80 203 DNA Damage and Repair

wea25324_ch20_636-676.indd Page 656
656
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
G-segment
T-segment
ATPase
ATP
B′
*
*
(b)
(a)
A′
(c)
ADP+Pi
(e)
*
*
*
(d)
*
Figure 20.23 Model of the segment-passing step in the
topoisomerase II reaction. Based on the crystal structure of
the enzyme, and other evidence, Wang and colleagues proposed the
following model: (a) The upper jaws of the enzyme open to bind
the DNA G-segment (a double-stranded DNA), which is the one that
will break to form a gate that will allow the other DNA segment to pass
through. This binding of DNA induces a conformational change in the
enzyme that brings the active-site tyrosines on the B9 domain into
position to attack the DNA. (b) The ATPase domain of each upper jaw
binds ATP (represented by an asterisk), and the upper jaw also binds
the double-stranded DNA T-segment, which will be passed through
force into the DNA. The stress of this force must be
overcome or it will resist progression of the replicating fork. The name given to this stress-release mechanism is the swivel. DNA gyrase is the leading
candidate for this role in E. coli. By pumping negative supercoils into the replicating DNA, DNA gyrase neutralizes the positive supercoils that would
otherwise halt replication.
20.3 DNA Damage and Repair
DNA can be damaged in many different ways, and this
damage, if left unrepaired, can lead to mutations: changes
in the base sequence of a DNA. This distinction is worth
emphasizing at the outset: DNA damage is not the same as
mutation, although it can lead to mutation. DNA damage
*
*
(c)
the G-segment. (c) In a series of conformational changes, including a
hypothetical intermediate (in brackets), the active site breaks the DNA
G-segment, and allows the T-segment to pass through into the lower
jaws. The front B9 domain during step (c) is transparent so the DNA
behind it can be seen. (d) The lower jaws open to release the
T-segment and the G-segment fragments are rejoined. (e) The enzyme
hydrolyzes the bound ATP, returning the enzyme to a state in which it
can accept another T-segment and repeat the segment-passing
process. (Source: Adapted from Berger, J.M., S.J. Gamblin, S.C. Harrison, and
J.C. Wang, Structure and mechanism of DNA topoisomerase II. Nature 379:231,
1996.)
is simply a chemical alteration to DNA. A mutation is a
change in a base pair. For example, the change from a G–C
pair to an ethyl-G–C pair is DNA damage; the change from
a G–C pair to any other natural base pair (A–T or T–A or
C–G) is a mutation. If a particular kind of DNA damage is
likely to lead to a mutation, we call it genotoxic. Indeed, we
will see in the next section that the ethyl-G in our example
is genotoxic because it is likely to mispair with T instead of
C during DNA replication. If this happens, then another
round of replication will place an A across from the mispaired T, and conversion of the normal G–C pair to an
A–T pair (a true mutation) will be complete. Notice that
this example illustrates the importance of DNA replication
in conversion of DNA damage to mutation.
Let us look at two common examples of DNA damage:
base modifications caused by alkylating agents and pyrimidine dimers caused by ultraviolet radiation. Then we will
examine the mechanisms that bacterial and eukaryotic cells
use to deal with such damage. Most of these mechanisms
involve DNA replication.
wea25324_ch20_636-676.indd Page 657
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
H
O
HC
O
H 2C
O
–O
P
N6
N7
C C
N1
N C
N3 CH
HC
O
O
CH3
C C
H N3
CH
C
O2
Thymine
N
dR
H
O
H
O6
N7
H2C
O4
Adenine
H
O
H
N4
C C
N1 H
N C
N3 C
N2 H
C CH
N3
CH
C N
O2
H
Guanine
H
dR
Cytosine
Figure 20.24 Electron-rich centers in DNA. The targets most
commonly attacked by electrophiles are the phosphodiester bonds,
N7 of guanine, and N3 of adenine (red); other targets are in blue.
Damage Caused by Alkylation of Bases
Some substances in our environment, both natural and
synthetic, are electrophilic, meaning electron- (or negative
charge-) loving. Thus, electrophiles seek centers of negative charge in other molecules and bind to them. Many
other environmental substances are metabolized in the
body to electrophilic compounds. One of the most obvious
centers of negative charge in biology is the DNA molecule.
Every nucleotide contains one full negative charge on the
phosphodiester bond and partial negative charges on the
bases. When electrophiles encounter these negative centers, they attack them, usually adding carbon-containing
groups called alkyl groups. Thus, we refer to this process
as alkylation.
Figure 20.24 shows the centers of negative charge in
DNA. Aside from the phosphodiester bonds, the favorite
sites of attack by alkylating agents are the N7 of guanine
and the N3 of adenine, but many other targets are available,
and different alkylating agents have different preferences
for these targets.
What are the consequences of alkylations at these DNA
sites? Consider the two predominant sites of alkylation, the
N7 of guanine and the N3 of adenine. N7 alkylation of
guanine does not change the base-pairing properties of the
target base and is generally harmless. Alkylation of the N3
of adenine is more serious because it creates a base (e.g.,
3-methyl adenine [3mA]) that cannot base-pair properly
with any other base—a so-called noncoding base. Because
a DNA polymerase does not recognize any base pair involving 3mA as correct, it stops at the 3mA damage, stalling DNA replication. Such blockage of DNA replication
can kill a cell, so we say it is cytotoxic. On the other hand,
as we will see later in this chapter, such stalled replication
can be resumed without repairing the damage, but the
mechanism of such resumption is error-prone and therefore leads to mutations.
Moreover, all of the nitrogen and oxygen atoms involved in base pairing (see Figure 20.24) are also subject to
alkylation, which can directly disrupt base pairing and lead
to mutation. The alkylation target that leads to most mutations is the O6 of guanine. Even though this atom is relatively rarely attacked by alkylating agents, such alkylations
are very mutagenic because they allow the product to basepair with thymine rather than cytosine. For example, consider the alkylation of the O6 of guanine by the common
laboratory mutagen ethylmethane sulfonate (EMS), which
transfers ethyl (CH3CH2) groups to DNA (Figure 20.25).
The alkylation of the guanine O6 changes the tautomeric
form (the pattern of double bonds) of the guanine so it
base-pairs naturally with thymine. This leads to the replacement of a G–C base pair by an A–T base pair.
Many environmental carcinogens, or cancer-causing
agents, are electrophiles that act by attacking DNA and
alkylating it. As we have just seen, this can lead to mutations. If the mutations occur in genes that control or otherwise influence cell division, they can cause a cell to lose
control over its replication and therefore change into a
cancer cell.
H
H
O
N
N
N
H
N
EMS
H
O
H3C
CH2
O
H
Guanine
CH2
N
O
N
O
N
N
H3C
N
N
S
O
Cytosine
Figure 20.25 Alkylation of guanine by EMS. At the left is a normal
guanine–cytosine base pair. Note the free O6 oxygen (red) on the
guanine. Ethylmethane sulfonate (EMS) donates an ethyl group (blue)
657
Transition
O
H
N
CH3
GC
N
N
AT
N
N
CH3
H
O
H
O6-ethylguanine
Thymine
to the O6 oxygen, creating O6-ethylguanine (right), which base-pairs
with thymine instead of cystosine. After one more round of replication,
an A–T base pair will have replaced a G–C pair.
wea25324_ch20_636-676.indd Page 658
658
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
SUMMARY Alkylating agents like ethylmethane sul-
fonate add alkyl groups to bases. Some of these
alkylations do not change base-pairing, so they are
innocuous. Others cause DNA replication to stall,
so they are cytotoxic, and can lead to mutations if
the cell attempts to replicate its DNA without
repairing the damage. Other alkylations change
the base-pairing properties of a base, so they are
mutagenic, and thus genotoxic.
Damage Caused by Ultraviolet Radiation
Ultraviolet (UV) radiation cross-links adjacent pyrimidines
on the same DNA strand, forming two major lesions.
Eighty to 90 percent of these are pyrimidine dimers (see
Figure 20.26), which are also called cyclobutane pyrimidine dimers (CPDs) because of the four-member cyclobutane ring that forms between the two bases. Ten to
20 percent of the lesions are (6-4) photoproducts, in which
A C T TGC
A C T=T GC
UV
TG A ACG
TG A ACG
(a)
Damage Caused by Gamma and X-Rays
5′
3′
H
O C
CH3
O C
CH3
N
C
O
C
H N
C
H
O
N
C
C
C
N
H
Cyclobutane ring
(b)
3′
the 6-carbon of one pyrimidine is linked to the 4-carbon of
an adjacent pyrimidine. Both of these products block DNA
replication because they are noninformative (non-coding):
The replication machinery cannot tell which bases to insert
opposite the lesion. As we will see, replication sometimes
proceeds anyway, and bases are inserted without benefit of
the base pairing that normally provides accuracy. If these
are the wrong bases, a mutation results.
Ultraviolet radiation has great biological significance; it
is present in sunlight, so most forms of life are exposed to it
to some extent. The mutagenicity of UV radiation explains
why sunlight can cause skin cancer: Its UV component
damages the DNA in skin cells, which leads to mutations
that sometimes cause those cells to lose control over their
division.
Given the dangers of UV radiation, we are fortunate to
have a shield—the ozone layer—in the earth’s upper atmosphere to absorb the bulk of such radiation. However,
scientists have noticed alarming holes in this protective
shield—the most prominent one located over Antarctica.
The causes of this ozone depletion are somewhat controversial, but they probably include the release of compounds
traditionally used in air conditioners and in plastics into
the atmosphere. Unless we can arrest the destruction of the
ozone layer, we are destined to suffer more of the effects of
UV radiation, including skin cancer.
The much more energetic gamma rays and x-rays, like UV
rays, can interact directly with the DNA molecule. However, they cause most of their damage by ionizing the molecules, especially water, surrounding the DNA. This forms
free radicals, chemical substances with an unpaired electron. These free radicals, especially those containing oxygen (e.g., OH?), are extremely reactive, and they immediately
attack neighboring molecules. When such a free radical attacks a DNA molecule, it can change a base, or it can cause
a single- or double-stranded break.
DNA bases are subject to at least 20 kinds of oxidative
damage, and these can be caused by reactive oxygen species
derived from ionizing radiation, or simply from normal
oxidative metabolism. The best-studied oxidatively damaged DNA base is 8-oxoguanine (oxoG), also known as
8-hydroxyguanine (Figure 20.27). DNA polymerases in
bacteria and eukaryotes misread oxoG as thymine and
O
5′
HN
Figure 20.26 Pyrimidine dimers. (a) Ultraviolet light cross-links two
pyrimidine bases (thymines in this case) on the top strand. This
distorts the DNA so that these two noncoding bases no longer pair
with their adenine partners. (b) The two bonds joining the two
pyrimidines form a four-member cyclobutane ring (pink).
H2N
Figure 20.27 8-oxoguanine.
H
N
O
N
N
H
wea25324_ch20_636-676.indd Page 659
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
insert adenine instead of cytosine, resulting in an oxoG–A
pair. Both bases in this pair are genotoxic because they
both will probably lead to mutations if they are not removed before the DNA replicates again.
Single-stranded breaks are ordinarily not serious because they are easily repaired, just by rejoining the ends of
the severed strand, but double-stranded breaks are very difficult to repair properly, so they frequently cause a lasting
mutation. Because ionizing radiation can break chromosomes, it is referred to not only as a mutagen, or mutationcausing substance, but also as a clastogen, which means
“breaker.”
