Fig. 6, A and B), exclusive cytoplasmic expression (Fig. 6, C and D), and a combination of nuclear and cytoplasmic expression (Fig. 6, E and F). It is significant to note that the different patterns of dUTPase expression illustrated are commonly observed in colon and gastric cancers. For example, a study of dUTPase expression in 41 gastric tumors revealed that 27% had nuclear and cytoplasmic staining, 12% had exclusive nuclear expression, 39% had exclusive cytoplasmic expression, and 22% were negative for dUTPase staining (data not shown). Additionally, the phenotype of expression appears to be stable within a given tumor specimen. A tumor sample exhibiting exclusive cytoplasmic expression within one field of view retains this phenotype throughout the specimen and tends not to display heterogeneous phenotypes. It is currently unknown whether metastatic lesions exhibit different patterns of expression relative to the primary tumor. The expression patterns of dUTPase within colon and gastric cancers provide distinct and recognizable phenotypes that together or independently may be of significant prognostic value.

To explore the possible role of dUTPase expression as a prognostic marker, we conducted a retrospective study of 20 patients with metastatic colorectal cancer who were treated with protracted infusion of 5-FU/LV. Patient samples were scored according to the percent cells staining and the pattern of dUTPase staining within the tumor cells(i.e., nuclear or cytoplasmic). Tumor samples were scored as positive for nuclear dUTPase expression if >10% of cells exhibited nuclear staining. The cutoff level of 10% was chosen arbitrarily based on cutoff levels of other markers such as p53 and p21. This threshold was chosen prior to the statistical evaluation of the data and appeared to be high enough to rule out any background staining. Independent,larger patient studies will be required to identify the clinical significance of different levels of positive staining, as well as to determine the optimal cut point for predicting response. Cytoplasmic staining was scored on a percentage basis; however, these results were not included because there was no apparent association between cytoplasmic staining and response to chemotherapy, time to progression,or survival in this limited study. TS gene expression was also determined by quantitative reverse transcription-PCR and used as a comparative marker in these samples according to the methods detailed by Lenz et al. (Ref. 22and references therein).

Of the 20 samples tested, we found that 8 patients’ tumors contained positive nuclear dUTPase staining (>10% positive nuclear expression). There was a variable amount of cytoplasmic staining (ranging from>75% to <5% tumor cells demonstrating cytoplasmic staining). None of these patients responded (0%) to 5-FU-based chemotherapy. Only 1 patient had stable disease, and 7 patients demonstrated progressive disease. From these 8 nuclear-positive patients, the overall median TS gene expression was 5.35 × 10⁻³ (expressed as a ratio of TS:β-actin). The median survival in this group was 6.9 months, and time to tumor progression was 2.7 months (Table 2).

We found that 12 patients’ tumors were negative for nuclear dUTPase staining. Again, there was a variable amount of cytoplasmic dUTPase expression associated with these tumors. Of these 12 nuclear-negative tumors, 6 responded to chemotherapy (at least 50%reduction of tumor size), 4 had stable disease, and 2 showed tumor progression. From these 12 tumors, the median TS expression was 2.20 × 10⁻³. The median survival in this group was 8.5 months, and time to tumor progression was 6.3 months (Table 2).

Comparative analysis of these data indicates that nuclear dUTPase expression may be a statistically significant predictor of response to 5-FU-based chemotherapy (P = 0.005), time to progression (P = 0.017), and overall survival(P = 0.09) in metastatic colon cancer. High nuclear dUTPase expression is associated with poor response, shorter time to progression, and poorer survival. Conversely, low nuclear dUTPase staining is associated with response to chemotherapy, longer time to progression, and greater overall survival.

Inhibitors of thymidylate metabolism (i.e., the fluoropyrimidines and anitfolates) represent an important class of antineoplastic agents used for the treatment of head and neck, breast,and gastrointestinal cancers (1, 2). Clinical response to 5-FU-based regimens is typically between 20 and 30%, and drug resistance, either intrinsic or acquired, is a common phenomenon. Because response is difficult to predict using anatomical staging alone, considerable effort has been directed toward understanding mechanisms of drug action and identifying markers that are better able to predict for response to therapy. Recent advances in this field have focused on determining intracellular levels of relevant enzymes,thereby generating a predictive molecular profile for each tumor specimen. Important markers of survival and response to 5-FU-based chemotherapy include not only the target enzyme TS but also enzymes involved in drug bioactivation and deactivation (i.e., TP and dihydropyrimidine dehydrogenase) and critical protein determinants of DNA damage response (p53 and p21; Refs. 22 and36, 37, 38, 39). Despite these advances, there has been no clinical evaluation of enzymes involved in uracil-DNA-mediated cytotoxicity as prognostic markers of fluoropyrimidine-based therapy.

