The Intracellular Localization of Deoxycytidine Kinase*

Deoxycytidine kinase (dCK) catalyzes the rate-limiting step of the deoxynucleoside salvage pathway in mammalian cells and plays a key role in the activation of several pharmacologically important nucleoside analogs. Using a highly specific polyclonal antibody raised against a C-terminal peptide of the human dCK, we analyzed its subcellular localization by Western blots of biochemically fractionated nuclear and cytoplasmic fractions as well as by in situ immunochemistry. Native dCK was found to be located mainly in the cytoplasm in several cell types, and the enzyme was more concentrated in the perinuclear and cellular membrane area. In contrast, when dCK was overexpressed in the cells, it was mainly located in the nucleus. The results demonstrate that native dCK is a cytoplasmic enzyme. However, it has the ability to enter the nucleus under certain conditions, suggesting the existence of a cytoplasmic retention mechanism that may have an important function in the regulation of the deoxynucleoside salvage pathway.

Deoxycytidine kinase (dCK) catalyzes the rate-limiting step of the deoxynucleoside salvage pathway in mammalian cells and plays a key role in the activation of several pharmacologically important nucleoside analogs. Using a highly specific polyclonal antibody raised against a C-terminal peptide of the human dCK, we analyzed its subcellular localization by Western blots of biochemically fractionated nuclear and cytoplasmic fractions as well as by in situ immunochemistry. Native dCK was found to be located mainly in the cytoplasm in several cell types, and the enzyme was more concentrated in the perinuclear and cellular membrane area. In contrast, when dCK was overexpressed in the cells, it was mainly located in the nucleus. The results demonstrate that native dCK is a cytoplasmic enzyme. However, it has the ability to enter the nucleus under certain conditions, suggesting the existence of a cytoplasmic retention mechanism that may have an important function in the regulation of the deoxynucleoside salvage pathway.
For these reasons, elucidation of potential regulatory mechanisms playing a role in dCK activation is of particular interest. Although much is known about the biochemical properties of dCK both in vitro and in vivo, little information is available on its regulation at the transcriptional or posttranslational levels. The constant amounts of dCK protein throughout the cell cycle suggest that its expression is not cell cycle-regulated (1,11). On the other hand, dCK is expressed predominantly in lymphoid cells, which indicates cell type-specific regulation (12)(13)(14)(15). In addition, dCK expression is up-regulated in certain solid tumors (16). The molecular mechanism that leads to tissue-specific and proliferation-dependent transcription of dCK is not yet clear.
Another level of control would involve regulated changes in the subcellular location of the enzyme. This assumption gained ground by the recent unexpected finding that a transfected dCK-green fluorescent protein (GFP) fusion protein was located mainly in the nucleus, contrasting with previous descriptions of dCK as a "cytoplasmic" protein (17). This finding may have important implications for the regulation of the deoxynucleotide metabolism, especially with respect to the functional compartmentalization of the dNTP pools. The determination of the actual location of native dCK is therefore of paramount interest.
In this paper we examined the intracellular localization of dCK in several cell types under physiological conditions. Using a highly specific C-terminal peptide antibody, we provide evidence that dCK is normally located in the cytoplasm. On the other hand, when dCK is overexpressed, it is found mainly in the nucleus, suggesting that a cytoplasmic retention regulatory mechanism may be functioning in living cells.

MATERIALS AND METHODS
dCK Antiserum Production and Immunoblot Assays-A peptide (NH 2 -YESLVEKVKEFLSTL-COOH) was synthesized that corresponded to amino acids 246 -260 of the human dCK protein (12), linked to keyhole limpet hemocyanin via maleimidobenzoyl-N-hydroxy succinimide ester, pH 7.0, as described (18). New Zealand White rabbits were injected subcutaneously with 0.1 mg of linked peptide in complete Freund's adjuvant and boosted five times with 0.04 mg of antigen in incomplete Freund's adjuvant at 2-week intervals. Antisera were collected and tested in Western blots containing various amounts of purified recombinant His-tagged human dCK (15), and whole cell extracts from CCRF-CEM cells or the araC-resistant, dCK-deficient cell line CEM Ϫ (AraC-8D (10)). For immunoblot analysis, proteins were separated on 12% SDS-polyacrylamide gels, transferred to nitrocellulose membranes, and probed with the dCK-pep antibody at 1:5000 dilution, followed by incubation with horseradish peroxidase-conjugated antirabbit IgG (Jackson Laboratories). Immuncomplexes were visualized by the enhanced chemiluminescence reaction (ECL, Amersham Pharmacia Biotech). The polyclonal antiserum for the P1 protein (TSG23, a kind gift from C. Hoog, Karolinska Institute) and the monoclonal antibody for the myc epitope (9E10 hybridoma, kindly provided by N. T. Ktistakis, Babraham Institute) were used at 1:4000 and 1:1000 dilutions, respectively.
