Mammalian enzymes for preventing transcriptional
errors caused by oxidative damage
Toru Ishibashi1,2,*, Hiroshi Hayakawa3, Riyoko Ito2, Masayuki Miyazawa4,
Yuriko Yamagata4and Mutsuo Sekiguchi1,2
1Biomolecular Engineering Research Institute, Suita, Osaka 565-0874, Japan,2Department of Physiological Science
and Molecular Biology and Frontier Research Center, Fukuoka Dental College, Fukuoka 814-0193, Japan,
3Department of Medical Biochemistry, Graduate School of Medical Science, Kyushu University, Fukuoka 812-8582,
Japan and4Graduate School of Pharmaceutical Sciences, Kumamoto University, Kumamoto 862-0973, Japan
Received March 31, 2005; Revised and Accepted June 13, 2005
8-Oxo-7,8-dihydroguanine (8-oxoGua) is produced
in cells by reactive oxygen species normally formed
during cellular metabolic processes. This oxidized
base can pair with both adenine and cytosine, and
thus the existence of this base in messenger RNA
would cause translational errors. The MutT protein
of Escherichia coli degrades 8-oxoGua-containing
ribonucleoside di- and triphosphates to the mono-
phosphate, thereby preventing the misincorporation
of 8-oxoGua into RNA. Here, we show that for human
the MutT-related proteins, NUDT5 and MTH1 have
the ability to prevent translational errors caused by
oxidative damage. The increase in the production of
erroneous proteins by oxidative damage is 28-fold
over the wild-type cells in E.coli mutT deficient
cells. By the expression of NUDT5 or MTH1 in the
cells, it is reduced to 1.4- or 1.2-fold, respectively.
NUDT5 and MTH1 hydrolyze 8-oxoGDP to 8-oxoGMP
with Vmax/Kmvalues of 1.3 3 10?3and 1.7 3 10?3,
respectively, values which are considerably higher
than those for its normal counterpart, GDP (0.1–
0.5 3 10?3). MTH1, but not NUDT5, possesses an
additional activity to degrade 8-oxoGTP to the mono-
phosphate. These results indicate that the elimina-
tion of 8-oxoGua-containing ribonucleotides from
the precursor pool is important to ensure accurate
protein synthesis and that both NUDT5 and MTH1
are involved in this process in human cells.
Oxidation of the components of nucleic acids occurs through
normal cellular metabolism, especially in aerobic states. More
than 20 different types of oxidatively altered purine and
pyrimidine bases have been detected in nucleic acids (1,2).
Among them, an oxidized form of guanine base, 8-oxo-7,8-
dihydroguanine (8-oxoGua), has a potential to alter genetic
information, since it can pair with adenine and cytosine
at almost equal efficiencies (3–5). 8-OxoGua-containing nuc-
leotides can be incorporated into DNA as well as RNA, and
Studies of Escherichia coli mutator mutants have revealed
that organisms possess mechanisms to prevent spontaneous
mutations caused by misincorporation of 8-oxoGua into
DNA. In this organism, MutT protein, which has a potent
activity to degrade 8-oxoGua-containing deoxyribonucleo-
side triphosphate, is almost solely responsible for reducing
the level of mutagenic nucleotides in cell (9,10). Mammalian
cells also possess enzymes capable of eliminating 8-oxoGua-
containing nucleotides from the DNA precursor pool. These
include MTH1 (11,12), MTH2 (NUDT15) (13) and NUDT5
(14). MTH1 and MTH2 degrade preferentially 8-oxodGTP,
whereas NUDT5 specifically hydrolyses 8-oxodGDP but
hardly 8-oxodGTP. Despite their different substrate speci-
ficities, all of these mammalian proteins have abilities to
replace the MutT function. When each of the cDNAs for
these proteins was expressed in mutT-defective E.coli mutant
The error rate of transcription is estimated to be ?10?5per
residue (16), and the fidelity of transcription is worse in an
aerobic state. It has been shown that 8-oxoGua can be incor-
porated into RNA by the normal action of RNA polymerase,
and that the E.coli MutT protein has the ability to prevent this
misincorporation (17). The MutT catalyzes the hydrolysis
of both 8-oxoGDP and 8-oxoGTP to the monophosphate,
thereby cleaning up the nucleotide pool to ensure accurate
transcription (18). 8-OxoGua can be incorporated into RNA
bymammalian RNA polymerase II(19).Formammalian cells,
particularly those which rarely undergo cell division, the
was almost completely
*To whom correspondence should be addressed. Tel: +81 6 6872 8200; Fax: +81 6 6872 8210; Email: email@example.com
? The Author 2005. Published by Oxford University Press. All rights reserved.
