(Received for publication, November 21, 1996, and in revised form, January 3, 1997)
From the Medical Nobel Institute, Department of Medical Biochemistry and Biophysics, Karolinska Institute, S-171 77 Stockholm, Sweden
Human thymidine kinase 2 (TK2) is a
deoxyribonucleoside kinase that phosphorylates thymidine,
deoxycytidine, and deoxyuridine. The enzyme also phosphorylates
anti-viral and anti-cancer nucleoside analogs. We have identified an
expressed sequence tag cDNA that encoded a 27.5-kDa protein 30%
similar to the human deoxycytidine kinase and deoxyguanosine kinase.
The protein was expressed in Escherichia coli and shown to
have similar substrate specificity as reported for purified native
human TK2. The recombinant TK2 was shown to phosphorylate the
anti-cancer nucleoside analog 2
,2
-difluorodeoxycytidine. Northern
blot analysis showed two mRNA species at 2.4 and 4.0 kilobases
predominantly expressed in liver, pancreas, muscle, and brain. We
identified a sequence-tagged site designed from the 3
region of the
TK2 cDNA. The sequence-tagged site has been mapped to 81-84
centimorgans from the top linkage group of chromosome 16, which
corresponds to the 16q22 region. Our data show that deoxycytidine
kinase, deoxyguanosine kinase, and TK2 belong to a family of closely
related enzymes. At the time of this report all four of the known human
deoxyribonucleoside kinases have been cloned. This provides the
opportunity to characterize their individual contribution to
therapeutic and toxic effects of nucleoside analogs.
Nucleoside analogs used in anti-cancer and anti-viral therapy are administered as pro-drugs, and their pharmacological actions are dependent upon intracellular phosphorylation by deoxyribonucleoside kinases. Biochemical studies show that there are four distinct human deoxyribonucleoside kinases with different subcellular locations (1). Deoxycytidine kinase (dCK)1 and thymidine kinase 1 (TK1) are cytosolic enzymes whereas deoxyguanosine kinase (dGK) and thymidine kinase 2 (TK2) are considered to be located in the mitochondria. The cDNAs encoding TK1, dCK, and dGK have been cloned (2-4).
Human TK2 purified from leukemic spleen phosphorylates the pyrimidines
thymidine, deoxycytidine, and deoxyuridine (5, 6). Purified TK2 also
phosphorylates the nucleoside analogs 3-azido-2
,3
-dideoxythymidine (AZT) and
1-(2-deoxy-2-fluoro-
-D-arabinofuranosyl)-5-iodouracil (FIAU) that are active against human immunodeficiency virus and hepatitis B virus, respectively (5-8). These nucleoside analogs are
also substrates of TK1, but the relative importance of the two
thymidine kinases for activation of these nucleoside analogs in
different tissues is not known. Both AZT and FIAU cause adverse effects
that are due to mitochondrial DNA damage (9). The suggested mitochondrial location of TK2 and the low TK1 expression in the affected tissues have implied that TK2-mediated phosphorylation may be
important.
We have recently cloned human mitochondrial dGK by identifying proteins
homologous to dCK encoded by expressed sequence tag (EST) cDNAs
(4). We now report the cloning of an EST cDNA that encoded an
enzyme with 30% similar primary structure as compared with dCK and
dGK. This enzyme had the same substrate specificity and size as
described for purified native human TK2. The cloning of all four human
deoxyribonucleoside kinases will make it possible to determine the
relative contribution of each enzyme to the therapeutic and toxic
effects of nucleoside analogs. These data are important for development
and evaluation of novel anti-cancer and anti-viral nucleoside analogs
to achieve optimal therapeutic index.
The GenBankTM sequence data base at the National Center of Biotechnology Information was accessed on the Internet World Wide Web.2 The Basic Local Alignment Search Tool (10) was used to identify the EST clone-encoded proteins homologous but not identical to predicted amino acid sequences of dCK (3) and dGK (4). The EST cDNA clones were ordered from Research Genetics Inc., and the DNA sequences were determined as described below.
Two primers were designed based upon the DNA sequence of the EST
clone (5-TCTGAATGGTGCCAGCTCCACTG and
5
-CTCCTATGGGCAATGCTTCCGATTCTCTG). The primers were used to
amplify the full-length cDNA from an adult liver cDNA Marathon
library (CLONTECH). A PCR with the Expand PCR Long Template System
(Boehringer Mannheim) was used as described in the CLONTECH manual. The
PCR products were cloned into pGEM-T plasmid vector (Promega). All DNA
sequences were determined with the automatic laser fluorescence
sequencer (Pharmacia Biotech Inc.).