Isolation of Subcellular Fractions-Preparations of nuclear and cytoplasmic extracts were carried out essentially as described (19), with some minor modifications. Briefly, 1-5 ϫ 10 7 cells were resuspended in isoosmotic lysis buffer containing 0.32 M sucrose, 3 mM CaCl 2 , 2 mM magnesium acetate, 0.1 mM EDTA, 10 mM Tris-HCl, pH 8.0, 1 mM * This work was supported by European Union Grant BIOMED BMH4-CT96-0479 and the Greek General Secretariat for Science and Technology. 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.
dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5% (v/v) Nonidet P-40. The lysates were centrifuged at 500 ϫ g for 5 min, and the supernatants were supplemented with 0.22 volume of cytoplasmic extraction buffer containing 0.15 M HEPES, pH 7.9, 0.7 M KCl, and 0.015 M MgCl 2 . After centrifugation at 12,000 ϫ g for 15 min, the resulting supernatants were supplemented with glycerol to 25% and stored at Ϫ70°C. The nuclear pellet was washed once in isoosmotic lysis buffer lacking Nonidet P-40 and incubated for 30 min at 37°C in a modified electrophoresis sample buffer containing 22% glycerol, 125 mM Tris-HCl, pH 6.8, 10 mM dithiothreitol, 1% SDS, 0.01% bromphenol blue, 5 mM MgCl 2 , and 75 units/ml Benzonase (Nicomed Pharma A/S). Before loading on the gel, samples were diluted with an equal volume of sample buffer without MgCl 2 and Benszonase.
Immunodepletion and Enzyme Activity Assays-100 l (50% suspension) of protein G-Sepharose (Amersham Pharmacia Biotech) beads were incubated with dCK-pep antiserum or preimmune serum (100 l of antiserum/mg of dry gel) for 1 h at 4°C, in a buffer containing 50 mM Tris-HCl, pH 7.5, 0.1 M KCl, 1 mM NaF, 1 mM EDTA, and 0.5% Nonidet P-40. The beads were washed with excess buffer and incubated with crude extracts from 5 ϫ 10 7 CCRF-CEM cells for 1 h. At the end of incubation, the beads were removed by centrifugation, and deoxycytidine kinase and thymidine kinase activities of the resulting supernatants were measured by the DEAE-cellulose filter binding assay as described previously (16,20).
Cell Culture and Transfections-A2780, HeLa, and CCRF-CEM cells were grown in Dulbecco's modified Eagle's Medium and RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum. The pCMV-myc-dCK expression vector was constructed by ligation of the blunt-ended NcoI-BamHI fragment from pET9d-dCK (15) into the blunt-ended EcoRI site of the pCMV-myc plasmid (Stratagene). The resulting construct contains the full-length human dCK cDNA in frame with the 10-amino acid-long myc epitope under the control of the strong cytomegalovirus (CMV) promoter. This construct was used to transfect CCRF-CEM cells by the Lipofectin method (Life Technologies, Inc.) or A2780 and HeLa cells by the calcium phosphate coprecipitation method as described previously (21,22). Transfected and untransfected cells as well as freshly isolated mouse thymocytes and splenocytes were transfered onto glass coverslips precoated with 50 g/ml poly-D-lysine and left to attach for 30 min. Nonattached cells were removed by washing the coverslips with phosphate-buffered saline (PBS) before fixation.
In Situ Immunofluorescence Analysis-Cells on coverslips were washed twice with PBS and fixed for 20 min with 3.7% formalin. After fixation the coverslips were washed twice with serum-free Dulbecco's modified Eagle's medium and incubated with PBS containing 1% bovine serum albumin (BSA/PBS). dCK-pep, myc-tag, and Sp1 ((Pep2)-G, Santa Cruz Biotechnology) antibodies were diluted 1:200, 1:20, and 1:150, respectively, in BSA/PBS and applied to the cells for 4 h. After washing with BSA/PBS, the cells were incubated with diluted (1:100) fluorescein isothiocyanate-conjugated anti-rabbit or anti-mouse secondary antibodies to IgG (Pierce) for 30 min, washed with BSA/PBS, rinsed in water, and mounted on glass slides with Mowiol (Polysciences) as described previously (21,22). Fluorescence signals were observed and photographed in a Leitz Dialux 20 EB microscope equipped with epifluorescence optics.

