Toxicity of Antiviral Nucleoside Analogs and the Human Mitochondrial DNA Polymerase . Allison A ...

46 Pages·2001·0.23 MB·English

JBC Papers in Press. Published on August 28, 2001 as Manuscript M106743200

Toxicity of Antiviral Nucleoside Analogs and the Human Mitochondrial DNA Polymerase†.

Allison A. Johnson‡, Adrian S. Ray§, Jeremiah Hanes‡, Zucai Suo(cid:1), Joseph M. Colacino(cid:1), Karen
S. Anderson§, and Kenneth A. Johnson‡*

‡Institute for Cellular and Molecular Biology
MBB 3.(cid:2)22, A4800
University of Texas at Austin
Austin, Texas 787(cid:2)2
D
o
w
n
lo
§Yale University School of Medicine ad
e
d
Department of Pharmacology fro
m
333 Cedar Street h
New Haven, Connecticut 06520 ttp://w
w
w
.jb
(cid:1)Infectious Diseases Research c.org
Lilly Research Laboratories b/
y
g
Eli Lilly and Company u
e
s
Indianapolis, Indiana 46285 t o
n
Ap
ril 5
*Author to whom correspondence should be addressed. , 20
1
9
Tel: (5(cid:2)2) 47(cid:2)-0434
Fax: (5(cid:2)2) 47(cid:2)-0435
E-mail: [email protected]

†This work was supported by National Institutes of Health Grants GM446(cid:2)3 to Kenneth A.
Johnson and GM4955(cid:2) to Karen S. Anderson and by the Welch Foundation (F-(cid:2)432) to Kenneth
A. Johnson .
Copyright 2001 by The American Society for Biochemistry and Molecular Biology, Inc.
2

ABBREVIATIONS
HIV-(cid:2) RT, Human Immunodeficiency Virus-(cid:2) reverse transcriptase; Pol γ, DNA polymerase γ;
dNTP, deoxynucleoside triphosphate; E, enzyme; d4T, 2’,3’-didehydro-2’,3’-dideoxythymidine;
ddA, 2’, 3’ -dideoxyadenosine; ddI, 2’,3’-dideoxyinosine; AZT, 3’-azido-2’,3’-
dideoxythymidine; FIAU, (cid:2)-(2-deoxy-2-fluoro-β-d-arabinofuranosyl)-5-iodouracil; CBV, (-)-cis-
2-amino-(cid:2),9-dihydro-9-(4-hydroxymethyl)-2-cyclopenten-(cid:2)-yl)-6H-purine-6-one; PMPA, (R)-9-
(2-phosphonylmethoxypropyl)adenine. The suffixes –MP or –TP are added to the drug
D
abbreviations to indicate their monophosphate or triphosphate forms, respectively. o
w
n
lo
a
d
e
d
fro
m
h
ttp
://w
w
w
.jb
c
.o
rg
b/
y
g
u
e
s
t o
n
A
p
ril 5
, 2
0
1
9
3
ABSTRACT
To examine the role of the mitochondrial polymerase (Pol γ) in clinically observed toxicity of
nucleoside analogs used to treat AIDS, we examined the kinetics of incorporation catalyzed by
Pol γ for each FDA-approved analog plus FIAU, (-)3TC and PMPA. We used recombinant
exonuclease-deficient (E200A), reconstituted human Pol γ holoenzyme in single turnover kinetic
studies to measure K (K ) and k (k ) to estimate the specificity constant (k /K ) for each
d m pol cat cat m
nucleoside-analog triphosphate. The specificity constants vary over 500,000-fold for the series:
ddC > FIAU > ddA (ddI) > d4T >> (+)3TC >> (-)3TC > PMPA > AZT >> CBV. Abacavir
D
(prodrug of CBV) and PMPA are two new drugs that are expected to be least toxic. Notably, the ow
n
lo
a
higher toxicities of d4T, ddC and ddA arose from their (cid:2)3 to 36-fold tighter binding relative to ded
fro
m
the normal dNTP even though their rates of incorporation were comparable to PMPApp and http
://w
w
AZT. We also examined the rate of exonuclease removal of each analog following w
.jb
c
incorporation. The rates varied from 0.06 to 0.0004 s-(cid:2) for the series: FIAU > (+)3TC ~ (-)3TC > .org
b/
y
g
u
CBV > AZT > PMPA ~ d4T >> ddA (ddI) >> ddC. Removal of ddC was too slow to measure es
t o
n
A
(<0.00002 s-(cid:2)). The high toxicity of dideoxy compounds, ddC and ddI (metabolized to ddA) may pril 5
, 2
0
be a combination of high rates of incorporation and ineffective exonuclease removal. 19
Conversely, the more effective excision of (-)3TC, CBV and AZT may contribute to lower
toxicity. FIAU is readily extended by the next correct base pair (0.(cid:2)3 s-(cid:2)) faster than it is
removed (0.06 s-(cid:2)) and therefore is stably incorporated and highly mutagenic. We define a
toxicity index for chain terminators to account for relative rates of incorporation versus removal.
These results provide a method to rapidly screen new analogs for potential toxicity.
4