Investigation of the underlying molecular mechanisms of cell killing induced by inhibitors of thymidylate metabolism suggest that aberrant uracil-DNA metabolism plays a significant role in initiating DNA damage and cell death (as reviewed by Aherne and Brown 4). A growing body of evidence suggests that variation in the expression of dUTPase, the chief regulator of dUTP pools, may be a critical factor in determining the efficacy of a broad range of chemotherapeutic agents that target de novo thymidylate biosynthesis. The model of uracil-DNA-mediated cytotoxicity predicts that overexpression of dUTPase induces resistance to TS-directed chemotherapeutics by limiting drug-induced dUTP pool accumulation, thereby preventing uracil misincorporation into DNA. In contrast, low dUTPase expression would promote dUTP accumulation, thereby inducing greater sensitivity to these agents. To test this hypothesis and establish the value of dUTPase expression as a prognostic marker, we have developed methodology to measure dUTPase expression by immunohistochemistry using the MAb DUT415. This investigation is the first to characterize dUTPase isoform expression in normal and neoplastic tissues and to correlate nuclear dUTPase expression with response and survival in colorectal cancer.

The data presented in this study establish the expression patterns of dUTPase isoforms in normal tissues. Immunohistochemical analysis of dUTPase throughout the human body demonstrates that nuclear staining occurs in replicating cell types. Examples include the stratum basale of epithelial tissue, cells at the base of the crypts of Lieberkühn in the mucosa of the gastrointestinal tract, and proliferating lymphocytes. This observation is confirmed by identical staining with the proliferation marker Ki67. These data are consistent with the expression of DUT-N in cell culture models (e.g.,PBLs and 34Lu human lung fibroblasts) where DUT-N protein and message are regulated in a growth-dependent manner (32). Taken together, these data strongly suggest that nuclear staining by DUT415 in normal human tissues is indicative of DUT-N expression.

dUTPase staining is also observed in the cytoplasm of normal tissues that have a high mitochondrial content. Examples include myocardial tissue and the cells of the proximal convoluted tubules in the kidney. Similar staining with the polyclonal antibody Mito-M that immunoreacts with mitochondria supports this observation. The cytoplasmic expression of dUTPase is independent of cellular proliferation status and is found in both replicating and nonreplicating cell types. For example, the replicating cells at the base of the crypts of Lieberkühn in the colon demonstrate cytoplasmic dUTPase staining as well as the nonreplicating cells of the myocardium (Fig. 4). These data are in agreement with the constitutive expression of DUT-M observed in both quiescent and replicating PBLs and 34Lu cells. We have also shown that DUT-M mRNA levels remain constant in both quiescent and replicating cells (32). Taken together, these data suggest that the cytoplasmic staining detected in normal human tissues is indicative of DUT-M expression. Thus, DUT415 may be a useful marker in normal tissues to simultaneously identify replicating cell populations by nuclear staining and mitochondrial content through cytoplasmic staining.

Immunohistochemical staining of dUTPase in colon cancers demonstrates that dUTPase expression varies dramatically in different tumor samples,both in magnitude and intracellular localization. Patterns of dUTPase expression range from exclusive nuclear or exclusive cytoplasmic staining to tandem expression of both. A recent report by Fleischmann et al. (40) confirms these data, where these authors observed variable levels of dUTPase expression in colorectal tumors. Thus, dUTPase expression among patient tumor samples is highly variable, exhibiting distinct and recognizable phenotypes of staining that may be of significant prognostic value.