659
CGTTAT
GCAATA
(a) UV
CGTTAT
GCAATA
(b)
Binding of DNA photolyase
CGTTAT
GCAATA
SUMMARY Different kinds of radiation cause dif-
ferent kinds of damage. Ultraviolet rays have comparatively low energy, and they cause a moderate
type of damage: pyrimidine dimers. Gamma and
x-rays are much more energetic. They ionize the
molecules around DNA and form highly reactive
free radicals that can attack DNA, altering bases or
breaking strands.
(c)
Absorption of light
(>300 nm)
CGTTAT
GCAATA
(d)
Breaking pyrimidine dimer
Release of enzyme
CGTTAT
Directly Undoing DNA Damage
One way to cope with DNA damage is to repair it, or restore it to its original, undamaged state. There are two
basic ways to do this: (1) Directly undo the damage, or
(2) remove the damaged section of DNA and fill it in with
new, undamaged DNA. Let us begin by looking at two
methods E. coli cells use to directly undo DNA damage.
In the late 1940s, Albert Kelner was trying to measure
the effect of temperature on repair of ultraviolet damage to
DNA in the bacterium Streptomyces. However, he noticed
that damage was repaired much faster in some bacterial
spores than in others kept at the same temperature. Obviously, some factor other than temperature was operating.
Finally, Kelner noticed that the spores whose damage was
repaired fastest were the ones kept most directly exposed to
light from a laboratory window. When he performed control
experiments with spores kept in the dark, he could detect
no repair at all. Renato Dulbecco soon observed the same
effect in bacteria infected with UV radiation-damaged
phages. It now appears that most forms of life share
this important mechanism of repair, which is termed photoreactivation, or light repair. However, placental mammals, including humans, do not have a photoreactivation
pathway.
It was discovered in the late 1950s that photoreactivation is catalyzed by an enzyme called photoreactivating
enzyme or photolyase. Actually, two separate enzymes catalyze the repair of CPDs and (6-4) photoproducts. The former is called CPD photolyase, or simply photolyase; the
latter is known as (6-4) photolyase. The CPD photolyase
GCAATA
Figure 20.28 Model for photoreactivation. (a) Ultraviolet radiation
causes a pyrimidine dimer to form. (b) The DNA photolyase enzyme
(red) binds to this region of the DNA. (c) The enzyme absorbs near-UV
to visible light. (d) The enzyme breaks the dimer and finally dissociates
from the repaired DNA.
operates by the mechanism sketched in Figure 20.28. First,
the enzyme detects and binds to the damaged DNA site (a
pyrimidine dimer). Then the enzyme absorbs light in the
UV-A to blue region of the spectrum, which activates it so
it can break the bonds holding the pyrimidine dimer together. This restores the pyrimidines to their original independent state. Finally, the enzyme dissociates from the
DNA; the damage is repaired.
Organisms ranging from E. coli to human beings can
directly reverse another kind of damage, alkylation of the
O6 of guanine. After DNA is methylated or ethylated, an
enzyme called O6-methylguanine methyltransferase
comes on the scene to repair the damage. It does this by
accepting the methyl or ethyl group itself, as outlined in
Figure 20.29.
The acceptor site on the enzyme for the alkyl group is
the sulfur atom of a cysteine residue. Strictly speaking, this
means that the methyltransferase does not fulfill one part
of the definition of an enzyme—that it be regenerated unchanged after the reaction. Instead, this protein seems to be
irreversibly inactivated, so we call it a “suicide enzyme” to
denote the fact that it “dies” in performing its function.
wea25324_ch20_636-676.indd Page 660
660
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
AGCGTA
CH3
TCGCAT
+
H
S
Enzyme
AGCGTA
TCGCAT
CH3
+
S
Enzyme
O6 -methylguanine methyltransferase
Figure 20.29 Mechanism of O6-methylguanine methyltransferase. A sulfhydryl group of the enzyme accepts the methyl group (blue) from a
guanine on the DNA, thus inactivating the enzyme.
The repair process is therefore expensive; each repair event
costs one protein molecule.
One more property of the O6-methylguanine methyltransferase is worth noting. The enzyme, at least in E. coli,
is induced by DNA alkylation. This means bacterial cells
that have already been exposed to alkylating agents are
more resistant to DNA damage than cells that have just
been exposed to such mutagens for the first time.
leaves an apurinic or apyrimidinic site (AP site), which is a
sugar without its purine or pyrimidine base. Once the AP
site is created, it is recognized by an AP endonuclease that
cuts, or nicks, the DNA strand on the 59-side of the AP site.
(The “endo” in endonuclease means the enzyme cuts inside
a DNA strand, not at a free end; Greek endo, meaning
within.) In E. coli, DNA phosphodiesterase removes the AP
SUMMARY Ultraviolet radiation damage to DNA
(pyrimidine dimers) can be directly repaired by a
DNA photolyase that uses energy from near-UV to
blue light to break the bonds holding the two
pyrimidines together. O6 alkylations on guanine
residues can be directly reversed by the suicide
enzyme O6-methylguanine methyltransferase, which
accepts the alkyl group onto one of its amino acids.
(a)
DNA glycosylase (base extrusion)
(b)
DNA glycosylase (base removal)
+
AP site
Excision Repair
The percentage of DNA damage products that can be handled by direct reversal is necessarily small. Most such
damage products involve neither pyrimidine dimers nor
O6-alkylguanine, so they must be handled by a different
mechanism. Most are removed by a process called excision
repair. The damaged DNA is first removed, then replaced
with fresh DNA, by one of two mechanisms: base excision
repair or nucleotide excision repair. Base excision repair is
more prevalent and usually works on common, relatively
subtle changes to DNA bases, such as chemical modifications caused by cellular agents. Nucleotide excision repair
generally deals with more drastic changes to bases, many of
which distort the DNA double helix. These changes tend to
be caused by mutagenic agents from outside of the cell. A
good example of such damage is a pyrimidine dimer caused
by UV light.
Base Excision Repair In base excision repair (BER), a
damaged base is recognized by an enzyme called DNA glycosylase, which distorts the DNA in such a way as to
extrude the damaged base out of its association with
its base-paired partner, then breaks the glycosidic bond
between the damaged base and its sugar (Figure 20.30). This
(c)
AP endonucleases
(d)
DNA phosphodiesterase
+
(e)
DNA polymerase I
(f)
DNA ligase
Figure 20.30 Base excision repair in E. coli. (a) DNA glycosylase
extrudes the damaged base (red). (b) DNA glycosylase removes the
extruded base, leaving an apurinic or apyrimidinic site on the bottom
DNA strand. (c) An AP endonuclease cuts the DNA on the 59-side of
the AP site. (d) DNA phosphodiesterase removes the AP-deoxyribose
phosphate (yellow block at right) that was left by the DNA glycosylase,
(e) DNA polymerase I fills in the gap and continues repair synthesis for
a few nucleotides downstream, degrading DNA and simultaneously
replacing it. (f) DNA ligase seals the nick left by the DNA polymerase.
wea25324_ch20_636-676.indd Page 661
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
sugar phosphate, then DNA polymerase I performs repair
synthesis by degrading DNA in the 59→39 direction, while
filling in with new DNA. But DNA polymerase cannot repair nicks, so DNA ligase seals the remaining nick to complete the job. Many different DNA glycosylases have
evolved to recognize different kinds of damaged bases.
Humans have at least eight of these enzymes. Because subtle
chemical modifications of bases frequently allow DNA
replication, but still cause miscoding, BER is important in
preventing mutations.
Most BER in eukaryotes proceeds by a pathway (Figure
20.31a–e), that is similar to BER in bacteria, except that
there is no participation by a DNA phosphodiesterase. Instead, DNA polymerase b fills in the gap left after AP-site
cleavage, and simultaneously removes the hanging sugarphosphate flap (blue). But this scheme has a fundamental
problem: Whereas DNA polymerase I in bacteria has a
built-in editing activity, DNA polymerase b does not. It
tends to make mistakes—about one every 4000 nt—and
3′
(a) Deamination
of cytosine
5′
ACGTGA A T C
TGCAU T T AG
5′
3′
5′
3′
3′
ACGTGA A T C
TGCAC T T AG
5′
3′
5′
5′
3′
(d) Gap filling
5′
5′
3′
(f) Inaccurate
gap filling
5′
3′
(g) Proofreading
ACGTGA A T C
TGCA T T AG
3′
5′
ACGTGA A T C
TGCA T T AG
(h) Accurate
gap filling
3′
3′
(c) Cleavage
at AP site
ACGTGA A T C
TGCAC T T AG
3′
5′
ACGTGA A T C
TGCA T T AG
(e) Ligation
3′
(b) Excision of uracil
5′
5′
ACGTGA A T C
TGCA T T T AG
5′
3′
Figure 20.31 The human BER pathway. (a) Spontaneous cytosine
deamination has converted a C (blue) to a U (orange) in the lower
strand of the DNA. (b) A glycosylase removes the uracil. (c) APE1
cleaves on the 59-side of the apyrimidinic site. (d) DNA polymerase
b correctly fills in the gap with a C (blue) and simultaneously removes
the hanging sugar-phosphate tag (green). (e) DNA ligase I seals the
nick, returning the DNA to normal. (f) Occasionally, the DNA polymerase
makes a mistake. This time it has incorporated a T (red) rather than a C,
leaving a mismatch at the 39-end of the fragment to the left of the nick.
(g) APE1 uses its 39-exonuclease to remove the mispaired T, again
leaving a gap. (h) This time, DNA polymerase b correctly places a C
(blue) across from the G. Now the mismatch is repaired, and the DNA
just needs to be ligated to be back to normal. (Source: Adapted from Jiricny,
J., An APE that proofreads. Nature 415 [2002] p. 593, f. 1.)
661
cannot repair them by itself. That may not sound so bad,
but considering that between 20,000 and 80,000 damaged
bases occur in our genomes every day, that error rate means
that the BER system would introduce about 5–20 mutations into our genome daily.
Fortunately, eukaryotic cells have a solution for that
problem. In 2002, Kai-Ming Chou and Yung-Chi Cheng
showed that the human apurinic/apyrimidinic (AP) endonuclease (APE1) works in conjunction with the DNA polymerase b to edit the latter enzyme’s mistakes. It had been
known for years that APE1 had a 39→59 exonuclease in
addition to its dominant endonuclease activity, but the exonuclease activity appeared to be too weak to be significant.
Chou and Cheng showed that, although the 39→59 exonuclease activity is indeed weak on properly base-paired
nucleotides, it is 50–150-fold stronger when faced with
a terminal mispair, such as would occur after DNA
polymerase b has performed inaccurate gap-filling
(Figure 20.31f).
DNA ligase I is relatively inefficient at ligating two adjacent DNA strands when one of them has a mispair at the
end, as in the structure after step f in Figure 20.31. In fact,
its efficiency in ligating such substrates is less than 10%. If
APE1 really does participate in repairing mispaired DNA
created by DNA polymerase b, one would expect it to
work with DNA ligase by repairing the mismatches and
stimulating the efficiency of the ligase. Chou and Cheng
used a reconstituted system with purified DNA ligase I,
DNA polymerase b, and APE1 to demonstrate that APE1
stimulated the efficiency of ligation in a concentrationdependent manner from ,10–95%. Thus, APE1 really does
appear to be the enzyme that repairs mismatches introduced by DNA polmerase b.
A special case of base excision repair occurs when cells
deal with 8-oxoguanine, which we encountered earlier in
this chapter as a consequence of oxidative damage to DNA.
Recall that oxoG tends to pair with A, forming oxoG–A
base pairs, and that both bases in this pair are genotoxic
because they both will probably take the wrong partner in
the next round of replication, causing mutations. In humans, these mutations lead to cancer. But aerobic organisms have evolved mechanisms for dealing with both of
these bases.