Current treatment of HIV includes a cocktail, which generally consists of a combination
of nucleoside and nonnucleoside analogs directed against HIV RT, plus an inhibitor of HIV
protease. Treatment with this cocktail allows patients to coexist with a low level of virus for
years, but treatments are limited by the development of resistance of HIV to the drugs on the one
hand and toxicity of nucleoside analogs on the other. Toxicity of nucleoside analogs is
particularly troublesome for the long-term management of the viral infection. Nucleoside
analogs function as chain terminators to suppress viral replication by HIV-(cid:2) RT and because
HIV RT lacks a proofreading exonuclease, the specificity of nucleoside analogs towards HIV RT D
o
w
n
lo
a
results from selective discrimination during incorporation and/or from removal by the d
e
d
fro
m
proofreading exonuclease of the host DNA polymerase. h
ttp
://w
Six nucleoside analogs have received FDA approval for treatment of HIV, and these w
w
.jb
c
analogs are illustrated with others in Figure (cid:2). All of these analogs act to halt viral replication by .org
b/
y
g
chain termination since they each lack a 3’ OH. There is growing evidence that these analogs u
e
s
t o
n
also act as substrates for human mitochondrial DNA polymerase (Pol γ), resulting in inhibition of Ap
ril 5
mitochondrial replication ((cid:2)-5). Patients experience side effects due to toxicity of these drugs, , 201
9
which are, in part, caused by interaction with Pol γ (2;6). The differences in analog toxicity may
result from a combination of cellular uptake (7), organellar transport (2), metabolic activation
(7), incorporation (7), and removal or degradation from the system (2). Generally, the individual
contributions of these factors to analog toxicity is unknown.
General symptoms of nucleoside analog toxicity include peripheral neuropathy (seen with
d4T, ddI and ddC), myopathy and bone marrow suppression (seen with AZT), and pancreatitis
(seen with ddI) (8). Many of the side effects resulting from nucleoside analog treatment mimic
5
the symptoms of mitochondrial diseases caused by genetic defects (4). Moreover, loss of
mitochondrial DNA and changes in mitochondrial ultrastructure have been observed in cell
culture studies after treatment with nucleoside analogs (9-(cid:2)6). A summary of symptoms
resulting from mitochondrial diseases and the nucleoside analogs that can cause these symptoms
is presented in Table (cid:2) ((cid:2)7). With the exception of FIAU, which is not a chain terminator, these
side effects are reversible upon withdrawal of drug treatment, suggesting removal of the chain
terminators from the mitochondrial DNA and restoration of normal rates of replication.
Nucleoside analog toxicity has been studied extensively in animal and cell culture
models. Studies using cell lines with compartment-specific kinases showed that nucleoside D
o
w
n
lo
a
analogs that were phosphorylated in the mitochondria became trapped in this compartment, and d
e
d
fro
m
that the location of phosphorylation was important in determining whether an analog was h
ttp
incorporated into mitochondrial or nuclear DNA ((cid:2)8). These data are useful in analyzing overall ://ww
w
.jb
c
drug toxicity, but the differences in nucleoside analog phosphorylation and the number of .org
b/
y
g
mitochondria per cell are based on the energetic requirements of each cell type, which can lead u
e
s
t o