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press
only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact firstname.lastname@example.org
Nucleic Acids Research, 2005, Vol. 33, No. 123779–3784
conservation of RNA seems to be important. In fact, a relat-
ively large amount of 8-oxoGua was formed in the RNA in
the neuronal tissues of patients suffering from some neuro-
degenerative diseases, such as Parkinson’s disease (20,21),
Alzheimer’s disease (21) and Down’s syndrome (22).
However, the mechanisms for cleaning up the ribonucleotide
pools of mammalian cells remain unclear. The actions of the
mammalian MutT-related proteins on 8-oxoGua-containing
ribonucleotides have not been reported, except for MTH1,
which has been shown to carry a weak hydrolytic activity
for 8-oxoGTP (19). Therefore, we initiated a survey of human
enzymes that act on 8-oxoGua-containing ribonucleotides.
Here, we report that among the three human MutT-related
proteins, NUDT5 and MTH1, but not MTH2, are capable
of eliminating oxidized ribonucleotides from the RNA
MATERIALS AND METHODS
Bacterial strains and plasmids
E.coli 101 (wild type) and 101T (mutT-deficient) carry an
amber mutation in codon 461 of the lacZ gene, where the
A:T to C:G transversion mutation is reversed phenotypically
to produce the b-galactosidase protein (17). Plasmid pTT100
is a derivative of pTrc99A (Amersham Pharmacia) lacking
the lacIqgene (15). pTT100::hMTH1 and pTT100::hNUDT5
were constructed as described previously (14,15). The human
MTH2 cDNA in the I.M.A.G.E. clone 4472716 (purchased
from Kurabo, gene bank accession no. 12761620) was PCR
amplified and cloned into the KpnI/HindIII site of pTT100
using two primers, 50-CGCTGGTACCATGACGGCCAGCG-
TAAACTG-30, to generate pTT100::hMTH2.
Measurement of b-galactosidase activity
5-Bromo-4-chloro-3-indolyl-b-D-galactoside (X-gal, Wako)
was dissolved in dimethylformamide (100 mg/ml) and
added to Luria–Bertani medium (10 g of bacto-tryptone, 5 g
of bacto-yeast extract and 10 g of NaCl per liter) to give a
final concentration of 0.5 mg/ml. Quantitative assays for
b-galactosidase activity were performed by measuring the
UV absorbance at 420 nm, which represents the hydro-
lysis of o-nitrophenyl-b-D-galactoside (ONPG, Sigma) to
o-nitrophenol. Three independent clones from each transform-
ant were used, and at least four experiments were performed
with each clone.
Hydrolysis of 8-oxoGua-containing ribonucleotides
8-OxoGTP, 8-oxoGDP and 8-oxoGMP were prepared as
described previously (23). The reaction mixture for the hydro-
lysisassaycontained20mM Tris–HCl, pH8.0,80mg/mlBSA,
8 mM MgCl2, 40 mM NaCl, 5 mM DTT, 2% glycerol and the
protein sample to be examined. The reaction was run at 30?C
for 10–30 min and was terminated by adding SDS to a final
concentration of0.1%.Thesampleswere fractionatedbyhigh-
performance liquid chromatography (HPLC) with a TSK-GEL
DEAE-2SW column (Tosoh) in an isocratic flow of 0.1 M
sodium phosphate (pH 6.0)–40% acetonitrile at a flow rate
of 0.9 ml/min. Nucleotides were quantified by measuring
the area of UV absorbance, using a Whatmann HPLC detec-
tion system and the Millennium program.The relative velocity
for hydrolysis was determined in time-course experiments.
The substrate concentrations for NUDT5 ranged from 1 to
50 mM for 8-oxoGDP, 20 to 500 mM for GDP and 50 to
2000 mM for 8-oxoGTP and GTP. For MTH1, they ranged
from 5 to 400 mM for 8-oxoGDP, 0.1 to 3 mM for GDP, 40 to
2000 mM for 8-oxoGTP and 0.2 to 4 mM for GTP. Kmand
Vmaxwere obtained from Lineweaver–Burk plots of the data.