Gregory Verdine and colleagues elucidated the mechanism for dealing with the mispaired A in 2004. The enzyme
responsible is an adenine DNA glycosylase called MutY in
bacteria and hMYH in humans. It can remove an A that is
mispaired with oxoG, but it leaves a correctly paired C
alone. Moreover, it ignores all the A’s that are correctly
paired with T’s. How does it make those distinctions?
X-ray crystallography of a complex between MutY and
model DNAs containing oxoG would shed considerable
light on this problem, but those complexes were apparently
too unstable to crystallize. So Verdine and colleagues
formed a covalent disulfide bond between oxoG-containing
wea25324_ch20_636-676.indd Page 662
662
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
oligonucleotides and MutY, and the complexes held
together and formed crystals.
The crystal structure revealed close and specific contacts between the oxoG–A pair and the enzyme. Furthermore the adenine base is extruded, or “flipped out” such
that it loses contact with its oxoG partner, and enters the
active site of the enzyme. There, the glycosidic bond linking
the adenine to the deoxyribose sugar is severed, and the
adenine is thus removed from the DNA. By contrast, an
ordinary T–A base pair does not make these close and specific contacts, so those base pairs are left alone. Furthermore, an oxoG–C pair makes the same contacts between
the enzyme and the oxoG base as an oxoG–A pair does,
but the cytosine base is not extruded, so it does not enter
the enzyme’s active site, and therefore is not removed.
What about removing the oxoG itself? That BER process is initiated by another DNA glycosylase, known as the
oxoG repair enzyme, which cleaves the glycosidic bond
linking oxoG to its deoxyribose. In humans, this enzyme is
called hOGG1, and it can distinguish an oxoG–C pair from
a normal G–C pair, extrude the oxoG out of its association
with its C partner, and excise it.
SUMMARY Base excision repair (BER) typically acts
on subtle base damage. This process begins with a
DNA glycosylase, which extrudes a base in a damaged base pair, then clips out the damaged base,
leaving an apurinic or apyrimidinic site that attracts
the DNA repair enzymes that remove the remaining
deoxyribose phosphate and replace it with a normal
nucleotide. In bacteria, DNA polymerase I is the
enzyme that fills in the missing nucleotide in BER; in
eukaryotes, DNA polymerase b plays this role.
However, this enzyme makes mistakes, and has no
proofreading activity, so APE1 carries out the necessary proofreading. Repair of 8-oxoguanine sites in
DNA is a special case of BER, that can happen in
two ways. Since oxoG mispairs with A, the A can be
removed after DNA replication by a specialized
adenine DNA glycosylase. However, if replication has
not yet occurred, the oxoG will still be paired with C,
and the oxoG can be removed by another DNA glycosylase, the oxoG repair enzyme.
Nucleotide Excision Repair Bulky base damage, including
pyrimidine dimers, can be removed directly, without help
from a DNA glycosylase. In this nucleotide excision repair
(NER) pathway (Figure 20.32), the incising enzyme system
recognizes the strand with the bulky damage and makes cuts
on either side of the damage, removing an oligonucleotide
with the damage. The key enzyme E. coli cells use in this process is called the uvrABC endonuclease because it contains
three polypeptides, the products of the uvrA, uvrB, and
(a)
Nick
Excinuclease (UvrABC)
Nick
(b)
+
(c)
DNA polymerase I,
DNA ligase
Figure 20.32 Nucleotide excision repair in E. coli (a) The UvrABC
excinuclease cuts on either side of a bulky damaged base (red). This
causes removal (b) of an oligonucleotide 12 nt long. If the damage
were a pyrimidine dimer, then the oligonucleotide would be a 13-mer
instead of a 12-mer. (c) DNA polymerase I fills in the missing
nucleotides, using the top strand as template, and then DNA ligase
seals the nick to complete the task, as in base excision repair.
uvrC genes. This enzyme cuts the damaged DNA, producing an oligonucleotide that is 12–13 bases long, depending
on whether the damage affects one nucleotide (alkylations)
or two (pyrimidine dimers). A more general term for the
enzyme system that catalyzes nucleotide excision repair is
excision nuclease, or excinuclease. As we will soon see, the
excinuclease in eukaryotic cells removes an oligonucleotide
about 24–32 nt long, rather than a 12- to 13-mer. In any
case, DNA polymerase fills in the gap left by the excised
oligonucleotide and DNA ligase seals the final nick.
Much of our information about repair mechanisms in
humans has come from the study of congenital defects in
DNA repair. These repair disorders cause a group of human
diseases, including Cockayne’s syndrome and xeroderma
pigmentosum (XP). Most XP patients are thousands of
times more likely to develop skin cancer than normal people
if they are exposed to the sun. In fact, their skin can become literally freckled with skin cancers. However, if XP
patients are kept out of sunlight, they suffer only normal
incidence of skin cancer. Even if XP patients are exposed to
sunlight, the parts of their skin that are shielded from light
have essentially no cancers. These findings underscore the
potency of sunlight as a mutating agent.
Why are XP patients so extraordinarily sensitive to sunlight? XP cells are defective in NER and therefore cannot
repair helix-distorting DNA damage, including pyrimidine
dimers, effectively. Thus, the damage persists and ultimately
wea25324_ch20_636-676.indd Page 663
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
leads to mutations, which ultimately lead to cancer. Because NER is also responsible for repairing chemically induced DNA damage that is helix-distorting, we would
expect XP patients to have a somewhat higher than average incidence of internal cancers caused by chemical mutagens, and they do. However, the incidence of such cancers
in XP patients is only marginally higher than that in normal people. This suggests that most internal DNA damage
in humans is not helix-distorting and we have an alternative pathway for correcting that milder kind of damage: the
BER pathway. But we have no alternative pathway for correcting UV damage because we do not have a photoreactivation system.
Nucleotide excision repair takes two forms in eukaryotes. It can involve all lesions throughout the genome
(global genome NER, or GG-NER), or it can be confined to
the transcribed strands in genetically active regions of the
genome (transcription-coupled NER, or TC-NER). The
mechanisms of these two forms of NER share many aspects
in common, but the method of recognition of the damage
differs, as we will see. Let us examine both processes as they
occur in humans.
Global Genome NER What repair steps are defective in
XP cells? There are at least eight answers to this question.
The problem has been investigated by fusing cells from different patients to see if the fused cells still show the defect.
(a) Damage
recognition
(b) TFΙΙH–helicase
melts DNA
663
Frequently they do not; instead, the genes from two different
patients complement each other. This probably means that a
different gene was defective in each patient. So far, seven different complementation groups affecting excision repair
have been identified this way. In addition, some patients
have a variant form of XP (XP-V) in which excision repair is
normal, and the patients’ cells are only slightly more sensitive to UV light than normal cells are. We will discuss the
gene responsible for XP-V later in this chapter. Taken together, these studies suggest that the defect can lie in any of
at least eight different genes. Seven of these genes are responsible for excision repair, and they are named XPA–XPG.
Most often, the first step in excision repair, incision, or cutting the affected DNA strand, seems to be defective.
The first step in human global genome NER (Figure 20.33) is the recognition of a distortion in the double helix
caused by DNA damage. This is where the first XP protein
(XPC) gets involved. XPC, together with another protein
called hHR23B, recognizes a lesion in the DNA, binds to it,
and causes melting of a small DNA region around the damage. This role in melting DNA is supported by in vitro studies performed in 1997 with templates that contain lesions
surrounded by or adjacent to a small “bubble” of melted
DNA. These templates do not require XPC, suggesting that
this protein’s job had already been performed when the
DNA was melted. Also, Jan Hoeijmakers and colleagues
used DNase footprinting in 1998 to show that XPC binds
XPA
RPA
XPC–hHR23B
TFΙΙH
(c) Incision by two
endonucleases
ERCCI–XPF
XPG
(d) DNA polymerase
ε/δ, DNA ligase
Figure 20.33 Human global genome NER. (a) In the damage
recognition step, the XPC–hHR23B complex recognizes the damage
(a pyrimidine dimer in this case), binds to it, and causes localized DNA
melting. XPA also aids this process. RPA binds to the undamaged
DNA strand across from the damage. (b) The DNA helicase activity of
TFIIH causes increased DNA melting. (c) RPA helps position two
endonucleases (the ERCC1–XPF complex and XPG) on either side
of the damage, and these endonucleases clip the DNA. (d) With
the damaged DNA removed on a fragment 24–32 nt long, DNA
polymerase fills in the gap with good DNA and DNA ligase seals the
final nick.
wea25324_ch20_636-676.indd Page 664
664
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
directly to a site of helix distortion in DNA and causes a
change in the DNA’s conformation (presumably a strand
separation).
XPA, which has an affinity for damaged DNA, is also
involved in an early stage of damage recognition. Because
both XPC and XPA can bind to damaged DNA, why do we
believe that XPC is the first factor on the scene? Competition studies performed by Hoeijmakers and colleagues,
with different sized templates, support this hypothesis.
These workers incubated XPC with one damaged template,
and all the other factors except XPC with the other damaged template. Then they mixed the two together. Repair
began first on the template that was originally incubated
with XPC alone, suggesting that XPC binds first to the
damaged DNA. Then what is the role of XPA? It can bind
to many of the other factors involved in NER, so it may
verify the presence of a DNA lesion in DNA that is already
denatured (by XPC or by other means), and help to recruit
the other NER factors.
At first, it may seem surprising to learn that two of the
other XP genes—XPB and XPD—code for two subunits of
the general transcription factor TFIIH, implicating this
general transcription factor in NER. However, we now
know that these two polypeptides have the DNA helicase
activity inherent in TFIIH (Chapter 11). So one role of
TFIIH is to enlarge the region of melted DNA around the
damage. But TFIIH is required for NER in vitro even with
damaged DNAs that have large melted regions, so this protein must have a function beyond providing DNA helicases. The fact that TFIIH interacts with a number of other
NER factors suggests that it serves as an organizer of the
NER complex.
The melting of the DNA by TFIIH attracts nucleases
that nick one strand on either side of the damage, excising
a 24–32-nt oligonucleotide that contains the damage. Two
excinucleases make the cuts on either side of the damaged
DNA. One is the XPG product, which cuts on the 39-side of
the damage. The other is a complex composed of a protein
called ERCC1 plus the XPF product, which cuts on the
59-side. These nucleases are ideally suited for their task:
They specifically cut DNA at the junction between doublestranded DNA and the single-stranded DNA created by the
TFIIH around the damage. Another protein known as RPA
helps position the two excinucleases for proper cleavage.
RPA is a single-strand-binding protein that binds preferentially to the undamaged strand across from the lesion. The
side of RPA facing toward the 39-end of this DNA strand
binds the ERCC1–XPF complex, and the other side of RPA
binds XPG. This automatically puts the two excinucleases
on the correct sides of the lesion.
Once the defective DNA is removed, DNA polymerase
ε or d fills in the gap, and DNA ligase seals the remaining
nick. The role of XPE is not clear yet. It appears not to
participate in NER, but it does bind to damaged DNA, so
it is presumably involved somehow in DNA repair.
Transcription-Coupled NER Transcription-coupled NER
uses all of the same factors as does global genome
NER, except for XPC. Because XPC appears to be responsible for initial damage recognition and limited DNA melting in GG-NER, what plays these roles in TC-NER? The
answer is RNA polymerase. When RNA polymerase encounters a distortion of the double helix caused by DNA
damage, it stalls. This places the bubble of melted DNA,
which is created by the polymerase, at the site of the lesion.
At that point, XPA could recognize the lesion in the denatured DNA and recruit the other factors. From that
point on, these factors would behave much as they do in
GG-NER, enlarging the melted region, clipping the DNA
in two places, and removing the piece of DNA containing
the lesion.