n
to a considerable variation of observed cell- and organ-specific toxicity. Ap
ril 5
Structure/function studies with nucleoside analogs and Pol γ may provide useful insight , 20
1
9
for design of less toxic, viral polymerase-specific analogs. We have examined the interactions
between human Pol γ and nucleoside analogs during in vitro polymerization to evaluate the
possible mechanistic basis of nucleoside analog toxicity. To provide directly quantifiable data,
we have performed transient kinetic studies to carefully examine the kinetics of incorporation of
each of the nucleoside analog triphosphates by Pol γ and to compare to our previous analysis of
HIV RT ((cid:2)9-2(cid:2)).
6
METHODS
The nucleoside analogs in this study were the triphosphate forms of: 2’,3’-didehydro-
2’,3’-dideoxythymidine (d4T, Stavudine); 2’,3’-dideoxyadenosine (active form of ddI,
Didanosine); 3’-azido-2’,3’-dideoxythymidine (AZT, Zidovudine); (cid:2)-(2-deoxy-2-fluoro-β-d-
arabinofuranosyl)-5-iodouracil (FIAU); (-)-cis-2-amino-(cid:2),9-dihydro-9-(4-hydroxymethyl)-2-
cyclopenten-(cid:2)-yl)-6H-purine-6-one (Carbovir, CBV, active form of Abacavir), and (R)-9-(2-
phosphonylmethoxypropyl)adenine (PMPA, Tenofovir DF). ddATP was obtained from Sigma,
while d4TTP and AZTTP were obtained from Moravek Biochemicals (Brea, CA). PMPApp and
D
CBVTP were kind gifts of Gilead Sciences (Foster City, CA) and Dr. William B. Parker o
w
n
lo
a
(Southern Research Institute, Birmingham, AL), respectively. de
d
fro
m
Overexpression and purification of recombinant human Pol γ were previously described h
ttp
://w
(22;23). Holoenzyme was reconstituted at a (cid:2):5 ratio of catalytic subunit to accessory subunit. ww
.jb
c
.o
All of the studies on the kinetics of incorporation of nucleoside analog triphosphates were rg
b/
y
g
u
conducted using an exonuclease deficient mutant (E200A). This mutant was selected based upon e
s
t o
n
A
studies showing that this single point mutation did not alter the kinetics of normal nucleotide p
ril 5
, 2
0
incorporation (data not shown). All other studies were conducted using wild-type enzyme. 1
9
A DNA primer/template of 25 and 45 nucleotides, respectively, was employed for all
experiments except for CBV incorporation and excision. The primer sequence was 5’-
GCCTCGCAGCCGTCCAACCAACTCA-3’; and the template sequence was 5’-
GGACGGCATTGGATCGAGGTTGAGTTGGTTGGACGGCTGCGAGGC-3’. Template
sequence opposite the 26th primer position was altered to allow correct base pairing for each
analog incorporation reaction. The 30-mer primer utilized for CBVTP experiments was 5’-
GCCTCGCAGCCGTCCAACCAACTCAACCTC-3’ and the corresponding 45-mer template
7
was used. Primers were 5’-32P labeled with T4 polynucleotide kinase according to the
manufacturer’s instructions (Life Technologies, Gaithersburg, MD). The reaction was
terminated by incubation at 95 o C for 5 min. Unincorporated nucleotide was removed using a
Bio-Spin 6 column (Biorad, Hercules, CA). Primer was annealed to 45mer template by
combining at an equimolar ratio, heating to 95 o C and cooling to room temperature.
Primers containing 3’ terminal nucleoside analog residues were created to examine
excision of each analog. A polymerase reaction was performed using (cid:2) µM polymerase, 3 µM
duplex DNA (containing appropriate template base opposite 26th primer position), and 50 µM
analog in standard reaction buffer (50 mM Tris-Cl pH 7.5, (cid:2)00 mM NaCl, 2.5 mM MgCl ). Do