Analysis of reaction products
The [g-32P]labeled 8-oxoGTP was prepared by the oxidation
of [g-32P]labeled GTP (MP Biomedicals, Inc.) and was puri-
fied as described previously (18). The enzyme reaction was
carried out as mentioned above, and an aliquot of the reaction
mixture was spotted onto a PEI-cellulose plate (MERCK) and
developed in 2 M LiCl/0.2 M Na2HPO4(1:1). Yeast pyrophos-
phatase was purchased from Sigma and the reaction was
performed at 30?C for 30 min.
Prevention of transcriptional errors caused
by oxidized guanine nucleotides
8-OxoGTP is produced by either the oxidation of GTP or the
phosphorylation of 8-oxoGDP, the latter of which is formed
by the oxidation of GDP (18). Once 8-oxoGTP is formed, this
oxidized nucleotide would be incorporated into RNA and
cause transcriptional errors (17). Since 8-oxoG is placed
opposite adenine in the DNA template, this misincorporation
would cause partial phenotypic suppression when an E.coli
lacZ?strain with a 50-T·A·G-30stop codon is used as a tester
strain (Figure 1A). In the presence of 8-oxoGTP, the 30-A·T·C-
50trinucleotide in the transcribed strand of the mutant would
be copied to 50-8-oxoG·A·G-30, which would then pair with
Figure 1. Partial phenotypic suppression by 8-oxoGua. (A) Schematic expla-
nation. a, Lac+cells; b, Lac?cells with an amber mutation at codon 461 of the
(G*) can be incorporated by the RNA polymerase opposite adenine in the
DNA template during transcription. The 8-oxoG·A·G could pair with the
CUC anticodon of tRNA for glutamic acid, thereby allowing elongation of
the b-galactosidase polypeptide chain. The cells would then have low levels of
(wild type) or 101T (mutT-deficient) cells harboring various plasmids were
cultures were placed in the wells. a, 101 (wild type); b, 101T (mutT?); c, 101T
with cDNA for human MTH2.
3780Nucleic Acids Research, 2005, Vol. 33, No. 12
50-C·U·C-30glutamic acid anticodon. This RNA transcript
would encode a wild-type b-galactosidase protein, whereas
the vast majority of the mRNAs encode truncated proteins.
We have used this assay system to examine the abilities of
human proteins to eliminate the mismatch-evoking oxidized
nucleotides from the RNA precursor pool.
The b-galactosidase activities produced by partial pheno-
typic suppression are relatively low, but can be detected
when cells are cultured in the presence of X-gal, as shown
in Figure 1B. E.coli 101 mutT+cells carrying an amber muta-
tion at codon 461 in the lacZ gene yield white colonies, since
they are unable to produce an active b-galactosidase protein.
On the other hand, 101T cells, which carry a mutT mutation in
addition to the lacZ amber mutation, produce blue colonies,
probably due to the partial phenotypic suppression of the lacZ
mutation caused by the misincorporation of 8-oxoGua into
mRNA (17) (Table 1). When the cDNA for either NUDT5
or MTH1 was introduced into the 101T cells, the formation of
blue colonies was almost completely suppressed, implying
that these human proteins can replace the defective MutT
function in E.coli cells. On the other hand, no suppression
was induced by the expression of MTH2.
More quantitative data were obtained by measuring the
actual b-galactosidase activities in the cultures of E.coli
101T cells harboring plasmids bearing various cDNAs. An
?30-fold increase in b-galactosidase activity was observed
in the mutT-defective 101T strain as compared with the
wild-type 101 strain. This increased level of enzyme activity
was almost completely suppressed by the expression of
the cDNA encoding either NUDT5 or MTH1 in the cells.
In accordance with the results of the colony color test, the
cDNA encoding MTH2 was unable to prevent the phenotypic
Actions of NUDT5 protein on oxidized guanine
The human NUDT5 protein was expressed as a His-tagged
protein in E.coli M15 cells and was purified to near homo-
geneity (14). Assays for enzyme activities were carried out
using 5 mM 8-oxoGDP or 8-oxoGTP, and the products were
analyzed by HPLC. As shown in Figure 2A, NUDT5
efficiently degraded 8-oxoGDP to its monophosphate form.