Two PCR primers with 5
BamHI and SalI restriction enzyme sites were
designed based on the cDNA sequences
(5
-ATCGGATCCGTAGATAAAGAACAGGAAAAAGAG and
5
-ATCGTCGACCCTATGGGCAATGCTTCCGATTC). The PCR-amplified product was
cloned into pGEX-4T-1 plasmid vector (Pharmacia) to express the
cDNA-encoded protein fused to glutathione S-transferase.
The expression plasmid vector was transformed into the
Escherichia coli strain BL21(DE3)pLysS (Stratagene). A
transformed colony was inoculated in LB medium supplemented with 100 µg/ml ampicillin and 34 µg/ml chloramphenicol. The bacteria were
grown to A600 = 0.7, and protein expression was
induced with 0.5 mM
isopropyl-1-thio-
-D-galactopyranoside for 4 h. The
cells were harvested by centrifugation at 6000 × g for
10 min and resuspended in phosphate-buffered saline. The bacteria were
lysed by sonication for 10 min on ice, and the extract was
recentrifuged at 10,000 × g for 10 min. The crude
extract was loaded onto a glutathione-Sepharose 4B column (Pharmacia), and the recombinant protein was eluted in 100 mM Tris (pH
7.6) supplemented with 10 mM reduced glutathione
(Sigma). The size and purity of the recombinant
protein were determined by sodium dodecyl sulfate-polyacrylamide gel
electrophoresis (Phast system, Pharmacia).
The substrate specificity of the recombinant
enzyme was determined by a phosphoryl transferase assay with
[-32P]ATP as described previously (6). The assay was
performed in 50 mM Tris (pH 7.6), 10 mM
dithiothreitol, 5 mM MgCl2, 1 mM unlabeled ATP, 25 µCi of [
-32P]ATP (3000 Ci/mmol,
Amersham Life Science, Inc.), and indicated concentrations of
deoxyadenosine, deoxycytidine, deoxyguanosine, thymidine, deoxyuridine,
deoxyinosine, 9-
-D-arabinofuranosyladenine, 9-
-D-arabinofuranosylcytosine,
9-
-D-arabinofuranosylguanine, 9-
-D-arabinofuranosylthymine,
2
,3
-didehydro-2
,3
-dideoxythymidine, 2
,3
-dideoxycytidine,
3
-fluoro-2
,3
-dideoxythymidine, AZT, 2
,2
-difluorodeoxycytidine (dFdC), and
2-chloro-2
-deoxyadenosine. The polyethyleneimine cellulose sheets were
autoradiographed for 1-12 h, and the products were quantified with the
Image Master system (Pharmacia).
Human multiple tissue Northern blot
and human immune system multiple tissue Northern blot with
poly(A)+ RNA from different human tissues were purchased
from CLONTECH. A probe including base pairs 400-1900 of the TK2
cDNA was labeled with [-32P]dCTP (6000 Ci/mmol,
Amersham) (Prime-A-Gene, Promega) and hybridized at 42 °C in 50%
formamide hybridization buffer following the protocol provided by
CLONTECH. The blots were scanned and quantified with the Image Master
system.
We compared the
predicted amino acid sequences of EST cDNA clones deposited in
GenBankTM with the amino acid sequences of human dCK and
dGK. By this analysis we found one EST cDNA clone from a human
fetal brain cDNA library that encoded a protein that was homologous
but not identical to the C-terminal region of both dCK and dGK
(Integrated Molecular Analysis of Genomes and their Expression cloneID
41853) (Ref. 11). We used PCR with primers designed from the 5
sequence of the EST clone to amplify the full-length cDNA from an
adult liver cDNA library. Two products with different lengths were
amplified (Fig. 1). The isoforms, A and B, were
identical in the 3
region but differed in the 5
part. The open
reading frame of the cDNA encoded a protein
30% similar to both
dCK and dGK. The shorter isoform A encoded a 234-amino acid residue
protein with a predicted molecular mass of 27.5 kDa (Fig.