2 w
n
lo
a
Primer terminated with ddAMP, ddIMP and FIAUMP were created using Pol γ, and HIV-(cid:2) RT ded
fro
m
was used for primers terminated in AZTMP, CBVMP or d4TMP. Reaction mixtures were http
://w
incubated at 37 o C for 30 minutes and then the products were gel purified to obtain the analog- ww
.jb
c
.o
terminated primer. Primers were labeled and DNA duplexes were annealed as described above. rg
b/
y
g
u
All polymerase assays were performed at 37 o C in buffer containing 50 mM Tris-Cl, pH es
t o
n
A
7.5, (cid:2)00 mM NaCl, 2.5 mM MgCl2. The holoenzyme was reconstituted by incubating the two pril 5
, 2
0
protein subunits for 20 minutes on ice in reaction buffer lacking magnesium. DNA was then 19
added to the proteins and the mixture was incubated for an additional 20 minutes on ice.
Reactions were initiated by the addition of nucleotide and magnesium in the same reaction buffer
at 37 o C, and then quenched with 0.3 M EDTA. Alternatively, only magnesium was added to
initiate excision reactions. All concentrations given are final reaction conditions.
Polymerase and excision assay products were separated on (cid:2)5% denaturing
polyacrylamide sequencing gels, imaged on a Storm 860 and quantified using ImageQuaNT
8
software (Molecular Dynamics, Sunnyvale, CA) or a Bio-Rad GS525 Imager using Molecular
Analyst (Bio-Rad Laboratories, Inc. Hercules, CA).
K and maximum rate of incorporation for nucleoside analog incorporation.
d
Single nucleotide incorporation assays were performed with various concentrations of
nucleotide to examine the nucleotide concentration dependence of the incorporation rate. Single
turnover conditions were employed, where the concentration of enzyme is greater than the
concentration of DNA. Specific reaction conditions are listed in each figure legend. A time
course was performed for each concentration of dNTP. The concentration of extended product
26-mer, or 3(cid:2)-mer (for CBVTP incorporation) DNA was plotted against time and fit to a single D
o
w
n
lo
exponential ([product] = A*e-kt + C). The observed rates were then plotted against nucleotide ad
e
d
fro
m
concentrations and the data were fit to a hyperbola (observed rate = kpol[dNTP]/(Kd + [dNTP]) to http
://w
obtain the K and the maximum rate of polymerization, k , for each nucleoside analog. A w
d pol w
.jb
c
quadratic equation was used to fit the rate data for d4TTP and ddCTP incorporation due to the .org
b/
y
tightness of nucleotide binding ([dNTP] = 0.5(K + [dNTP] + k ) – [0.25(K + [dNTP] + k )2 – gu
d pol d pol es
t o
n
([dNTP]kpol)](cid:2)/2). In each case, because of the established mechanism for the polymerase (24), April 5
, 2
the measured K and k define the K and k for polymerization, respectively. However, our 0
d pol m cat 1
9
methods circumvent the numerous errors in attempting to measure steady state parameters for
incorporation with defined oligonucleotides and a processive enzyme, where DNA release from
the enzyme often limits the observed rate of polymerization.
Excision Reactions
Wild-type polymerase holoenzyme ((cid:2)00 nM) was preincubated with 75 nM 26/45 or
3(cid:2)/45 DNA containing a 3’ terminal nucleoside analog and MgCl was added to initiate the
2
hydrolysis reaction. The loss of full length substrate was quantified and plotted against time.
9
Data were fit to a single exponential ([primer] = A*e-kt + C) where k represents the rate of
excision (k ) in a single turnover.
exo