Under the same conditions, the hydrolysis of 8-oxoGTP was
The kinetic parameters of the NUDT5 enzyme (Kmand
Vmax) were measured for the hydrolysis of the normal and
oxidized forms of guanine ribonucleotides (Table 2). The
Kmfor the hydrolysis of 8-oxoGDP is ten times lower than
Table 1. Partial phenotypic suppression of an amber mutation of the lacZ gene
E.coli strain Plasmid
101 (wild type)
0.012 – 0.002
0.34 – 0.17
0.017 – 0.005
0.014 – 0.002
0.41 – 0.14
aAverage – standard error of all independent experiments for each
Figure 2. Actions of the NUDT5 and MTH1 proteins toward 8-oxoGTP and
8-oxoGDP. (A) Action of NUDT5. a, Reference nucleotides; b, treatment of
8-oxoGDP with NUDT5 protein; c, treatment of 8-oxoGTP with NUDT5 pro-
tein. (B) Action of MTH1. a, treatment of 8-oxoGDP with MTH1 protein; b,
treatment of 8-oxoGTP with MTH1 protein. In these experiments, 5 mM of
human N-terminal His-tagged NUDT5 protein at 30?C for 30 min for (A), or
10 mM 8-oxoGDP or 8-oxoGTP were incubated with a purified preparation of
Arrowheads indicate peaks for 8-oxoGMP (single black triangle), 8-oxoGDP
(double black triangles) and 8-oxoGTP (triple black triangles).
Nucleic Acids Research, 2005, Vol. 33, No. 12 3781
that for GDP. The Kmvalues for both 8-oxoGTP and GTP are
extremely high and the Vmaxvalues are very low, indicating
that these nucleotides are hardly degraded by NUDT5. On the
basis of these results, we may conclude that 8-oxoGDP is a
specific substrate for the human NUDT5 protein.
Hydrolysis of oxidized guanine ribonucleotides
by the MTH1 protein
The human MTH1 protein was purified to homogeneity (24).
The data presented in Figure 2B indicate that both 8-oxoGDP
and 8-oxoGTP can be hydrolyzed by the MTH1 protein. The
incubation of 8-oxoGDP with MTH1 effectively converted it
to the monophosphate. A more rapid conversion was observed
with 8-oxoGTP; almost all of the triphosphate was converted
to the monophosphate, under the same conditions.
The kinetic parameters of the MTH1 protein are also given
in Table 2. The Kmfor the hydrolysis of 8-oxoGDP is 10 times
lower than that for the normal counterpart, GDP, as observed
with NUDT5. However, the actual Kmvalues of the MTH1
protein are considerably higher than those of the NUDT5
protein. This low affinity of MTH1 for the substrates appears
to be compensated by its high velocity in the enzyme reaction.
It vastly exceeds the values of NUDT5 for 8-oxoGDP, thus
providing almost the same Vmax/Kmvalue for 8-oxoGDP. In
addition, the Kmof MTH1 for8-oxoGTPisrather high,and the
high Vmax/Kmvalue of this protein was attained by its extre-
mely high Vmaxin the 8-oxoGTP cleavage reaction. These
characteristics of the MTH1 protein could imply that the pro-
tein may act on the two types of error-evoking nucleotides,
8-oxoGDP and 8-oxoGTP, under different circumstances.
Another question regarding the action of MTH1 is whether
the 8-oxoGTP is directly converted to 8-oxoGMP or degraded
through the formation of 8-oxoGDP as an intermediate. To
8-oxoGTP was treated with a purified preparation of MTH1
protein and the products were analyzed on thin-layer chro-
matography. As shown in Figure 3, a spot for [32P]labeled
pyrophosphate appeared, which was further converted to
orthophosphate by a treatment with pyrophosphatase. It
appears that the MTH1 protein has the ability to cleave the
phosphoanhydride bond between the a and the b phosphate of
8-oxoGua-containing nucleoside di- and triphosphates.
The basal level of spontaneous errors in RNA synthesis is
estimated to be 10?5per residues, which is much higher
than the error rate of DNA replication (10?9per residues).