Consider the usefulness of RNA polymerase as a DNA
damage detector. It is constantly scanning the genome as it
transcribes, and lesions block its passage, demanding attention. Lesions in parts of the DNA that are not transcribed
(or even on the nontranscribed strand in a transcribed region) would not be detected this way, but they can wait
longer to be repaired because they are not blocking gene
expression. Thus, the fact that noncoding lesions such as
pyrimidine dimers and 3mA block transcription as well as
DNA replication is useful to the cell in that these lesions
stall the transcribing polymerase, which recruits the repair
machinery.
SUMMARY Nucleotide excision repair typically han-
dles bulky damage that distorts the DNA double
helix. NER in E. coli begins when the damaged DNA
is clipped by an endonuclease on either side of the
lesion, at sites 12–13 nt apart. This allows the damaged DNA to be removed as part of the resulting
12–13-base oligonucleotide. DNA polymerase I fills
the gap and DNA ligase seals the final nick. Eukaryotic NER follows two pathways. In GG-NER, a
complex composed of XPC and hHR23B initiates
repair by binding to a lesion anywhere in the genome
and causing a limited amount of DNA melting. This
protein apparently recruits XPA and RPA. TFIIH
then joins the complex, and two of its subunits (XPB
and XPD) use their DNA helicase activities to expand the melted region. RPA binds two excinucleases (XPF and XPG) and positions them for cleavage
of the DNA strand on either side of the lesion. This
releases the damage on a fragment between 24 and
32 nt long. TC-NER is very similar to GG-NER, except that RNA polymerase plays the role of XPC in
damage sensing and initial DNA melting. In either
kind of NER, DNA polymerase ε or d fills in the gap
left by the removal of the damaged fragment, and
DNA ligase seals the DNA.
wea25324_ch20_636-676.indd Page 665
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
665
Double-Strand Break Repair in Eukaryotes
Double-strand breaks in eukaryotes are probably the most
dangerous form of DNA damage. They are really broken
chromosomes, and if they are not repaired, they can lead to
cell death or, in vertebrates, to cancer. Eukaryotic cells deal
with double-strand breaks in DNA (DSBs) in two ways:
First, they can use homologous recombination, with the
unbroken sister chromatid as the recombining partner. This
mechanism is similar to recombination repair in bacteria,
discussed later in this chapter, except that both strands
must participate in recombination. Second, eukaryotic cells
can use nonhomologous end-joining (NHEJ). In replicating
cells in S and G2 phases, homologous recombination is the
dominant mechanism, because only one DNA copy is broken and the other is available to align the breaks properly.
Yeast cells, which divide frequently, rely primarily on homologous recombination to repair their double-strand
breaks. On the other hand, mammalian cells in G1 phase
preferentially use nonhomologous end-joining because the
DNA has not replicated and no second, homologous chromosome is yet available to serve as a template for repair. In
this section, we will focus on the latter mechanism.
Nonhomologous End-Joining J. Phillips and W. Morgan
investigated nonhomologous end-joining in 1994 by introducing a restriction endonuclease into Chinese hamster
ovary cells. This enzyme made double-stranded cuts in
chromosomes, including a site within the adenine phosphoribosyltransferase (APRT) gene, which was present in only
one copy in these cells. Then these workers looked for viable cells with mutations in the APRT gene and sequenced
the mutated genes to see what had happened during the rejoining process. They found mostly short insertions and deletions of DNA around the cleavage site. Furthermore, these
insertions and deletions appeared to have been directed by
microhomology—small areas of homology (1–6 bp)—in the
DNA ends. Figure 20.34 shows a model for nonhomologous end-joining that explains these and other findings.
First, the DNA ends attract Ku, a dimer of two polypeptides (Ku70 [Mr 5 69 kD] and Ku80 [Mr 5 83 kD]). One of
the important functions of this protein is to protect the
DNA ends from degradation until end-joining is complete.
Ku has DNA-dependent ATPase activity and is the regulatory subunit for DNA protein kinase (DNA-PK), whose
catalytic subunit is known as DNA-PKcs. X-ray crystallography studies have shown that Ku binds to DNA ends like
a ring on a finger. Its two subunits form a ring that is lined
with basic amino acids, which help it bind to acidic DNA.
Once Ku has bound to a DNA end, it can recruit the
DNA-PKcs and perhaps other proteins, completing the
DNA-PK complex. The protein complexes on each DNA
end have binding sites, not only for the DNA ends, but
also for double-stranded DNA adjacent to the ends. Thus,
these DNA-PK complexes, by binding to the other DNA
(a)
Binding Ku
(b)
Binding DNA-PKcs
(c)
Synapsis and transphosphorylation
P
P
P
(d)
P
Loss of catalic subunits and unwinding
P
P
(e)
Alignment
(f)
Flap resolution,
ligation
Figure 20.34 Model for nonhomologous end-joining. (a) Free DNA
ends attract Ku (blue), which protects them from degradation. (b) Ku
attracts DNA-PKcs (red), constituting the full DNA-PK complex.
(c) The DNA-PK complexes promote synapsis, or lining up of
regions of microhomology near the DNA ends. The two DNA-PK
complexes phosphorylate each other on both the regulatory (Ku)
and catalytic subunits. (d) The phosphorylation from step (c) has two
effects: (1) The phosphorylated catalytic subunits dissociate from
the complex. (2) Phosphorylation activates the DNA helicase activity
of Ku, which unwinds the two DNA ends. The phosophorylation of
Ku activates its DNA helicase activity, which unwinds the DNA
of the two ends. (e) Regions of microhomology in the two ends
base-pair with each other in the alignment step. (f) Flap resolution
removes extra flaps of DNA, and fills in gaps. Finally, DNA ligase
joins the ends of the DNA strands together permanently.
(Source: Adapted from Chu, G., Double strand break repair. Journal of Biological
Chemistry 272 [1997] p. 24099, f. 4.)
wea25324_ch20_636-676.indd Page 666
666
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
fragment, can promote synapsis, or lining up of regions of
microhomology.
The two DNA-PK complexes also phosphorylate each
other, which has two effects: First, the phosphorylation of
DNA-PKcs promotes dissociation of that catalytic subunit,
whose job is done. The phosphorylation of Ku activates its
DNA helicase activity, so it can promote unwinding of the
DNA ends. This unwinding allows regions of microhomology
to base-pair, leaving flaps composed of the ends of the other,
nonpairing strands. Finally, the flaps are removed by nucleases,
gaps are filled in, and the DNA strands are ligated together.
When the flaps are removed a few nucleotides of DNA
are lost, but this process is inherently inaccurate, and nucleotides can also be added. We will encounter nonhomologous end-joining again in Chapter 23 when we discuss
recombination of antibody genes. This scheme deliberately
introduces double-strand breaks into DNA and then rearranges the DNA fragments by joining selected free DNA
ends in a process that requires Ku.
SUMMARY Double-strand DNA breaks in mammals
can be repaired by homologous recombination or by
nonhomologous end joining. The latter process requires Ku and DNA-PKcs, which bind together at the
DNA ends, constituting active DNA-PK complexes
that allow the ends to find regions of microhomology
with each other. Once the regions of microhomology
line up, the two DNA-PK complexes phosphorylate
each other. This phosphorylation activates the catalytic subunit (DNA-PKcs) to dissociate, and it also activates the DNA helicase activity of Ku to unwind the
DNA ends so the microhomology regions can base-pair.
Finally, extra flaps of DNA are removed, gaps are filled,
and the DNA ends are ligated permanently together.
The Role of Chromatin Remodeling in Double-Stranded
Break Repair We learned in Chapter 13 that nucleosomes
can block association of gene control regions with transcription factors, and therefore that chromatin remodeling
is required for activation of eukaryotic genes. By the same
token, it seems reasonable to expect that nucleosomes
would block association between damaged DNA and repair factors, and therefore that chromatin remodeling
would be required for DNA repair. Indeed, work in 2004
by Susan Gasser and colleagues and by Xuetong Shen and
colleagues showed that double-stranded chromosome
break (DSB) repair in yeast, which is accomplished primarily by homologous recombination, depends on a chromatin
remodeling complex known as INO80.
INO80, a member of the SWI/SNF family of chromatin
remodelers (Chapter 13), is composed of 12 polypeptides,
including the ino80 gene product Ino80. This polypeptide
has the ATPase/translocase domain characteristic of
chromatin remodeling proteins. Mutations in ino80 block
both transcription and DSB repair, presumably because of
chromatin remodeling defects in both cases.
Both groups of investigators induced a unique doublestranded break at a defined site at the MAT locus in yeast
chromatin, then used chromatin immunoprecipitation
(ChIP, Chapter 13) to measure recruitment of proteins to
the break. INO80 appeared at the break within 30–60 min,
suggesting that it is involved in DSB repair. The next question concerns the other proteins that are required to recruit
INO80. One clue to the answer is that two yeast protein
kinases, Mec1 and Tel1, were already known to phosphorylate serine 129 of histone H2A on nucleosomes near DSBs,
and that replacement of serine 129 with alanine renders
yeast cells sensitive to radiation and chemicals that damage
DNA. Because alanine, unlike serine, cannot be phosphorylated, this finding indicates that phosphorylation of serine
129 on histone H2A promotes DSB repair.
Moreover, both groups showed that mutations in the
genes encoding Mec1 and Tel1, or mutations that changed
serine 129 to alanine, inhibited recruitment of INO80 to
DSBs. These findings suggested a direct interaction
between phosphorylated H2A and INO80. Indeed, Shen
and colleagues showed that INO80 co-purified with phosphorylated H2A and other histones, but not with unphosphorylated H2A.
What roles does INO80 play in DSB repair? Gasser
and colleagues showed that yeast strains with mutations
in genes encoding the subunits of INO80, or mutations
that changed serine 129 of histone H2A, do not form the
39-single-stranded overhangs at the broken ends of chromosomes with DSBs. Thus, formation of these essential
overhangs appears to be one of the functions of INO80,
and it could help in this process by sliding nucleosomes
away from the broken ends.
A suggestion for how INO80 could perform this remodeling comes from the finding that INO80 contains two
ATPases, Rvb1 and Rvb2, that are similar to RuvB, a protein involved in recombination and DSB repair in E. coli.
RuvB is composed of two cyclic hexamers of identical subunits (Chapter 22) and it uses its DNA helicase activity to
drive “branch migration,” the sliding of a branch connecting two recombining DNA duplexes. Similarly, Rvb1/Rvb2
has DNA helicase activity, and the human homolog has
been proposed to have a double hexamer structure, although the yeast protein appears to be a single heterohexamer. Because a DNA helicase tracks along a DNA duplex
as it unwinds the DNA, it is possible to imagine that INO80
uses its DNA helicase activity to nudge aside nucleosomes
as it tracks along the DNA, pushing the nucleosomes away
from a DSB.
Another chromatin remodeler, SWR1, is also recruited to
DSBs. Like INO80, SWR1 contains Rvb1/Rvb2, but it has
an additional intriguing activity: the ability to replace histone H2A with the H2A variant Htz1. Thus, SWR1 might
wea25324_ch20_636-676.indd Page 667
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
replace phospho-H2A with Htz1, which cannot be phosphorylated. In this way, SWR1 would return the histone
phosphorylation in nucleosomes near DSBs to the pre-broken
state once DSB repair is at least underway. In support of this
hypothesis, Jerry Workman and colleagues have shown that
Domino/p400, the Drosophila homolog of SWR1, replaces
phospho-H2A with unphosphorylated H2A in vitro.
