You are on page 1of 12

Trends in

OPEN ACCESS Genetics


Review

Underappreciated Roles of DNA Polymerase δ


in Replication Stress Survival
Jeannette Fuchs,1 Anais Cheblal,1,2 and Susan M. Gasser1,2,3,*,@

Recent structural analysis of Fe-S centers in replication proteins and insights into Highlights
the structure and function of DNA polymerase δ (DNA Pol δ) subunits have shed DNA polymerase δ is key to surviving
light on the key role played by this polymerase at replication forks under stress. replication stress, and is found in almost
every oncogene-transformed mamma-
The sequencing of cancer genomes reveals multiple point mutations that com-
lian cell line.
promise the activity of POLD1, the DNA Pol δ catalytic subunit, whereas the
loci encoding the accessory subunits POLD2 and POLD3 are amplified in a Cancer-linked point mutations are found
very high proportion of human tumors. Consistently, DNA Pol δ is key for the sur- in POLD1, whereas POLD2 and POLD3
are found amplified in the vast majority of
vival of replication stress and is involved in multiple long-patch repair pathways. cancers.
Synthetic lethality arises from compromising the function and availability of the
noncatalytic subunits of DNA Pol δ under conditions of replication stress, opening Hydroxyurea destabilizes Fe-S centers,
the door to novel therapies. which also compromises the stability of
the DNA polymerase δ complex.

Introduction Inhibitors of checkpoint kinases and re-


The replication checkpoint is a highly conserved surveillance mechanism for defects or stalling of the agents that compromise DNA polymer-
ase δ activity may act synergistically to
replication fork. It plays a crucial role in the early stages of human tumorigenesis. Interestingly, pre-
arrest growth in transformed cells.
cancerous lesions show a constitutive activation of the DNA damage response due to replication
fork dysfunction even in the absence of exogenous damage [1,2]. The activation of this replication
checkpoint provides an important barrier to the suppression of cancer development because it
stabilizes the replication fork, promotes repair, and prevents late origin firing [3–5]. The damage
that triggers this intra-S checkpoint stems from premature entry into S phase and the inappropriate
initiation of DNA replication [6], which leads to replication fork stalling and RNA–DNA polymerase col-
lision, generating further DNA damage. In contrast to the DNA double-strand break (DSB) (see
Glossary) response mediated by the checkpoint axis ATM (ataxia telangiectasia mutated)–CHK2
(checkpoint kinase 2) and p53 [7], the response to DNA replication stress does not require p53
[8–10]. Indeed, p53-deficient cancer cells rely on the ATR (ataxia telangiectasia and Rad3
related)–CHK1 (checkpoint kinase 1) branch of the checkpoint cascade for their survival when
challenged by replication stress or fork-associated DNA damage. ATR inhibition has been shown
to exacerbate the toxicity of replication stress in p53-deficient cells [11] and in yeast [12]. Finally, be-
cause common therapeutic agents such as methyl methanesulfonate (MMS), cisplatin, aphidicolin
and hydroxyurea (HU) interfere with DNA replication [13], replication stress is enhanced during
many commonly used chemotherapeutic protocols. Not surprisingly, ATR and CHK1 inhibitors 1
Friedrich Miescher Institute for
have emerged as promising cancer treatments because cancer cells are particularly dependent Biomedical Research, Maulbeerstrasse
on this pathway for cell division [14,15]. 66, CH-4058 Basel, Switzerland
2
Faculty of Sciences, University of Basel,
Klingelbergstrasse 90, CH-4056 Basel,
Many targets and functions downstream of ATR have been identified in budding yeast, which has Switzerland
3
a highly analogous intra-S checkpoint that is fully dependent on the yeast ATR homologue Mec1. www.fmi.ch
Both ATR and Mec1 are known to modify replication fork processivity factors, as well as other
kinases, to downregulate replication initiation, including the DDK (Dbf4-dependent kinase),
formed by Cdc7 (cell division cycle 7) and Dbf4 (dumbbell forming 4). Moreover, a number of
*Correspondence:
subunits of the replication polymerases themselves are phosphorylated upon checkpoint activa- susan.gasser@fmi.ch (S.M. Gasser).
tion (reviewed in [16]; see also [17–23]). For example, DNA polymerase (Pol) α, which primes @
Twitter: @susan_gasser