2). Isoform B contained an additional 0.7-kilobase pair
5
region but had no ATG start codon in-frame with the translated part
of the cDNA sequence. We recloned isoform B from a human fetal
brain cDNA library to exclude the possibility of artificial
mutations induced during cDNA synthesis or PCR amplification. The
sequences of isoform B cDNA from the adult liver and fetal brain
cDNA libraries were, however, identical. A splice acceptor site
(12) was identified in the isoform B sequence at the position where it
differed from isoform A (Fig. 1). We believe that the isoform B
cDNA is the product of a defectively spliced mRNA.
We expressed the cDNA-encoded protein as a fusion to glutathione
S-transferase. After purification, a major band at 60 kDa was detected by sodium dodecyl sulfate-polyacrylamide gel
electrophoresis (data not shown). A phosphoryl transferase assay was
used to determine the substrate specificity of the enzyme (Table
I). The enzyme efficiently phosphorylated the
pyrimidines deoxycytidine, thymidine, and deoxyuridine but not the
purines deoxyadenosine, deoxyguanosine, and deoxyinosine. Among the
tested nucleoside analogs, only
9-
-D-arabinofuranosylthymine and dFdC were substrates of
the enzyme. 9-
-D-Arabinofuranosyladenine, 9-
-D-arabinofuranosylcytosine,
9-
-D-arabinofuranosylguanine, 2
,3
-didehydro-2
,3
-dideoxythymidine, 2
,3
-dideoxycytidine, 3
-fluoro-2
,3
dideoxythymidine, AZT, and
2-chloro-2
-deoxyadenosine were not phosphorylated by the enzyme. The
substrate specificity of the recombinant enzyme is identical to
that described for purified native human TK2 with the exception
that the tissue-purified native enzyme has been reported to
phosphorylate AZT (5, 6). The nucleoside analog dFdC has not, to our
knowledge, previously been shown to be a substrate for TK2. Based on
the substrate specificity, the predicted molecular size, and the
sequence similarities to dCK and dGK, we conclude that the cloned
cDNA encodes human TK2.
|
We have previously compared
the sequences of dCK, dGK, and herpes simplex virus type-1 thymidine
kinase (4). The comparison shows that five of the six primary structure
motifs that are conserved in most herpesvirus thymidine kinases (13)
match the regions that are highly conserved between dCK and dGK.
Alignment of the predicted amino acid sequences of dCK, dGK, and TK2
showed that the five regions are also conserved in the TK2 sequence
(Fig. 3).
TK2 is considered to be located in the mitochondria (1, 5). Most
proteins that are translocated into the mitochondria have an N-terminal
signal sequence with characteristic properties such as high content of
basic and hydrophobic amino acid residues, few acidic residues, and an
amphipathic -helical structure (14). These features are present in
the N-terminal sequence of human mitochondrial dGK (4). The N-terminal
sequence of TK2, however, lacks several of these features, and we were
not able to identify a mitochondrial translocation signal motif in the
TK2 sequence.
We used multiple
tissue Northern blots to determine the length and expression pattern of
TK2 mRNA. Northern blot analysis of the same tissues has previously
been used to determine the expression patterns of dCK and dGK mRNAs
(4). The TK2 probe hybridized with two bands at 2.4 and
4.0 kb (Fig. 4). Both TK2 mRNA species were expressed in
all tissues with the highest levels in liver, pancreas, muscle, and
brain. No cross-hybridization with the 1.0-1.3-kb dGK mRNA or the
2.8-kb dCK mRNA was observed.
Chromosome Localization of the TK2 Gene
The PCR primers that
were used to map two of the sequence-tagged sites deposited in the
GenBankTM data base were derived from the 3 region of an
EST TK2 cDNA (sequence-tagged sites clone WI-11439 and WI-15863)
(15). Both sequence-tagged sites have been mapped between the gene
markers D16S400 and D16S421 by PCR screening of a human-rodent
radiation hybrid panel. The gene markers are located on chromosome 16 at 81-84 centimorgans from the top linkage group (16). We compared this location with the locations of other genes that have been mapped
with both fluorescence in situ hybridization and PCR
screening of the same radiation hybrid panel (16). The comparison
showed that the position of the human TK2 gene can be assigned to
chromosome 16q22 (Fig. 5).