Another chromatin remodeler recruited to doublestrand breaks, and other sites of DNA damage, is ALC1
(amplified in liver cancer). This protein contains a macrodomain, which binds specifically to poly(ADP-ribose)
(Chapter 13) that is formed at the sites of DNA damage by
poly(ADP-ribose) polymerase (PARP-1). This binding to
poly(ADP-ribose) also stimulates the remodeling activity of
ALC1. A histone H2A variant known as macroH2A1.1
also has a macrodomain, and is also attracted to
poly(ADP-ribose) at sites of damaged DNA. The substitution of macroH2A1.1 for ordinary H2A may facilitate the
remodeling catalyzed by ALC1, or other chromatin remodelers. Assuming that this remodeling aids in DNA repair, it appears that PARP-1 plays a role in DNA repair.
The fact that PARP-1 inhibitors are highly toxic to cells
defective in homologous recombination repair supports
this hypothesis. So does the fact that cells with excessive
DNA damage have hyperactive PARP-1.
Both of these findings have important clinical implications. Cancer cells, especially breast cancer cells with impaired homologous recombination repair due to faulty
BRCA1 and BRCA2 genes, are readily killed by PARP-1 inhibitors. And heart and brain cells can have their DNA
damaged by the oxidative stress of a cut-off blood supply
(ischemia) due to heart attack or stroke, respectively; the
sudden return of oxygen-rich blood (reperfusion) can result
in hyperactive PARP-1 in these cells. This is good for repairing the DNA, but making so much poly(ADP-ribose) depletes the ATP stores of the cells, which can rapidly kill
them. PARP-1 inhibitors could protect such cells.
SUMMARY Two protein kinases, Mec1 and Tel1,
are recruited to DSBs, where they phosphorylate
serine 129 of histone H2A in nearby nucleosomes.
This phosphorylation recruits the chromatin remodeler INO80 to the DSB, where it appears to use
its DNA helicase activity to push nucleosomes away
from the ends of the DSB, enabling formation of
single-stranded 39-DNA overhangs, which are essential for both nonhomologous end-joining and homologous recombination. Another chromatin remodeler
known as SWR1, which shares many components
with INO80, also appears at DSBs, and replaces
phospho-H2A with the H2A variant Htz1, which
cannot be phosphorylated. This returns the phosphorylation state of H2A on nucleosomes near DSBs
667
to normal. PARP-1 is recruited to DSBs and other
damaged DNA sites. It poly(ADP-ribosyl)ates itself
and other proteins at the damage site, which recruits
chromatin remodelers such as ALC1 and the histone
variant macroH2A1.1, both via their macrodomains.
Mismatch Repair
So far, we have been discussing repair of DNA damage caused
by mutagenic agents. What about DNA that simply has a
mismatch due to incorporation of the wrong base and failure
of the proofreading system? At first, it would seem tricky to
repair such a mistake because of the apparent difficulty in
determining which strand is the newly synthesized one that
has the mistake and which is the parental one that should be
left alone. At least in E. coli this is not a problem because the
parental strand has identification tags that distinguish it from
the progeny strand. These tags are methylated adenines, created by a methylating enzyme that recognizes the sequence
GATC and places a methyl group on the A. Because this
4-base sequence occurs approximately every 250 bp, one is
usually not far from a newly created mismatch.
Moreover, GATC is a palindrome, so the opposite
strand also reads GATC in its 59→39 direction. This means
that a newly synthesized strand across from a methylated
GATC is also destined to become methylated, but a little
time elapses before that can happen. The mismatch repair
system (Figure 20.35) takes advantage of this delay; it uses
the methylation on the parental strand as a signal to leave
that strand alone and correct the nearby mismatch in the
unmethylated progeny strand. This process must occur
fairly soon after the mismatch is created, or both strands
will be methylated and no distinction between them will be
possible. Eukaryotic mismatch repair is not as well understood as that in E. coli. The genes encoding the mismatch
recognition and excision enzymes (MutS and MutL) are
very well conserved, so the mechanisms that depend on
these enzymes are likely to be similar in eukaryotes and
bacteria. However, the gene encoding the strand recognition protein (MutH) is not found in eukaryotes, so eukaryotes appear not to use the methylation recognition trick. It
is not clear yet how eukaryotic cells distinguish the progeny strand from the parental strand at a mismatch.
SUMMARY The E. coli mismatch repair system
recognizes the parental strand by its methylated
adenines in GATC sequences. Then it corrects the
mismatch in the complementary (progeny) strand.
Eukaryotes use part of this repair system, but they
rely on a different, uncharacterized method for distinguishing the strands at a mismatch.
wea25324_ch20_636-676.indd Page 668
668
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
CH3
CH3
3′
5′
5′
3′
(a)
MutH, MutL, MutS, ATP
CH3
CH3
Nick
(b)
Exonuclease I, MutL, MutS,
helicase, ATP
CH3
CH3
(c)
DNA polymerase III holoenzyme,
SSB, DNA ligase
CH3
CH3
(d)
CH3
CH3
number of repeats in a given microsatellite may differ from
one normal individual to another, but it should be the same
in all tissues and remain constant throughout an individual’s
lifetime. The relationship between microsatellite instability
and mismatch repair is that the mismatch repair system is
responsible for recognizing and repairing the “bubble” created by the inaccurate insertion of too many or too few
copies of a short repeat because of “slippage” during DNA
replication. When this system breaks down, such slippage
goes unrepaired, leading to mutations in many genes whenever DNA replicates in preparation for cell division. This
kind of genetic instability presumably leads to cancer, by
mechanisms involving mutated genes (oncogenes and
tumor suppressor genes) that are responsible for control of
cell division.
SUMMARY The failure of human mismatch repair
leads to microsatellite instability, and ultimately to
cancer.
Methyltransferase
CH3
CH3
Figure 20.35 Mismatch repair in E. coli. (a) The products of the
mutH, L, and S genes along with ATP, recognize a base mismatch
(center), identify the newly synthesized strand by the absence of
methyl groups on GATC sequences, and introduce a nick into that
new strand, across from a methylated GATC and upstream of the
incorrect nucleotide. (b) Exonuclease I, along with MutL, MutS, DNA
helicase, and ATP, removes DNA downstream of the nick, including
the incorrect nucleotide. (c) DNA polymerase III holoenzyme, with
help from single-stranded binding protein (SSB), fills in the gap left
by the exonuclease, and DNA ligase seals the remaining nick.
(d) A methyltransferase methylates GATC sequences in the progeny
strand across from methylated GATC sequences in the parental
strand. Once this happens, mismatch repair nearby cannot occur
because the progeny and parental strands are indistinguishable.
Failure of Mismatch Repair in Humans
Failure of human mismatch repair has serious consequences, including cancer. One of the most common forms
of hereditary cancer is hereditary nonpolyposis colon cancer (HNPCC), also known as Lynch syndrome. Approximately 1 American in 200 is affected by this disease, and it
accounts for about 15% of all colon cancers. One of the
characteristics of HNPCC patients is microsatellite instability, which means that DNA microsatellites, tandem
repeats of 1–4-bp sequences, change in size (number of
repeats) during the patient’s lifetime. This is unusual; the
Coping with DNA Damage
Without Repairing It
The direct reversal and excision repair mechanisms described so far are all true repair processes. They eliminate
the defective DNA entirely. However, cells have other
means of coping with damage that do not remove it but
simply skirt around it. These are sometimes called repair
mechanisms, even though they really are not. A better term
might be damage bypass mechanism. These mechanisms
come into play when a cell has not performed true repair of
a lesion, but has either replicated its DNA or both replicated its DNA and divided before repairing the lesion. At
each of these steps (DNA replication and cell division), the
cell loses attractive options for dealing with DNA damage
and is increasingly faced with more dangerous options.
Recombination Repair Recombination repair is the most
important of these mechanisms. It is also sometimes called
postreplication repair because replication past a pyrimidine dimer can leave a problem: a gap opposite the dimer
that must be repaired. Excision repair will not work any
longer because there is no undamaged DNA opposite the
dimer—only a gap—so recombination repair is one of the
few alternatives left. Figure 20.36 shows how recombination repair works. First, the DNA is replicated. This creates
a problem for DNA with pyrimidine dimers because the
dimers stop the replication machinery. Nevertheless, after a
pause, replication continues, leaving a gap (a daughter
strand gap) across from the dimer. (A new primer is presumably required to restart DNA synthesis.) Next, recombination occurs between the gapped strand and its homolog
wea25324_ch20_636-676.indd Page 669
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
20.3 DNA Damage and Repair
669
RecA
(a)
UV light
Replication
Cleaved LexA
RecA coprotease
RNA polymerase
LexA
+
umuDC operon (repressed)
(b)
Active umuDC operon
Strand exchange
UmuC and UmuD
GCAT TCG A
GCAT TCG A
C GT
(c)
Recombination completed
+
(d)
Gap filled in
Figure 20.36 Recombination repair. We begin with DNA with a
pyrimidine dimer, represented by a V shape. (a) During replication,
the replication machinery skips over the region with the dimer,
leaving a gap; the complementary strand is replicated normally.
The two newly synthesized strands are shown in pink. (b) Strand
exchange between homologous strands occurs. (c) Recombination
is completed, filling in the gap opposite the pyrimidine dimer, but
leaving a gap in the other daughter duplex. The duplex with the
pyrimidine dimer has not been repaired, but it has replicated
successfully and may be repaired properly in the next generation.
(d) This last gap is easily filled, using the normal complementary
strand as the template.
on the other daughter DNA duplex. This recombination
depends on the recA gene product, which exchanges the
homologous DNA strands. We have encountered recA before in our discussion of the induction of a l prophage
during the SOS response (Chapter 8)—and we will discuss
it more fully in our consideration of recombination in
Chapter 22. The net effect of this recombination is to fill in
the gap across from the pyrimidine dimer and to create a
new gap in the other DNA duplex. However, because the
other duplex has no dimer, the gap can easily be filled in by
DNA polymerase and ligase. Note that the DNA damage
still exists, but the cell has at least managed to replicate its
DNA. Sooner or later, true DNA repair could presumably
occur.
Replication continues
C GTGAG CT
Replication
stalled
at dimer
Figure 20.37 Error-prone (SOS) bypass. Ultraviolet light activates
the RecA coprotease, which stimulates the LexA protein (purple) to
cleave itself, releasing it from the umuDC operon. This results in
synthesis of UmuC and UmuD proteins, which allow DNA synthesis
across from a pyrimidine dimer, even though mistakes (blue) will
frequently be made.
Error-Prone Bypass So-called error-prone bypass is another way of dealing with damage without really repairing
it. In E. coli, this pathway is induced as part of the SOS
response by DNA damage, including UV damage, and depends on the product of the recA gene. The chain of events
seems to be as follows (Figure 20.37): UV light or another
mutagenic treatment somehow activates the RecA coprotease activity. This coprotease has several targets. One we
have studied already is the l repressor, but its main target
is the product of the lexA gene. This product, LexA, is a
repressor for many genes, including repair genes; when it is
stimulated by RecA coprotease to cleave itself, all these
genes are induced.
Two of the newly induced genes are umuC and umuD,
which make up a single operon (umuDC). The product of
the umuD gene (UmuD) is clipped by a protease to form
UmuD9, which associates with the umuC product, UmuC,
to form a complex UmuD92C. This complex has DNA
polymerase activity, so it is also referred to as DNA pol V.
Pol V can cause error-prone bypass of DNA lesions in vitro
on its own, but it is activated by RecA-ATP. This RecA-ATP
comes from the 39-end of a nucleoprotein filament of RecA
and DNA (RecA*), which may have assembled at a site
remote from the site of error-prone bypass. Such bypass
involves replication of DNA across from the DNA lesion
even though correct “reading” of the lesion itself is impossible. This avoids leaving a gap, but it frequently puts the
wea25324_ch20_636-676.indd Page 670
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
wrong bases into the new DNA strand (hence the name
“error-prone”). When the DNA replicates again, these errors will be perpetuated. Error-prone bypass and other,
more error-free bypass mechanisms found in eukaryotes,
are also called translesion synthesis (TLS).