476 Trends in Genetics, May 2021, Vol. 37, No. 5 https://doi.org/10.1016/j.tig.2020.12.003


© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Trends in Genetics
OPEN ACCESS

replication initiation on the leading strand and later the lagging strand, is both a trigger [24] and a Glossary
target of the replication checkpoint [25], whereas DNA Pol ε is modified by ATR/Mec1 or by ATM ATM: ataxia telangiectasia mutated;
(also by the yeast homologue Tel1) [17,22]. In yeast, the Pol31 subunit of DNA Pol δ is modified central DNA damage response
by Mec1 in response to MMS [16,17,22] and HU [16], whereas the Pol32 subunit was shown to checkpoint kinase in mammals.
ATR: ataxia telangiectasia and Rad3
be phosphorylated by cyclin-dependent kinase 1 (CDK1) [26] during the cell cycle; yet it is also
related; central intra-S and replication
modified in response to osmotic stress [27]. checkpoint kinase in mammals.
BER: base excision repair; two-
In this review, we focus on DNA Pol δ, important for its roles in repair and genomic replication. It pronged repair pathway for removal and
replacement of damaged bases.
has been implicated in recovery from replication fork collapse through multiple pathways, includ- BIR: break-induced replication;
ing break-induced replication (BIR), error-free template switch and error-prone translesion DNA mechanism of fork restart after
synthesis (TLS; reviewed in [28]). Moreover, all DNA Pol δ subunits have been found to be replication fork collapse.
overexpressed at the protein level in human cancers (as indicated by Human Protein Atlas), BRCA1/2: Breast cancer type 1/2
susceptibility protein; proteins
and the POLD3 gene itself is frequently found amplified in human tumors [29]. It is unclear whether involved in multiple steps of DSB
DNA Pol δ amplification is oncogenic per se or whether it is a protection mechanism triggered to repair by homologous recombination,
improve tolerance to replication stress. Here we describe the main structural features of DNA Pol but BRCA1 is also involved in
resolving RNA-DNA hybrids that
δ and review its roles in DNA replication stress survival.
generate replication stress.
CDC7: cell division cycle 7; essential
DNA Polymerases in DNA Replication kinase working with DBF4 to regulate
DNA replication is carried out by the DNA polymerases α, δ and ε. The DNA Pol α–primase com- the initiation of DNA replication, it
modifies the minichromosome
plex establishes RNA primer templates that are extended by Pol ε on the leading strand, whereas maintenance complex (MCM) and other
Pol δ extends primers on the lagging strand and is heavily implicated in DNA repair. In contrast to proteins involved in DNA replication.
leading strand synthesis, and because of the universal 5′ to 3′ directionality of nucleotide addition, CHK1: checkpoint kinase 1; key
checkpoint kinase downstream of ATR
synthesis of the lagging strand is carried out in a discontinuous manner requiring sequential nu-
and ATM.
cleation of Okazaki fragments by the Pol α–primase complex, which is then elongated by CFS: chromosomal common fragile