RESULTS
Incorporation of nucleoside analogs into DNA by Pol γ was investigated to determine the
kinetic parameters governing enzyme discrimination against incorporation of these substrates
based upon the supposition that those most easily incorporated and most difficult to remove
D
would be the most toxic. Single turnover conditions were employed, in that the concentration of o
w
n
lo
a
enzyme was in excess of the concentration of DNA. Kinetic studies on Pol γ and every other de
d
fro
m
DNA polymerase examined with natural deoxynucleoside triphosphate substrates have shown h
ttp
://w
that the presteady-state burst of polymerization defines the rate of incorporation while the slower ww
.jb
c
.o
steady state rate is limited by the rate of dissociation of DNA from the enzyme (25) (23;24;26- rg
b/
y
g
u
30). By restricting our attention to the first turnover and examining the reaction on a millisecond e
s
t o
n
A
time scale, we circumvent the slower reactions limiting the steady state measurements on p
ril 5
, 2
0
processive enzymes. 1
9
For each analog, the concentration dependence of the rate of incorporation was obtained
from a series of polymerization experiments performed at various nucleotide concentrations.
The concentration dependence defines the K for ground state nucleotide binding and the
d
maximum rate, k , for incorporation. Based upon knowledge of the mechanism of the reaction,
pol
these parameters define the K and k , respectively, for incorporation and the ratio k /K =
m cat pol d
k /K , the specificity constant, because nucleotide binding is a rapid equilibrium step and is
cat m
followed by a single rate-limiting incorporation reaction (25).
(cid:2)0
The concentration dependence of d4TTP incorporation by Pol γ is illustrated in Figure
2A. The rate of incorporation was established by a fit to a single exponential at each nucleotide
concentration. The nucleotide concentration dependence of the rate (Fig 2B) was fit to a
quadratic equation to yield a K for ground state binding for d4TTP of 44.7 ± 7.5 nM, and a
d
maximum rate of polymerization of 0.24 ± 0.0(cid:2) s-(cid:2) (Figure 2B). The data had to be fit to a
quadratic equation rather than a hyperbola because the K was less than the concentration of
d
enzyme-DNA used in the experiment. Importantly, the tight binding of d4TTP makes this
analog highly competitive for incorporation by Pol γ as defined by the specificity constant (Table
D
o
2) even though its rate of incorporation is not large. w
n
lo
a
d
e
Approximately 3% of ddI intracellularly converts to ddATP, the active form of the drug d
fro
m
(3(cid:2);32). Therefore, to examine the toxicity of ddI, we examined the kinetics of incorporation of http
://w
w
ddATP. The ground state nucleotide binding and maximum polymerization rate for ddATP w
.jb
c
.o
incorporation were 0.022 ± 0.009 µM and 0.3(cid:2) ± 0.02 s-(cid:2), respectively (Figure 3, Table 2). For brg/
y
g
u
e
completeness, we include the results of studies on ddI in summary Tables 2 and 4 (primary data st o
n
A
p
not shown). ril 5
, 2
0
1
The active form of Abacavir is CBVTP. Incorporation of CBVTP occurred at a 9
maximum polymerization rate of 0.00(cid:2)8 ± 0.000(cid:2) s-(cid:2), with a ground state nucleotide binding of
(cid:2)3 ± (cid:2).5 µM (Figure 4, Table 2). The maximum rate of incorporation of CBVTP was the
slowest we observed for the analogs in this study (Figure 4, Table 2). Pre-steady state burst
experiments in which primer/template is in slight excess of enzyme were used to show that the
rate limiting step for CBVTP incorporation had changed from the release of product to catalysis
(Figure 4A) due to the lack of a burst. With DNA in excess, natural dNTPs and most nucleoside
analog triphosphates show a pre-steady-state burst under these conditions reflecting fast catalysis