DNA polymerase V can efficiently bypass the three
most common types of DNA lesion: pyrimidine dimers,
related lesions also caused by UV light—(6-4) photoproducts, and abasic (AP) sites. However, this enzyme performs this translesion synthesis with varying degrees of
fidelity. In 2000, Myron Goodman and colleagues measured the incorporation of A and G across from the two T’s
of a thymine dimer, or of a (6-4) photoproduct, and
across from an AP site. Opposite a pyrimidine dimer,
DNA polymerase V tended to incorporate A’s in both
positions, which is fine for thymine dimers, but not if the
dimer contains cytosines. Opposite a (6-4) photoproduct
containing two thymines, DNA polymerase V tended to
incorporate a G in the first position and an A in the second—
obviously not very faithful replication. Opposite an AP
site, DNA polymerase V incorporated about two-thirds A
and about one-third G. All of these ratios, and the fact
that pyrimidines were not detectably incorporated, agree
with in vivo observations, suggesting that DNA polymerase V is indeed the enzyme that performs translesion
synthesis in vivo.
If the umu genes are really responsible for error-prone
bypass, we might expect mutations in one of these genes to
make E. coli cells less susceptible to mutation. These mutant cells would be just as prone to DNA damage, but the
damage would not be as readily converted into mutations.
In 1981, Graham Walker and colleagues verified this expectation by creating a null allele of the umuC gene (a version of the gene with no activity), and showing that bacteria
harboring this gene were essentially unmutable. In fact,
“umu” stands for “unmutable.”
These workers established an E. coli strain carrying the
umuC mutant, and a his2 mutation that is ordinarily revertable by UV radiation. Then they challenged this bacterial strain with UV radiation and counted the his1
revertants. The more revertants, the more mutation was allowed because a reversion is just a back-mutation. Figure 20.38
shows the results. A reasonable number of revertants
occurred in wild-type cells (about 200 at the highest UV
dose). By stark contrast, in umuC2 cells almost no revertants occur. Furthermore, addition of a plasmid bearing
the muc gene, which can suppress the unmutable phenotype of umuC2 cells, caused a dramatic increase in the
number of revertants (about 500, even at a relatively low
UV dose).
The null allele in this experiment was created by insertion of the lac structural genes, without the lac promoter,
into the umuC gene, then screening for lac1 cells. The cells
were originally lac2, so the appearance of lac1 cells
indicated that the lac genes had inserted downstream of a
600
Revertants/108 survivors
670
12/17/10
umuC – + muc +
400
200
umuC +
umuC –
1
3
2
UV dose (J/m2)
4
5
Figure 20.38 An umuC strain of E. coli is unmutable. Walker and
colleagues tested three his2 strains of bacteria for the ability to generate
his1 revertants after UV irradiation. The strains were: wild-type with
respect to umuC (blue), a umuC2 strain (red), and a umuC2 strain
supplemented with a plasmid containing the muc gene (green). (Source:
Adapted from Bagg, A., C.J. Kenyon, and G.C. Walker, Inducibility of a gene product
required for UV and chemical mutagenesis in Escherichia. coli. Proceedings of the
National Academy of Sciences USA 78:5750, 1981.)
promoter—the umuDC promoter, in this case. The fact
that the lac genes fell under control of the umuDC promoter
allowed Walker and colleagues to test the inducibility of
this promoter by UV radiation, simply by measuring
b-galactosidase activity. Figure 20.39 shows that the promoter was indeed inducible by UV radiation at a dose of
10 J/m2 (blue curve). But the promoter was not inducible in
lexA mutant or recA2 cells (green and red curves). The
lexA mutant cells used in this experiment encoded a LexA
protein that was not cleavable and therefore could not be
removed from the umuDC operator.
Wild-type E. coli cells can tolerate as many as 50 pyrimidine dimers in their genome without ill effect because
of their active repair mechanisms. Bacteria lacking one of
the uvr genes cannot carry out excision repair, so their susceptibility to UV damage is greater. However, they are still
somewhat resistant to DNA damage. On the other hand,
double mutants in uvr and recA can perform neither excision repair nor recombination repair, and they are very
sensitive to UV damage, perhaps because they have to rely
on error-prone bypass. Under these conditions, only one to
two pyrimidine dimers per genome is a lethal dose.
Obviously, if bacterial cells had evolved without the
error-prone bypass, they would be subject to many fewer
mutations. If that is the case, then why have they retained
this mutation-causing mechanism? It is likely that the
error-prone bypass system does more good than harm by allowing an organism to replicate its damaged genome even
at the risk of mutation. This is especially obvious if the
wea25324_ch20_636-676.indd Page 671
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
β-Galactosidase
(units/A600 unit)
20.3 DNA Damage and Repair
100
75
50
25
0
0
1
2
Time (h)
3
Figure 20.39 The umuDC promoter is UV-inducible. Walker and
colleagues irradiated cells with the lac genes under control of the
umuDC promoter with a UV dose of 10 J/m2. They performed the
irradiation at 1 h, as indicated by the arrow. Then they measured the
accumulation of b-galactosidase activity (blue) per OD600 unit (an index
of turbidity and therefore of cell density). They also performed the same
experiment in lexA mutant (green) and recA– cells (red). The lexA mutant
was an “uninducible” one encoding a LexA protein that cannot be
cleaved and therefore cannot be removed from the umuDC operator.
(Source: Adapted from Bagg, A., C.J. Kenyon, and G.C. Walker, Inducibility of a
gene product required for UV and chemical mutagenesis in Escherichia coli.
Proceedings of the National Academy of Sciences USA 78:5751, 1981.)
price for failure to replicate is death, as would be the case
after a cell replicates its damaged DNA and then divides
without repairing the damage. This chain of events would
produce one daughter cell with a DNA gap across from a
lesion. By this time, excision repair and even recombination repair are no longer possible. So the last resort is errorprone bypass to stave off cell death.
It is also true that a certain level of mutation is good
for a species because it allows the genomes of a group of
organisms to diverge so they do not all have equal susceptibility to disease and other insults. That way, when a
new challenge arises, some of the members of a population have evolved resistance and can survive to perpetuate the species.
SUMMARY Cells can employ nonrepair methods to
circumvent DNA damage. One of these is recombination repair, in which the gapped DNA strand
across from a damaged strand recombines with a
normal strand in the other daughter DNA duplex
after replication. This solves the gap problem but
leaves the original damage unrepaired. Another
mechanism to deal with DNA damage, at least in
E. coli, is to induce the SOS response, which causes
the DNA to replicate even though the damaged
region cannot be read correctly. This results in errors
in the newly made DNA, so the process is called
error-prone bypass.
671
Error-Prone and Error-Free Bypass in Humans All of the
DNA repair processes are well conserved throughout all
kingdoms of life, probably because DNA damage has been
part of life from the very beginning, so damage repair had
to evolve early, before the three kingdoms diverged. Errorprone bypass is no exception: Human cells have systems
similar to those in prokaryotes to deal with lesions like
pyrimidine dimers. These bypass systems depend on specialized DNA polymerases, including DNA polymerases z
(zeta), h (eta), u (theta), i (iota), and k (kappa). These specialized polymerases take over from polymerases d and ε,
which synthesize the lagging and leading strands, respectively, but stall at uninstructive DNA lesions like pyrimidine dimers.
Some of these enzymes insert bases at random to get
past the lesion, which is obviously an error-prone strategy. But some of them have specificities that minimize
errors and are therefore relatively error-free. For example, DNA polymerase h automatically inserts two dAMPs
into the DNA strand across from a pyrimidine dimmer.
Thus, even though the bases in the dimer cannot basepair, this system is able to make the correct choice if both
bases in the dimer are thymines—which is often the case.
DNA polymerase h can also bypass adjacent guanines
(Pt-GGs) that have been cross-linked via platinum by the
anti-cancer drug cisplatin. It does a good job of replicating the 39-dG, usually inserting a dC in the opposite
strand, but it randomly inserts either dC or dA opposite
the 59-dG.
In 1999, Fumio Hanaoka and colleagues discovered
that the defective gene in patients with the variant form of
XP (XP-V) is the gene that codes for DNA polymerase h.
Thus, these patients cannot carry out the comparatively
error-free bypass of pyrimidine dimers catalyzed by DNA
polymerase h and must therefore rely on the error-prone
bypass catalyzed by other specialized DNA polymerases,
including DNA polymerase z. This error-prone system introduces mutations during replication of pyrimidine dimers
not removed by the excision repair system. However, because these patients have normal excision repair, few dimers are left for the error-prone system to deal with. This
argument accounts for the relatively low sensitivity of
XP-V cells to ultraviolet radiation.
Polymerase h cannot carry out error-free bypass by itself. After it inserts two A’s across from a pyrimidine dimer,
the 39-end of the newly synthesized strand is not basepaired to a T because the T’s in the template strand are
locked up in the pyrimidine dimer. Without a base-paired
nucleotide to add to, the replicative DNA polymerases
(ε and d) cannot resume DNA synthesis. Thus, another
polymerase, perhaps polymerase z, must do the job.
Why doesn’t polymerase h simply continue synthesizing enough DNA for one of the replicative polymerases to
get started again? The answer is that this would be a very
error-prone process. Although the term “error-free” for
wea25324_ch20_636-676.indd Page 672
672
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
(a)
α
η
(b)
nt
–30
α
η
α
η
DNA polymerase h is justified in terms of its ability to
deal with thymine dimers, this enzyme is remarkably
error-prone when replicating ordinary DNA. When Hanaoka,
Thomas Kunkel, and colleagues tested the fidelity of this
enzyme in vitro, using a double-stranded DNA with a gap
in it, they found that DNA polymerase h had a lower
fidelity than any other template-dependent DNA polymerase ever studied until that time: one mistake per
18–380 nt incorporated. By contrast; DNA polymerase z
is about 20 times more accurate. Thus, it is a good thing
that cells normally have the NER system. Without it,
DNA polymerase h would be a very poor backstop for
dealing with anything but thymine dimers—as typical XP
patients can attest.
DNA polymerase h is specific for translesion synthesis
at certain kinds of DNA damage. This enzyme can perform
TLS at a pyrimidine dimer, but not at a (6-4) photoproduct.
DNA polymerase h can also bypass an abasic (AP) site.
Hanaoka and colleagues performed an assay to measure
TLS in vitro at each of these kinds of DNA damage, using
either polymerase a or polymerase h. They used templates
that contained one damaged strand and one 32P-labeled
primer strand that had its 39-end just upstream of the damage. Then they added nucleotides to allow TLS and electrophoresed the products.
Figure 20.40 depicts the results. Panel (a) shows that
polymerases a and h could both extend the primer on an
undamaged template, but polymerase a was ineffective in
extending the primer past any of the DNA lesions. This
failure of polymerase a is not surprising because it is designed for accurate copying of normal DNA to make primers, not for dealing with the noninformative DNA in these
lesions. Panels (b–d) show that polymerase h could extend
the primer past a cyclic pyrimidine dimer (CPD) and an AP
site, but not past a (6-4) photoproduct.