We cloned and recombinantly expressed human TK2 cDNA. Thereby the cDNA sequences of all four known human deoxyribonucleoside kinases are available. The sequence data showed that human dCK, dGK, and TK2 are closely related. These three enzymes are also sequence-related to both prokaryotic and viral deoxyribonucleoside kinases (17, 18). Human TK1 cannot be included in this enzyme family based on sequence similarity as most of the substrate binding regions are different (2). The cell cycle regulation of TK1 also separates it from the other deoxyribonucleoside kinases. TK1 is strictly cell cycle- regulated and expressed only in the S-phase, whereas dCK, dGK, and TK2 are considered to be constitutively expressed throughout the cell cycle (1). The constitutively expressed enzymes are believed to supply nondividing cells with deoxyribonucleotides for DNA repair. As a result of the broad substrate specificities of dCK, dGK, and TK2, these enzymes are sufficient for the phosphorylation of all the naturally occurring purine and pyrimidine deoxyribonucleosides.
We were not able to identify a mitochondrial translocation signal in the primary structure of TK2. This does not exclude that the TK2 protein is translocated into the mitochondria, as there are other signals in addition to the typical N-terminal motif (14). There are, however, reports that TK2 activity also has been detected in the cytosol (19). One possibility is that both cytosolic and mitochondrial forms of TK2 exist, and the presently cloned TK2 cDNA encodes the cytosolic enzyme. By Northern blot analysis we have found two mRNA species with different lengths; further studies will be required to determine the differences between these TK2 mRNAs. The cloning of TK2 will allow immunocytochemistry analysis to determine the true intracellular location of TK2.
Human dCK and dGK have been mapped by fluorescence in situ hybridization to chromosomes 4q13.3-q21.1 and 2p13, respectively (20, 21). The TK2 gene has been located to chromosome 16 by electrophoretic separation of the human and mouse TK2 isoenzymes in a human-mouse somatic cell hybrid panel (22). Our data confirmed the location of the TK2 gene to chromosome 16 and showed that it was located at q22. This region is shared with the genes causing Bardet-Biedl syndrome 2, Marner type of cataract, and macular dystrophy type 1 (23, 24). No phenotypic features of these diseases imply, however, that alterations of the TK2 protein or TK2 expression is involved in the pathogenesis.
The nucleoside analog FIAU is a potent inhibitor of hepatitis B virus replication (25). However, long term treatment with this drug causes liver failure, pancreatitis, myopathy, and peripheral neuropathy (26). The adverse effects of FIAU therapy are due to mitochondrial DNA damage (9, 27). FIAU is phosphorylated in vitro by both TK1 and TK2 (8). The low or absent expression of TK1 in mature hepatocytes and the proposed mitochondrial localization of TK2 imply that TK2-mediated FIAU phosphorylation is important for both the therapeutic and adverse effects of FIAU (25). Our data showed that the TK2 mRNAs were predominantly expressed in liver, pancreas, muscle, and brain. Since these organs are mainly affected by FIAU toxicity, there may be a correlation between TK2 expression levels and FIAU toxicity. The anti-human immunodeficiency virus nucleoside analog AZT also causes adverse effects due to mitochondrial DNA damage (28). Low levels of AZT phosphorylating activity have been reported for native purified TK2 (5, 6). However, we were not able to detect any AZT phosphorylating activity by the recombinant TK2. Possible explanations of this discrepancy include post-translational modifications of TK2 that are not present in the recombinant enzyme or the possibility that the tissue-purified TK2 was contaminated with a small amount of TK1. Another possibility would be that we have cloned a new, previously not identified, isoenzyme of thymidine kinase. The identical size, specificity of natural substrates, and chromosomal localization of native TK2 and the cloned enzyme, however, strongly suggest that these proteins are identical.
dFdC is a nucleoside analog that is used in the therapy of several solid malignant tumors (29). The compound is phosphorylated by dCK (30). Our data showed that dFdC is also a substrate of the recombinant TK2. The pharmacological effects of dFdC phosphorylation by TK2 remain to be elucidated. The cloning of all the known human deoxyribonucleoside kinases will enable us to address the questions of the role of TK2 role in activation of FIAU, AZT, and dFdC. It will also be possible to determine the relative contribution of each deoxyribonucleoside kinase to the pharmacological effects of novel nucleoside analogs. These data are important for the development of new anti-cancer and anti-virus therapies on a rational basis.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U77088[GenBank] and U80628[GenBank].