This means that numerous erroneous proteins are synthesized
in a normal cell. Furthermore, the transcriptional fidelity
would become lower when the RNA bases are modified
by internal or external agents. Among such modifications,
8-oxoGua is particularly important since this modified base
can pair with adenine and cytosine at almost equal efficiencies
(3–5). It has been shown that the incorporation of 8-oxoGua
opposite the adenine residues of DNA by RNA polymerase
causes partial phenotypic suppression (17). In this case,
although the majority of the proteins are normal, some of
the products are abnormal; erroneous proteins may lose
their functions or exhibit dominant characteristics to cause
disorders of some cellular functions, which could lead to cata-
strophic consequences. In mammals, many differentiated cells
remain in the G1/G0state and exert their cellular functions via
interactions with sophisticated networks. The dysfunction of a
single cell, caused by the accumulation of proteins translated
from the erroneous RNA, may be amplified with increasing
age. Therefore, mechanisms to control the RNA quality would
be important in facilitating the normal functions of organisms.
8-OxoGua can be formed in RNA by the direct oxidation of
the base and also by the incorporation of the oxidized base into
RNA (17). Once 8-oxoGua is formed in RNA, it cannot be
eliminated, in contrast to the case of in DNA, in which dam-
aged bases are excised by a specific glycosylase and repaired
(25). Thus, organisms must be equipped with other mechan-
isms to maintain the high quality of RNA against oxidative
stress. Proteins that specifically bind to oxidized RNA have
been implicated in a mechanism to scavenge damaged RNA.
The E.coli polynucleotide phosphorylase (PNP) protein and
the humanYB1 protein function in such mechanisms (26,27).
Another mechanism to prevent transcriptional errors caused
by oxidative damage is the sanitization of nucleotide pools.
Table 2. Substrate specificity of human NUDT5 and MTH1
Figure 3. Analysis of products cleaved by MTH1. Reaction mixtures (total of
5 ml) containing 50 pmol of [g-32P]labeled 8-oxoGTP were incubated with
on thin-layer chromatography. Lane 1, no enzyme; lane 2, treated with 5 ng
of purified MTH1 protein; lane 3, treated with 0.05 unit of yeast inorganic
pyrophosphatase; lane 4, treated with both 5 ng of MTH1 protein and 0.05 U
of yeast inorganic pyrophosphatase.
3782 Nucleic Acids Research, 2005, Vol. 33, No. 12
The E.coli MutT protein eliminates 8-oxoGTP and prevents
the occurrence of transcriptional errors, which are induced
particularly in the aerobic state (17). Recently, it was revealed
that the MutT protein has an additional activity to sanitize
the nucleotide pools; it degrades 8-oxoGDP as efficiently as
8-oxoGTP (18). However, in mammalian cells it is unclear
what types of enzymes are involved in this process. These
situations prompted us to search for enzymes that eliminate
8-oxoGua-containing ribonucleotides from the RNA precursor
pool in mammalian cells. In the present study, we identified
two MutT-related proteins, MTH1 and NUDT5, as the can-
didates responsible for this process.
Both MTH1 and NUDT5 exhibit considerably lower Km
values for 8-oxoGDP than for 8-oxoGTP. In the case of
MTH1,the selectivecleavage of8-oxoGTPwasalsoobserved.
Since 8-oxoGTP and 8-oxoGDP are interconvertible within
the cell (19), NUDT5 and MTH1 may collaborate to prevent
the misincorporation of 8-oxoGua into RNA. These situations
can be seen in Figure 4. The relatively high Vmax/Kmvalues
for the hydrolysis of 8-oxoGDP by NUDT5 and MTH1
(1.3 · 10?3and 1.7 · 10?3, respectively) support this view
(Figure 1). On the other hand, the Vmax/Kmof MTH1 for the
hydrolysis of 8-oxoGTP is considerably high, implying that
MTH1 is capable of eliminating 8-oxoGTP from the precursor
pool. In contrast to MTH1, NUDT5 hardly acts on 8-oxoGTP,
yet the expression of NUDT5 prevented of partial phenotypic
suppression in the mutT-deficient background. Taking all of
these results into account, it is conceivable that the hydrolysis
of 8-oxoGDP may be sufficient for the high fidelity of RNA
synthesis. To establish the biological significance of these
enzymes, it is necessary to construct mouse models deficient
in one or both of the activities. These studies are in progress in
Funding to pay the Open Access publication charges for this
article was provided by Biomolecular Engineering Research
Conflict of interest statement. None declared.