RESULTS
We have raised polyclonal antibodies against a 15-amino acid-long peptide corresponding to the very C-terminal region of human dCK. Purified recombinant human dCK, as well as crude extracts from CCRF-CEM cells and the araC-resistant cell line CEM Ϫ (AraC-8D (10)) were analyzed in Western blots using this antibody. The antibody readily reacted with recombinant dCK and with one single species of CCRF-CEM proteins, which had a molecular mass of ϳ30 kDa (Fig. 1, A and B). This band was missing in extracts prepared from the dCKdeficient CEM Ϫ cell line (Fig. 1B). Although smaller antibody dilutions or longer exposure times revealed some small background reactivity, the 30-kDa band was always predominant. These results demonstrate that the antibody is highly specific for dCK. Further evidence for the specificity was provided by dCK activity measurements in CCRF-CEM crude extracts. When these extracts were preincubated with dCK-pep antibody coupled to protein G-Sepharose, in vitro phosphorylation of deoxycytidine was greatly reduced, whereas phosphorylation of thymidine was not affected (Fig. 1C).
To study the intracellular localization of dCK, we first performed biochemical fractionation of cytoplasms and nuclei from three cell types of different origin, such as the human malignant T lymphoblastoid leukemia (CCRF-CEM), human ovarian carcinoma (A2780), and human cervical carcinoma (HeLa) cell lines. We used an isoosmotic/Nonidet P-40 lysis procedure (19) as an alternative to the commonly used hypotonic/Nonidet P-40 method to isolate intact nuclei and accurately compare the nuclear and cytoplasmic contents of dCK. Immunoblot analysis of these extracts showed that the bulk of cellular dCK protein was in the cytoplasmic fractions, whereas nuclei contained only a minor portion of the protein (Fig. 2A). To exclude the possibility of potential leakage of dCK from the nuclei to the cytoplasm during the fractionation procedure, we examined the distribution of the nuclear protein P1. P1 is a 105-kDa protein that is present only in the nucleus throughout the cell cycle in mammalian cells (23). We chose as a control P1 because it is a "free-floating" protein, not associated tightly with chromatin or structural components of the nucleus, making it a reliable marker for assaying nuclear protein leakage (23). Two strong bands at ϳ105 kDa were detected in the nuclear fraction of A2780 cells with P1 antiserum (Fig. 2B). Only trace amounts of these bands were seen in the cytoplasmic fractions, together with some smaller cross-reactive species (Fig. 2B), indicating that the integrity of nuclei during the fractionation procedure has been preserved to a high degree.
The above results suggest that dCK is a cytoplasmic protein. This is in contradiction with the findings of a previous study using a dCK-GFP fusion protein, which was located mainly in the nucleus (17). One possible reason for this discrepancy could be that when dCK is artificially overexpressed, its localization may be altered compared with its localization in cells expressing physiological levels of the protein. To test this possibility we have transiently transfected CCRF-CEM, A2780, and HeLa cells with an expression vector containing the human dCK cDNA under the control of the CMV promoter. We also fused a short, 10-amino acid-long myc-tag sequence to the N terminus of the dCK cDNA to follow the expression of the transgenederived, overexpressed protein independently of its endogenous counterpart. Western blot analysis using an antibody recognizing the myc epitope showed mainly nuclear localization of the overexpressed protein supplied in trans by the expression vector (Fig. 2C). Therefore, it seems likely that the nuclear localization of dCK observed by the dCK-GFP fusion protein (17) is attributable to its overexpression in the cells.
To verify these findings in situ, we performed indirect immunofluorescence experiments with the above cell lines, as well as with primary lymphocytes from mouse thymus and spleen. A ring-like peripheral staining pattern characteristic of the thin cytoplasmic region of lymphoid cells was observed in CCRF-CEM cells, splenocytes, and thymocytes with the dCKpep antiserum (Fig. 3, A, D, and E). Strong cytoplasmic but not nuclear staining was seen in A2780 and HeLa cells (Fig. 3, B and C). In these latter two cell types, especially in cells where as a consequence of less intense staining the resolution was relatively higher, an uneven cytoplasmic distribution of dCK was observed. dCK was mainly accumulated in the cytoplasmic membrane and perinuclear areas, locations that may relate to distinct functional compartments for deoxynucleoside metabolism. The specificity of the signal was verified by the lack of immunoreactivity when the antibody was applied to the cells in the presence of excess purified recombinant dCK protein (Fig.  3F).
In situ localization of the overexpressed dCK was analyzed by immunostaining of transfected cells with myc-tag antibody. The transgene-derived, overexpressed enzyme was found al-most exclusively in the nucleus in all three cell lines, as evidenced by the similar staining pattern obtained by the myc-tag antibody (Fig. 4, A-C) and the control antibody recognizing the nuclear protein Sp-1 (Fig. 4, D-F). The same distribution was observed when the cells were transfected with an untagged expression vector and stained with dCK-pep antibody (data not shown), excluding the possibility of artificial translocation of dCK from the cytoplasm to the nucleus as a result of the short myc epitope at the N terminus of the protein. Taken together, these results demonstrate that the nuclear localization of dCK in tranfected cells is a consequence of its highly elevated intracellular levels.