PCNA (proliferating cell nuclear antigen)–Pol δ. In order to ensure proper ligation of consecutive sites; specific chromosomal locations
Okazaki fragments, Pol δ displaces the RNA primer downstream of the newly synthesized Oka- that are susceptible gaps or breaks
when the cell is exposed to replication
zaki fragment, whereas exonuclease FEN1 (flap endonuclease 1) cleaves the displaced RNA and
stress.
then Pol δ again fills the gap. Finally, the nicked ends/phosphate backbones are connected by CNA: copy number alterations are
DNA ligase I. During replication fork restart, which often uses BIR as a mechanism, Pol δ is somatic changes to chromosome
also used for leading strand synthesis [30]. Indeed, in Schizosaccharomyces pombe (fission structure (such as mutations) that result
in gain or loss in copies of sections of
yeast), Pol δ has been shown to replicate both strands after strand invasion–mediated fork re-
DNA and are prevalent in many types of
start, which resembles BIR [31]. cancer.
CysA-CysB: DNA polymerases (Pol α,
δ, and ε) and DNA Pol ζ contain two
Structure of DNA Pol δ
conserved cysteine-rich metal-binding
Budding yeast Pol δ comprises three subunits: the catalytic subunit Pol3, which contains a poly- (Zn2+) motifs (CysA and CysB) in the
merase and an exonuclease domain, and the two accessory subunits Pol31 and Pol32 [32,33] C-terminal domain of their catalytic
(Figure 1 and Table 1). Recently, the structure of the yeast DNA-bound Pol δ holoenzyme has subunits.
DBF4: dumbbell forming 4; regulatory
been resolved by cryogenic electron microscopy (cryo-EM). In contrast to previous speculation,
subunit of CDC7 kinase, required for
the data demonstrate that only the catalytic Pol3 subunit directly interacts with DNA. Despite initiation of DNA replication.
the fact that the regulatory subunits Pol31 and Pol32 contain domains that are potentially capable DDK: DBF4-dependent kinase; the
of DNA binding, their position deep within the holoenzyme prevents direct DNA interaction [34]. DBF4–CDC7 complex is an essential,
conserved Ser/Thr protein kinase that
regulates DNA replication.
The human Pol δ complex exhibits a very similar structure consisting of the catalytic subunit DSB: DNA double-strand break;
POLD1 and structural subunits POLD2 and POLD3, respectively (Figure 1). In contrast to bud- cytotoxic lesion of DNA where both DNA
ding yeast, human Pol δ contains a fourth subunit, POLD4. This subunit was first discovered in strands are broken, for example, by
ionizing radiation or at collapsed
fission yeast, where it is called Cdm1 or p12 (Table 1). However, S-phase entry and genotoxic replication forks.
stress trigger dissociation and proteasomal degradation of POLD4, leaving Pol δ as a trimer for dsDNA: double stranded DNA; stable
DNA replication and repair [35,36]. Compared with the tetrameric Pol δ, the trimeric form has a form of two antiparallel polynucleotide
40-fold better proofreading capability and thus has higher fidelity and a reduced capacity strands connected via hydrogen bonds
(A-T and C-G); unzipped into single
to carry out TLS [37]. Loss of POLD4 per se does not affect the viability of human cells, but