To examine the association between dUTPase expression and response to 5-FU-based chemotherapy and overall survival, we initiated a retrospective study including tumor specimens from 20 patients who had received protracted infusion of 5-FU and LV for treatment of metastatic colon cancer. In these colon cancer cases, nuclear dUTPase expression was associated with resistance to 5-FU-based chemotherapy(P = 0.005), shorter time to progression(P = 0.017), and a shorter median survival(P = 0.09). In contrast, tumors lacking nuclear dUTPase expression were more responsive to chemotherapy and had a longer time to progression and longer overall survival period (see Table 2). Significantly, no patients with nuclear positive dUTPase expression responded to 5-FU-based chemotherapy, and all responders demonstrated a nuclear negative phenotype. These data suggest that elevated expression of dUTPase in the nucleus of tumor cells may protect cells from the cytotoxic effect of uracil misincorporation induced by inhibition of thymidylate metabolism. Variable cytoplasmic expression of dUTPase was also observed in these tumor specimens;however, no association between cytoplasmic staining and response or survival was evident in this limited study.

The expression of nuclear dUTPase was also associated with TS gene expression within these tumors (P= 0.06). This observation may be expected because TS gene expression has already been shown to be a predictive marker of response to 5-FU-based chemotherapy in gastrointestinal cancers (19, 21, 28). Larger studies will be required to assess whether tandem determination of dUTPase and TS will provide a more accurate prediction of tumor response to chemotherapy and overall survival compared with each marker independently.

The association between dUTPase expression and response to chemotherapy has several implications: (a) this study represents the first clinical data that support uracil-DNA-mediated cytotoxicity as a molecular mechanism of response to 5-FU-based chemotherapy;(b) the utility of dUTPase as a prognostic marker is not limited to 5-FU alone. Because the action of dUTPase is a downstream event of TS inhibition, dUTPase may be a useful marker for a number of commonly used chemotherapeutics that target either TS or DHFR; and(c) dUTPase has long been considered a viable chemotherapeutic target (reviewed by McIntosh and Haynes41). Immunohistochemical detection of dUTPase isoforms will aid in the development and evaluation of dUTPase inhibitory compounds.

Variable cytoplasmic versus nuclear expression of dUTPase between tumor specimens raises a question about the consequences of TS inhibition on nuclear versus mtDNA. The existence of nuclear and mitochondrial isoforms of both dUTPase and uracil-DNA glycosylase suggest that the maintenance of uracil-free DNA is critical for the integrity of both nuclear DNA and mtDNA. Although there have been many studies of the effect of TS inhibition on nuclear DNA, little is known about the contribution of mtDNA degradation toward cell killing in tumors. Studies of the effects of thymidylate deprivation on lower eukaryotes and human cell lines suggest that there is a significant bias toward organelle DNA-specific mutagenesis and degradation(42, 43, 44, 45). Studies of HeLa cells and normal human fibroblasts demonstrate that methotrexate and FUdR both induce mitochondria-specific mutagenesis and DNA degradation, an effect that is reversed by the addition of thymidine (42). These data suggest that in certain cases, mtDNA degradation and not nuclear DNA degradation may be critical for response to chemotherapy.

Within this study, nuclear dUTPase expression was the important variable in predicting response and survival, suggesting that drug-induced nuclear DNA damage may be the ultimate mediator of cytotoxicity. Although this limited study failed to correlate cytoplasmic dUTPase expression with clinical response or survival, the striking variability of dUTPase expression in the cytoplasm of tumor cells derived from different patients suggest that overexpression of DUT-M may play a significant role in mediating protection of mtDNA from uracil-DNA-mediated degradation. The role of mtDNA degradation in overall fluoropyrimidine cytotoxicity and DUT-M-induced resistance awaits further investigation.

The results of this study demonstrate that DUT415 is capable of detecting human dUTPase isoforms in formalin-fixed,paraffin-embedded tissues by immunohistochemistry. In normal tissues, nuclear expression of dUTPase is observed in replicating cell types, whereas cytoplasmic expression is observed in mitochondria-rich tissues. In addition, we have measured dUTPase expression in colon cancer tumor specimens and correlated these data with clinical outcome and response to chemotherapy. Although additional studies will be needed to further establish the clinical utility of this marker, these data strongly implicate dUTPase as a significant predictor of survival and response to 5-FU-based chemotherapy in metastatic colon cancer. The ability to measure dUTPase expression in patient tumor specimens by immunohistochemistry will aid in determining the role of dUTPase in drug resistance and may lead to improved therapeutic strategies for patients treated with inhibitors of thymidylate metabolism.