1. Gajewski,E., Rao,G., Nackerdien,Z. and Dizdaroglu,M. (1990)
Modification of DNA bases in mammalian chromatin by
radiation-generated free radicals. Biochemistry, 29, 7876–7882.
2. Demple,B. and Harrison,L. (1994) Repair of oxidative damage to DNA:
enzymology and biology. Annu. Rev. Biochem., 63, 915–948.
3. Wood,M.L., Dizdaroglu,M., Gajewski,E. and Essigmann,J.M. (1990)
Mechanistic studies of ionizing radiation and oxidative mutagenesis:
genetic effects of a single 8-hydroxyguanine (7-hydro-8-oxoguanine)
residue inserted at a unique site in a viral genome. Biochemistry,
4. Moriya,M., Ou,C., Bodepudi,V., Johnson,F., Takeshita,M. and
Grollman,A.P. (1991) Site-specific mutagenesis using a gapped duplex
vector: a study of translesion synthesis past 8-oxodeoxyguanosine in
E. coli. Mutat. Res., 254, 281–288.
bases during DNA synthesis past the oxidation-damaged base 8-oxodG.
Nature, 349, 431–434.
6. Kasai,H., Crain,P.F., Kuchino,Y., Nishimura,S., Ootsuyama,A. and
Tanooka,H. (1986) Formation of 8-hydroxyguanine moiety in cellular
DNA by agents producing oxygen radicals and evidence for its repair.
Carcinogenesis, 7, 1849–1851.
7. Ames,B.N. and Gold,L.S. (1991) Endogenous mutagens and the causes
of aging and cancer. Mutat. Res., 250, 3–16.
8. Sekiguchi,M. and Tsuzuki,T. (2002) Oxidative nucleotide damage:
consequences and prevention. Oncogene, 21, 8895–8904.
9. Yanofsky,C., Cox,E.C. and Horn,V. (1966) The unusual mutagenic
specificity of E. coli mutator gene. Proc. Natl Acad. Sci. USA, 55,
potent mutagenic substrate for DNA synthesis. Nature, 355, 273–275.
11. Mo,J.Y., Maki,H. and Sekiguchi,M. (1992) Hydrolytic elimination of a
mutagenic nucleotide, 8-oxodGTP, by human 18-kilodalton protein:
sanitization of nucleotide pool. Proc. Natl Acad. Sci. USA, 89,
12. Sakumi,K., Furuichi,M., Tsuzuki,T., Kakuma,T., Kawabata,S., Maki,H.
and Sekiguchi,M. (1993) Cloning and expression of cDNA for a human
enzyme that hydrolyzes 8-oxo-dGTP, a mutagenic substrate for DNA
synthesis. J. Biol. Chem., 268, 23524–23530.
13. Cai,J.P., Ishibashi,T., Takagi,Y., Hayakawa,H. and Sekiguchi,M. (2003)
nucleotides. Biochem. Biophys. Res. Commun., 305, 1073–1077.
14. Ishibashi,T.,Hayakawa,H.andSekiguchi,M. (2003)Anovelmechanism
for preventing mutations caused by oxidation of guanine nucleotides.
EMBO Rep., 4, 479–483.
and Sekiguchi,M. (1994) Genomic structure and chromosome location
of the human mutT homologue gene MTH1 encoding 8-oxo-dGTPase
for prevention of A:T to C:G transversion. Genomics, 24, 485–490.
Figure 4. A model for the exclusion of 8-oxoGua-containing ribonucleotides
from the RNA precursor pools in mammalian cells. 8-OxoGTP and 8-oxoGDP
side triphosphatase. Partial phenotypic suppression is caused by the mis-
incorporation of 8-oxoGTP into RNA by RNA polymerase II. 8-OxoGDP is
hydrolyzed to 8-oxoGMP by NUDT5 and MTH1, while 8-oxoGTP is cleaved
directly to 8-oxoGMP by MTH1. 8-OxoGMP, a non-utilizable form for RNA
are readily transported through the cell membrane and can be excreted into
Nucleic Acids Research, 2005, Vol. 33, No. 123783