Trends in Genetics, May 2021, Vol. 37, No. 5 477


Trends in Genetics
OPEN ACCESS

stranded DNA during replication and


CysB Fe-S transcription.
Pol32/POLD3 FEN1: flap endonuclease 1; cleaves
RNA flap during Okazaki fragment
Pol3/POLD1 Pol31/POLD2 POLD2
POLD1 processing.
Fe-S: iron–sulfur cluster; the Fe centers
PCNA POLD4 are tetrahedral and the terminal ligands
are thiolated sulfur centers from cysteinyl
residues. Iron–sulfur clusters are found
in a variety of metalloproteins.
PCNA
FANCJ: Fanconi anemia
complementation group J is implicated
in DNA damage detection and repair
POLD3
CysA and cell cycle checkpoint control.
HU: hydroxyurea; an inhibitor of dNTP
DNA synthesis.
HR: homologous recombination; a DSB
repair pathway that relies on the strand
Pol3/POLD1 Exonuclease Polymerase PIP CysA CysB
invasion of a homologous template for
repair.
Pol31/POLD2 Pol32-BD OB PIP PDE Mec1: mitosis entry checkpoint 1; a
yeast homologue of ATR.
MMR: DNA mismatch repair; DNA
Pol32/POLD3 Pol31-BD PIP
repair mechanism to resolve mis-paired
nucleotides and small insertions or
POLD4 PIP deletions.
MMS: methyl methanesulfonate; an
Trends in Genetics
alkylating mutagen.
Figure 1. Model and Structure of Yeast (Pol3, Pol31, and Pol32) and Human (POLD1, POLD2, POLD3, and NHEJ: nonhomologous end joining;
POLD4) Replication DNA Polymerase (Pol) δ Complex. Upper panel: on the left is a model of the human and yeast DNA DSB repair pathway dependent on
Pol δ complex based on cryogenic electron microscopy (cryo-EM) data (with POLD4 existing only in the human Pol δ Ku70/80 dimer.
complex) [34,39]. On the right is the corresponding 3D structure of the processive human Pol δ holoenzyme (Protein Data NER: nucleotide excision repair is an
Bank identifier 6TNY) as provided by Lancey et al. [39] and created with the Research Collaboratory for Structural excision mechanism that senses helix
Bioinformatics Protein Data Bankii,iii [104]; Pol δ subunits are colored accordingly. For accessibility reasons, POLD4 has distortions and repairs DNA damage
been removed from the graphic. The cysteine-rich metal-binding motif A (CysA; C1012–C1029) and CysB (C1057– induced by UV light.
C1076) motifs in POLD1 (highlighted in yellow) harbor the main interaction sites for interaction with proliferating cell nuclear p53: tumor protein 53; tumor
antigen (PCNA) and POLD2, respectively. The iron sulfur cluster (Fe-S) within the CysB motif is highlighted in red. Lower suppressor protein that regulates the cell
panel: domain organization of Pol δ subunits, PCNA-interacting protein (PIP), cysteine-rich metal-binding domains CysA cycle.
and CysB, oligonucleotide-binding domain (OB), and phosphodiesterase domain (PDE). PBMCs: peripheral blood mononuclear
cells; these are lymphocytes (T cells,
B cells, natural killer cells) and
D-loop displacement synthesis in homologous recombination is apparently reduced [38]. monocytes.
PCNA: proliferating cell nuclear
Similar to that of budding yeast, the atomic structure of the human Pol δ complex has also
antigen; processivity factor for DNA
now been resolved (Figure 1) [34,39]. Complementing the yeast study, Lancey et al. [39] Pol δ and ε.
PIP: PCNA-interacting protein motif.
POLD: DNA polymerase δ subunits
Table 1. Subunits of DNA Pol δ across Speciesa,b POLD1, POLD2, POLD3, or POLD4.
DNA polymerase δ Mammalian Schizosaccharomyces Saccharomyces Repair pathway RPA: replication protein A; binds
pombe cerevisiae ssDNA during DNA repair.
TLS: translesion DNA synthesis; DNA
Catalytic subunit POLD1, p125 125 kDa, Pol3 125 kDa, Pol3 BIR, TS, MMR, NER,
damage tolerance process that allows
DSB repair
the DNA replication machinery to
Second subunit POLD2, p50 55 kDa, Cdc1 58 kDa Pol31 TLS replicate DNA lesions or distorted
templates.
Third subunit POLD3, p68 54 kDa, Cdc27 55 kDa Pol32 TLS, BIR
TS: error-free template switch is a DNA
Fourth subunit POLD4, p12 22 kDa, Cdm1 Not found HR, BIR damage tolerance whereby the stalled
nascent strand switches from the
a
Shown are the subunits of DNA polymerase δ, their size, and species’ specific names. Note that the repair pathways only damaged template to the undamaged,
indicate pathways compromised upon mutation of the subunit reported in the literature; yet, because POLD2 is essential for newly synthesized sister strand for
POLD1 activity, the repair pathways ascribed to POLD1 would naturally also depend on POLD2. Listing is based on
replication over the lesion.
[33,60,105,106].
b Tel1: telomere maintenance 1; budding
Abbreviations: BIR, break-induced replication; DSB, double-strand break; HR, homologous recombination; MMR, mismatch
yeast homologue of ATM kinase.
repair; NER, nucleotide excision repair; POLD, polymerase delta; TLS, translesion synthesis; TS, error-free template switch.