DISCUSSION
In mammalian cells deoxyribonucleotides for nuclear DNA replication and repair are synthesized via the de novo pathway, which involves reduction of ribonucleotide diphosphates by ribonucleotide reductase, or via direct phosphorylation of deoxyribonucleosides imported from the extracellular space or derived from dephosphorylation of dNTPs (24). Earlier studies trying to explain the functional link between dNTP precursor metabolism and DNA replication processes showed a physical association of DNA polymerase ␣, with several key enzymes involved deoxynucleotide metabolism (25). The "replitase model" has since been challenged by several observations, such as the findings that the enzymes thought to be components of the multienzyme complex were localized in different cellular compartments (26,27). In addition, a number of kinetic isotope incorporation experiments argue against the direct channeling of precursors to the sites of DNA replication (28). These studies led to the prevailing view that dNTPs are synthesized in the cytoplasm and subsequently translocate to the nucleus to participate in DNA replication or repair. The recent finding that a GFP-dCK fusion protein was located in the nucleus raised a potentially interesting exception to the above rule (17).
However, the possibility that native dCK is an exclusively nuclear protein generates a number of questions. Although this location would favor the direct supply of dCTP for DNA replication, it is difficult to explain certain results from kinetic isotope incorporation experiments. These studies showed the existence of two dCTP pools: one labeled mainly via the de novo pathway and used directly for DNA replication, and the other labeled via the salvage pathway and preferentially used for liponucleotide synthesis (29 -32). This second "cytoplasmic" dCTP pool is in rapid equilibrium with the one arising from the de novo synthesis and can be used for DNA synthesis efficiently (28 -32). It is, however, difficult to explain why the de novo pool, which is the primary source of nuclear DNA synthesis precursors, is built up by cytoplasmic enzymes such as ribonucleotide reductase, whereas the salvage pool, which has to equilibrate first with the de novo pool to be used for DNA synthesis, would be synthesized by an enzyme located in the nucleus. In addition, a nuclear dCK as a rate-limiting enzyme for the dCTP supply of liponucleotides is not compatible with the rapid labeling of these phospholipid precursors from deoxycytidine (29 -32), even if one takes into account a free passage of dCTP across the nuclear membrane.
The results presented in this paper clearly establish that native dCK is mainly located in the cytoplasm, a location more consistent with the previously mentioned metabolic studies. This conclusion is based on two independent approaches using biochemical cell fractionation and in situ immunochemistry experiments. A cytoplasmic localization was seen in three cell lines of different origin, as well as in primary thymocytes and splenocytes. In adherent cells, such as A2780 and HeLa, where the intracytoplasmic distribution is better resolved, we noticed denser staining of the cytoplasmic membrane and the perinu- clear area. It is tempting to speculate that this uneven distribution of dCK may be of functional relevance, corresponding to a compartmentalization that is related to the dCTP supplies for membrane phospholipid precursor synthesis and DNA replication, respectively. Alternatively, these sites may represent independent compartments for the phosphorylation of exogenous deoxycytidine immediately after its transport into the cell and deoxycytidine derived from endogenous dCTP breakdown, respectively. To resolve the contradiction between the localization of the GFP-dCK fusion protein (17) and the localization of the endogenous native protein, we asked whether the nuclear location of the former is attributable to its overexpression in the cells. We found that this is indeed the case, because after transfection of an expression vector coding for human dCK, which results in higher intracellular concentrations of dCK by several orders of magnitude, the overexpressed protein was mainly localized in the nucleus in all cell types tested. This finding strongly suggests that dCK has the ability to enter the nucleus. Further evidence for this notion was provided by the identification of a consensus nuclear import sequence at the N terminus of the protein, the mutation of which prevents nuclear entry even of the overexpressed GFP-dCK fusion protein (17).
Taking these findings together, we propose a cytoplasmic retention mechanism for dCK, which may involve the action of another protein complexed with dCK, that keeps it in the cytoplasm. When dCK is overexpressed, because of the limiting amounts of the cytoplasmic retention partner, the majority of dCK molecules will contain an accessible NLS sequence for recognition by the nuclear transport machinery. An analogous situation has been described for nuclear factor B (33, 34), the regulation of which involves cytokine-induced phosphorylation of its cytoplasmic retention partner, IB, which leads to the dissociation of the nuclear factor B/IB complex and the translocation of active nuclear factor B subunits into the nucleus. An alternative model would involve posttranslational modification of dCK itself, which may be required for nuclear entry in analogy to the transcription factors involved in the interferon signaling pathways (35). Although at this point the actual mechanism that retains dCK in the cytoplasm is not known, our results raise the interesting possibility that under certain physiological conditions transport of dCK between subcellular compartments may play an important role in the regulation of its known function in the supply of deoxynucleotides for different biological processes or for unknown additional functions of the protein.