1 2 3
α
η
4567
nt
–30
AP
X
Figure 20.40 Activities of DNA polymerases a and h on
undamaged and damaged templates. Hanaoka and colleagues
prepared double-stranded DNAs containing on the template strand:
(a) no damage; (b) a cyclobutane pyrimidine dimer (CPD); (c) a (6-4)
photoproduct [(6-4)PP]; or (d) an AP site. The nontemplate strand of
these DNAs was a 32P-labeled primer that was poised to be extended
through the damage (or normal pair of thymines) on the template
TT
1 23 4567
(d)
nt
–30
(6-4)PP
CPD
TT
TT
1 23 4567
(c)
nt
–30
1 2 3
4 567
strand. The DNAs are illustrated with cartoons adjacent to each panel.
The workers added increasing amounts of either DNA polymerase
a or h, along with nucleotides, and electrophoresed the products on
polyacrylamide gels. If translesion synthesis was successful, the
primer was extended to the full length of the template strand, 30 nt. If
not, synthesis stalled at the lesion. (Source: From Masutani et al., Cold
spring Harbor Symposia p. 76. © 2000.)
SUMMARY Humans have a relatively error-free by-
pass system that inserts dAMPs across from a pyrimidine dimer, thus replicating thymine dimers (but
not dimers involving cytosines) correctly. This system uses DNA polymerase h plus another enzyme
to replicate a few bases beyond the lesion. When the
gene for DNA polymerase h is defective, DNA polymerase z and perhaps other DNA polymerases take
over. But these polymerases insert random nucleotides across from a pyrimidine dimer, so they are
error-prone. These errors in correcting UV damage
lead to a variant form of XP known as XP-V. DNA
polymerase h is active on templates with thymine
dimers and AP sites, but not on (6-4) photoproducts. This polymersase is not really error-free. With
a gapped template it is one of the least accurate
template-dependent polymerases known.
S U M M A RY
Several principles apply to all (or most) DNA replication:
(1) Double-stranded DNA replicates in a semiconservative
manner. When the parental strands separate, each serves
as the template for making a new, complementary strand.
(2) DNA replication in E. coli (and in other organisms) is
at least semidiscontinuous. One strand is replicated in the
direction of the movement of the replicating fork; This
strand is commonly thought to replicate continuously,
though there is evidence that it replicates discontinuously.
the other is replicated discontinuously, forming 1–2 kb
Okazaki fragments in the opposite direction. This allows
wea25324_ch20_636-676.indd Page 673
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Summary
both strands to be replicated in the 59→39 direction.
(3) Initiation of DNA replication requires a primer. Okazaki
fragments in E. coli are initiated with RNA primers
10–12 nt long. (4) Most eukaryotic and bacterial DNAs
replicate bidirectionally. ColE1 is an example of a DNA
that replicates unidirectionally.
Circular DNAs can replicate by a rolling circle
mechanism. One strand of a double-stranded DNA is
nicked and the 39-end is extended, using the intact DNA
strand as template. This displaces the 59-end. In phage l,
the displaced strand serves as the template for
discontinuous, lagging strand synthesis.
Pol I is a versatile enzyme with three distinct activities:
DNA polymerase; 39→59 exonuclease; and 59→39
exonuclease. The first two activities are found on a large
domain of the enzyme, and the last is on a separate, small
domain. The large domain (the Klenow fragment) can be
separated from the small by mild protease treatment,
yielding two protein fragments with all three activities
intact. The structure of the Klenow fragment shows a
wide cleft for binding to DNA. This polymerase active site
is remote from the 39→59 exonuclease active site on the
Klenow fragment.
Of the three DNA polymerases in E. coli cells, pol I,
pol II, and pol III, only pol III is required for DNA
replication. Thus, this polymerase is the enzyme that
replicates the bacterial DNA. The pol III core is composed
of three subunits, a, ε, and u. The a-subunit has the DNA
polymerase activity. The ε-subunit has the 39→59 activity
that carries out proofreading.
Faithful DNA replication is essential to life. To help
provide this fidelity, the E. coli DNA replication
machinery has a built-in proofreading system that requires
priming. Only a base-paired nucleotide can serve as a
primer for the pol III holenzyme. Therefore, if the wrong
nucleotide is incorporated by accident, replication stalls
until the 39→59 exonuclease of the pol III holoenzyme
removes it. The fact that the primers are made of RNA
may help mark them for degradation.
Mammalian cells contain five different DNA
polymerases. Polymerases ε, d, and a appear to participate
in replicating both DNA strands. Polymerase a makes the
primers for both strands, polymerase ε elongates the
leading strand, and polymerase d elongates the lagging
strand. Polymerase b seems to function in DNA repair.
Polymerase g probably replicates mitochondrial DNA.
The helicase that unwinds double-stranded DNA at
the replicating fork is encoded by the E. coli dnaB gene.
The bacterial single-strand DNA-binding proteins bind
much more strongly to single-stranded than to doublestranded DNA. They aid helicase action by binding tightly
and cooperatively to newly formed single-stranded DNA
and keeping it from annealing with its partner. By coating
the single-stranded DNA, SSBs also protect it from
degradation. They also stimulate their homologous DNA
673
polymerases. These activities make SSBs essential for
bacterial DNA replication.
As a helicase unwinds the two parental strands of a
closed circular DNA, it introduces a compensating
positive supercoiling force into the DNA. The stress of
this force must be overcome or it will resist progression
of the replicating fork. A name given to this stress-release
mechanism is the swivel. DNA gyrase, a bacterial
topoisomerase, is the leading candidate for this role in
E. coli.
Alkylating agents like ethylmethane sulfonate add
bulky alkyl groups to bases, either disrupting base pairing
directly or causing loss of bases, either of which can lead
to faulty DNA replication or repair.
Different kinds of radiation cause different kinds of
damage. Ultraviolet rays have comparatively low energy,
and they cause a moderate type of damage: pyrimidine
dimers. Gamma and x-rays are much more energetic. They
ionize the molecules around DNA and form highly
reactive free radicals that can attack DNA, altering bases
or breaking strands.
Ultraviolet radiation damage to DNA (pyrimidine
dimers) can be directly repaired by a DNA photolyase
that uses energy from visible light to break the bonds
holding the two pyrimidines together. O6 alkylations on
guanine residues can be directly reversed by the suicide
enzyme O6-methylguanine methyltransferase, which
accepts the alkyl group onto one of its amino acids.
Base excision repair (BER) typically acts on subtle
base damage. This process begins with a DNA
glycosylase, which extrudes a base in a damaged base
pair, then clips out the damaged base, leaving an apurinic
or apyrimidinic site that attracts the DNA repair enzymes
that remove the remaining deoxyribose phosphate and
replace it with a normal nucleotide. In bacteria, DNA
polymerase I is the enzyme that fills in the missing
nucleotide in BER, in eukaryotes, DNA polymerase b plays
this role. However, this enzyme makes mistakes, and has
no proofreading activity, so APE1 carries out the necessary
proofreading. Repair of 8-oxoguanine sites in DNA is a
special case of BER that can happen in two ways. Since
oxoG mispairs with A, the A can be removed after DNA
replication by a specialized adenine DNA glycosylase.
However, if replication has not yet occurred, the oxoG will
still be paired with C, and the oxoG can be removed by
another DNA glycosylase, the oxoG repair enzyme.
Nucleotide excision repair (NER) generally deals with
drastic, helix-distorting base changes. In bacterial NER,
the damaged DNA is clipped out directly by cutting on
both sides of the lesion with an endonuclease to remove
the damaged DNA as part of an oligonucleotide. DNA
polymerase I fills in the gap and DNA ligase seals the
final nick.
Eukaryotic NER follows two pathways. In global
genome NER (GG-NER), a complex composed of XPC
wea25324_ch20_636-676.indd Page 674
674
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
and hHR23B initiates repair by binding to a lesion
anywhere in the genome and causing a limited amount of
DNA melting. This protein apparently recruits XPA and
RPA. TFIIH then joins the complex, and two of its
subunits (XPB and XPD) use their DNA helicase activities
to expand the melted region. RPA binds two excinucleases
(XPF and XPG) and positions them for cleavage of the
DNA strand on either side of the lesion. This releases the
damage on a fragment between 24 and 32 nt long.
Transcription-coupled NER (TC-NER) is very similar to
global genome NER, except that RNA polymerase plays
the role of XPC in damage sensing and initial DNA
melting. In either kind of NER, DNA polymerase ε or d
fills in the gap left by the removal of the damaged
fragment, and DNA ligase seals the DNA.
Double-strand DNA breaks can be repaired by
homologous recombination or by nonhomologous end
joining. The latter process requires Ku and DNA–PKcs,
which bind together at the DNA ends, constituting active
DNA–PK complexes that allow the ends to find regions
of microhomology with each other. Once the regions of
microhomology line up, the two DNA–PK complexes
phosphorylate each other. This phosphorylation activates
the catalytic subunit (DNA–PKcs) to dissociate, and it
also activates the DNA helicase activity of Ku to unwind
the DNA ends so the microhomology regions can basepair. Finally, extra flaps of DNA are removed, gaps are
filled, and the DNA ends are ligated permanently
together.
Chromatin remodeling is required for both
nonhomologous end-joining and homologous
recombination. In yeast, two protein kinases, Mec1 and
Tel1, are recruited to DSBs, where they phosphorylate
serine 129 of histone H2A in nearby nucleosomes. This
phosphorylation recruits the chromatin remodeler
INO80 to the DSB, where it appears to use its DNA
helicase activity to push nucleosomes away from the
ends of the DSB, enabling formation of single-stranded
39-DNA overhangs, which are essential for both NHEJ
and homologous recombination. Another chromatin
remodeler known as SWR1, which shares many
components with INO80, also appears at DSBs, and
replaces phospho-H2A with the H2A variant Htz1,
which cannot be phosphorylated. This returns the
phosphorylation state of H2A on nucleosomes near
DSBs to normal.
Errors in DNA replication leave mismatches that can
be detected and repaired. The E. coli mismatch repair
system recognizes the parental strand by its methylated
adenines in GATC sequences. Then it corrects the
mismatch in the complementary (progeny) strand. The
failure of human mismatch repair leads to microsatellite
instability, and ultimately to cancer.
Cells can employ nonrepair methods to circumvent
DNA damage. One of these is recombination repair, in
which the gapped DNA strand across from a damaged
strand recombines with a normal strand in the other
daughter DNA duplex after replication. This solves the
gap problem but leaves the original damage unrepaired.
Another mechanism to deal with DNA damage, at least
in E. coli, is to induce the SOS response, which causes
the DNA to replicate even though the damaged region
cannot be read correctly. This results in errors in the
newly made DNA, so the process is called error-prone
bypass.
Humans have a relatively error-free bypass system
that inserts dAMPs across from a pryimidine dimer,
thus replicating thymine dimers (but not dimers
involving cytosines) correctly. This system uses DNA
polymerase h plus another enzyme to replicate a few
bases beyond the lesion. When the gene for DNA polymerase
h is defective, DNA polymerase z, and perhaps other
DNA polmerases, take over. But these polymerases
insert random nucleotides across from a pryimidine
dimer, so they are error-prone. These errors in
correcting UV damage lead to a variant form of XP
known as XP-V.
REVIEW QUESTIONS
1. Compare and contrast the conservative, semiconservative,
and dispersive mechanisms of DNA replication.
2. Describe and give the results of an experiment that shows
that DNA replication is semiconservative.
3. Compare and contrast the continuous, discontinuous, and
semidiscontinuous modes of DNA replication.
4. Describe and give the results of an experiment that shows
that DNA replication is at least semidiscontinuous.
5. What is the evidence for fully discontinuous DNA
replication in E. coli cells?
6. Describe and give the results of an experiment that
measures the size of the primers on Okazaki fragments.
7. Present electron microscopic evidence that DNA replication
of the B. subtilis chromosome is bidirectional, whereas
replication of the colE1 plasmid is unidirectional.
8. Diagram the rolling circle replication mechanism used by
the l phage.
9. Diagram the proofreading process used by E. coli DNA
polymerases.
10. What activities are contained in E. coli DNA polymerase I?
What is the role of each in DNA replication?
11. How does the Klenow fragment differ from the intact
E. coli DNA polymerase I? Which enzyme would you use in
nick translation? DNA end-filling? Why?
12. Of the three DNA polymerases in E. coli, which is essential
for DNA replication? Present evidence.
13. Which pol III core subunit has the DNA polymerase
activity? How do we know?
wea25324_ch20_636-676.indd Page 675
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Suggested Readings
14. Which pol III core subunit has the proofreading activity?
How do we know?
15. Explain how the necessity for proofreading rationalizes the
existence of priming in DNA replication.
16. List the eukaryotic DNA polymerases and their roles.
Outline evidence for these roles.
17. Compare and contrast the activity of a helicase with that of
a topoisomerase in the context of DNA replication.
18. What roles do SSBs play in DNA replication?
19. Explain why nicking one strand of a supercoiled DNA
removes the supercoiling.
20. How do we know that DNA gyrase forms a covalent bond
between an enzyme tyrosine and DNA? What is the
advantage of forming this bond?
21. Present a model, based on the structure of yeast DNA
topoisomerase II, for the DNA segment-passing step.
22. Compare and contrast the DNA damage done by UV rays
and x-rays or gamma rays.
23. What two enzymes catalyze direct reversal of DNA
damage? Diagram the mechanisms they use.
24. Compare and contrast base excision repair and nucleotide
excision repair. Diagram both processes. For what types of
damage is each primarily responsible?
675
A N A LY T I C A L Q U E S T I O N S
1. Why is it improbable that we will ever observe continuous
DNA replication of both strands in nature?
2. You are studying a protein that you suspect has DNA helicase activity. Describe how you would assay the protein for
this activity and show sample positive results.
3. You are studying a protein that you suspect has DNA
topoisomerase activity. Describe how you would assay the
protein for the activity and show sample positive results.
4. Explain the difference between DNA damage and mutation.
How do mutations in E. coli DNA polymerase V illustrate
this difference?
5. Recently, as a post-doc in a highly reputable laboratory, you
designed a new single-celled organism only capable of three
DNA repair mechanisms. You have been asked to present
your research at a prestigious Molecular Biology conference
Describe how you will support your reason for choosing
the three repair mechanisms and discuss if there are overlaps or gaps between the chosen mechanisms. Additionally,
explain the types of mutations your cell can overcome and
the types of damage that may potentially destroy your new
organism. You may assume that your organism already has
a homologous recombination system.
25. What enzyme performs proofreading in human base
excision repair? Outline the evidence supporting your
answer.
26. Briefly describe the crystal structures of complexes between
the human oxoG repair enzyme (hOGG1) and an oxoG–C
pair, or a normal G–C pair. How do these structures explain
why oxoG is removed, while ordinary G is not.
27. How does transcription-coupled NER differ from global
genome NER?
28. Outline the nonhomologous end-joining mechanism
mammals use to repair double-stand DNA breaks. Show how
this process can lead to loss of nucleotides at the repair site.
29. What DNA repair system is missing in most cases of
xeroderma pigmentosum? Why does that make XP patients
so sensitive to UV light? What is the primary backup system
for these patients?
30. What DNA repair system is missing in XP-V patients? Why
is the incidence of skin cancer lower in these people than in
typical XP patients? What is the backup system for lesions
missed by the NER system in XP-V patients?
31. Why is chromatin remodeling needed for double-strand
break repair in eukaryotes?
32. Diagram the mismatch repair mechanism in E. coli.
33. Diagram the recombination repair mechanism in E. coli.
34. Diagram the error-prone bypass system in E. coli.
35. Explain why recombination repair and error-prone bypass
are not real repair systems.
36. Present evidence that shows that DNA polymerase h can
bypass a thymine dimer and an AP site but not a (6-4)
photoproduct, and that DNA polymerase a cannot bypass
any of these lesions.
SUGGESTED READINGS
General References and Reviews
Cairns, B.R. 2004. Around the world of DNA damage INO80
days. Cell 119:733–34.
Chu, G. 1997. Double strand break repair. Journal of Biological
Chemistry. 272:24097–100.
Citterio, E., W. Vermeulen, and J.H.J. Hoeijmakers. 2000.
Transcriptional healing. Cell 101:447–50.
David, S.S. 2005. DNA search and rescue. Nature 434:569–70.
de Latt, W.L., N.G.J. Jaspers, and J.H.J. Hoeijmakers. 1999.
Molecular mechanism of nucleotide excision repair. Genes
and Development 13:768–85.
Friedberg, E.C., R. Wagner, and M. Radman. 2002. Specialized
DNA polymerases, cellular survival, and the genesis of
mutations. Science 296:1627–30.
Herendeen, D.R. and T.J. Kelly. 1996. DNA polymerase III:
Running rings around the fork. Cell 84:5–8.
Jiricny, J. 2002. An APE that proofreads. Nature 415:593–94.
Joyce, C.M. and T.A. Steitz. 1987. DNA polymerase I: From
crystal structure to function via genetics. Trends in
Biochemical Sciences 12:288–92.
Kornberg, A. and T. Baker. 1992. DNA Replication. New York:
W.H. Freeman and Company.
Lindahl, T. 2004. Molecular biology: Ensuring error-free DNA
repair. Nature 427:598.
Lindahl, T. and R.D. Wood. 1999. Quality control by DNA
repair. Science 286:1897–1905.
Maxwell, A. 1996. Protein gates in DNA topoisomerase II.
Nature Structural Biology. 3:109–12.
wea25324_ch20_636-676.indd Page 676
676
12/17/10
1:40 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 20 / DNA Replication, Damage, and Repair
Sharma, A. and A. Mondragón. 1995. DNA topoisomerases.
Current Opinion in Structural Biology 5:39–47.
Wood, R.D. 1997. Nucleotide excision repair in mammalian
cells. Journal of Biological Chemistry 272:23465–68.
Wood, R.D. 1999. Variants on a theme. Nature 399:639–70.
Research Articles
Bagg, A., C.J. Kenyon, and G.C. Walker. 1981. Inducibility of a
gene product required for UV and chemical mutagenesis in
Escherichia coli. Proceedings of the National Academy of
Sciences USA 78:5749–53.
Banerjee, A., W. Yang, M. Karplus, and G.L. Verdine. 2005.
Structure of a repair enzyme interrogating undamaged DNA
elucidates recognition of damaged DNA. Nature 434:612–18.
Berger, J.M., S.J. Gamblin, S.C. Harrison, and J.C. Wang. 1996.
Structure and mechanism of DNA topoisomerase II. Nature
379:225–32.
Cairns, J. 1963. The chromosome of Escherichia coli. Cold
Spring Harbor Symposia on Quantitative Biology 28:43–46.
Chou, K.-M. and Y.-C. Cheng. 2002. An exonucleolytic activity
of human apurinic/apyrimidinic endonuclease on 39 mispaired
DNA. Nature 415:655–59.
Curtis, M.J. and B. Alberts. 1976. Studies on the structure of
intracellular bacteriophage T4 DNA. Journal of Molecular
Biology 102:793–816.
Drapkin, R., J.T. Reardon, A. Ansari, J.-C. Huang, L. Zawel,
K. Ahn, A. Sancar, and D. Reinberg. 1994. Dual role of TFIIH
in DNA excision repair and in transcription by RNA
polymerase II. Nature 368:769–72.
Eom, S.H., T. Wang, and T.A. Steitz. 1996. Structure of Taq
polymerase with DNA at the active site. Nature 382:
278–281.
Gefter, M.L., Y. Hirota, T. Kornberg, J.A. Wechster, and
C. Barnoux. 1971. Analysis of DNA polymerases II and III
in mutants of Escherichia coli thermosensitive for DNA
synthesis. Proceedings of the National Academy of Sciences
USA 68:3150–53.
Gellert, M., K. Mizuuchi, M.H. O’Dea, and H.A. Nash. 1976.
DNA gyrase: An enzyme that introduces superhelical turns
into DNA. Proceedings of the National Academy of Sciences
USA 73:3872–76.
Gyurasits, E.B. and R.J. Wake. 1973. Bidirectional chromosome
replication in Bacillus subtilis. Journal of Molecular Biology
73:55–63.
Hirota, G.H., A. Ryter, and F. Jacob. 1968. Thermosensitive
mutants in E. coli affected in the processes of DNA synthesis
and cellular division. Cold Spring Harbor Symposia on
Quantitative Biology 33:677–93.
Huberman, J.A., A. Kornberg, and B.M. Alberts. 1971.
Stimulation of T4 bacteriophage DNA polymerase by the
protein product of T4 gene 32. Journal of Molecular Biology
62:39–52.
Kitani, T., K.-Y. Yoda, T. Ogawa, and T. Okazaki. 1985. Evidence
that discontinuous DNA replication in Escherichia coli is
primed by approximately 10 to 12 residues of RNA starting
with a purine. Journal of Molecular Biology 184:45–52.
LeBowitz, J.H. and R. McMacken. 1986. The Escherichia coli
dnaB replication protein is a DNA helicase. Journal of
Biological Chemistry 261:4738–48.
Maki, H. and A. Kornberg. 1985. The polymerase subunit of
DNA polymerase III of Escherichia coli. Journal of Biological
Chemistry 260:12987–92.
Masutani, C., R. Kusumoto, A. Yamada, N. Dohmae, M. Yokoi,
M. Yuasa, M. Araki, S. Iwai, K. Takio, and F. Hanaoka. 1999.
The XPV (xeroderma pigmentosum variant) gene encodes
human DNA polymerase h. Nature 399: 700–04.
Masutani, C., R. Kusumoto, A. Yamada, M. Yuasa, M. Araki,
T. Nogimori, M. Yokoi, T. Eki, S. Iwai, and F. Hanaoka. 2000.
Xeroderma pigmentosum variant: From a human genetic
disorder to a novel DNA polymerase. Cold Spring Harbor
Symposia on Quantitative Biology. 65:71–80.
Matsuda, T., K. Bebenek, C. Masutani, F. Hanaoka, and T.A.
Kunkel. 2000. Low fidelity DNA synthesis by human DNA
polymerase-h. Nature 404:1011–13.
Meselson, M. and F. Stahl. 1958. The replication of DNA in
Escherichia coli. Proceedings of the National Academy of
Sciences USA 44:671–82.
Okazaki, R., T. Okazaki, K. Sakabe, K. Sugimoto, R. Kainuma,
A. Sugino, and N. Iwatsuki. 1968. In vivo mechanism of
DNA chain growth. Cold Spring Harbor Symposia on
Quantitative Biology 33:129–43.
Scheuermann, R.H. and H. Echols. 1984. A separate editing
exonuclease for DNA replication: The ε subunit of
Escherichia coli DNA polymerase III holoenzyme.
Proceedings of the National Academy of Sciences USA
81:7747–57.
Sugasawa, K., J.M.Y. Ng, C. Masutani, S. Iwai, P.J. van der
Spek, A.P.M. Eker, F. Hanaoka, D. Bootsma, and J.H.J.
Hoeijmakers. 1998. Xeroderma pigmentosum group C
protein complex is the initiator of global genome nucleotide
excision repair. Molecular Cell 2:223–32.
Tse, Y.-C., K. Kirkegaard, and J.C. Wang. 1980. Covalent bonds
between protein and DNA. Journal of Biological Chemistry
255:5560–65.
Wakasugi, M. and A. Sancar. 1999. Order of assembly of human
DNA repair excision nuclease. Journal of Biological
Chemistry 274:18759–68.