478 Trends in Genetics, May 2021, Vol. 37, No. 5


Trends in Genetics
OPEN ACCESS

assessed the structure of DNA-bound human Pol δ in the presence of PCNA and FEN1. XPD: xeroderma pigmentosum group
D protein (also named ERCC2); a DNA
Strikingly, they identified a hitherto unknown noncanonical PIP (PCNA-interacting protein) be-
helicase containing an Fe-S cluster
tween amino acids 1001 and 1005 in POLD1, upstream of the CysA–CysB (cysteine-rich domain.
metal-binding motifs A and B) motifs (see later). This tethers POLD1 to PCNA and was
shown to be essential for Pol δ activity and processivity. A previously known PIP motif in
POLD4 also contributes to Pol δ processivity [40], most likely by modulating the relative posi-
tioning of the catalytic and accessory subunits, thereby playing off the polymerase velocity
against fidelity. Finally, the orientation and tilting of PCNA orchestrates Okazaki fragment
maturation by coordinating FEN1, Pol δ and DNA ligase I [39].

Crucial Structural Features of DNA Pol δ


All four subunits of human Pol δ contain putative PCNA interaction motifs (Figure 1). The catalytic
subunit POLD1 (Pol3) harbors a C-terminal Zn2+-binding CysA at amino acids 1012–1030 in the
human protein, which is critical for interaction with the sliding clamp PCNA (Figure 1). In yeast, be-
sides Pol3, Pol31 and Pol32 are reported to bind PCNA; yet, in our hands, yeast Pol31 can only
bind PCNA in two-hybrid assays in the absence of Pol32 [41]. Consistently, human Pol δ compo-
nent POLD2 (Pol31) has only low affinity for PCNA, and the interaction is supported instead by
PCNA interaction with POLD3 and POLD4. Despite a multiplicity of PIP domains in the complex,
POLD1 is clearly the key interaction partner for PCNA [39,42], and its activity increases 10-fold
in vitro upon addition of PCNA [43]. In yeast, the mutation of various PIPs in individual subunits
does not affect survival in unstressed conditions, indicating that the loss of any one of the three
PCNA interaction domains is compensated by interaction with others [44].

The C-terminal domain of POLD1 (Pol3) contains a second metal-binding CysB motif that was ini-
tially believed to bind Zn2+ ion similarly to CysA. More recent data, however, suggest that this is a
[4Fe-4S] cluster coordinated by the four cysteines C1056, C1059, C1069 and C1074 and
Arg1080 (Figure 1) [34,39,45,46]. The CysB domain is crucial for the Pol δ holoenzyme structure
because the α1 helix upstream and α2 helix downstream of the [4Fe-4S] cluster interact with the
phosphodiesterase- and oligonucleotide-binding domain of accessory subunit Pol31 (POLD2). In
addition, this domain contacts the exonuclease and thumb domain of Pol3 (POLD1), creating a
more compact structure that is likely the driver of increased enzyme activity and fidelity [34,47].
Human POLD4, which has no yeast homolog, also associates with the α2 helix downstream of
the [4Fe-4S] cluster.

DNA Polymerase δ in DNA Replication and Repair


Accumulating data from yeast and chicken DT40 cells implicate POLD2 (Pol31) and POLD3 (Pol32)
in DNA repair, particularly at the replication fork. The Pol32 subunit of Pol δ is required for the initi-
ation of BIR, a recovery pathway that allows restart of stalled or broken replication forks through
invading a template sequence with homology to one end of the DSB. Pol31 and Pol32 are also
found as core subunits of the DNA polymerase zeta (Pol ζ), which mediates error-prone TLS
[47,48]. Intriguingly, a recent study in chicken DT40 cells found that POLD3 has an additive contri-
bution to TLS, independent of Pol ζ [49], and a similar finding was made in yeast [50]. It was later
shown that inactivating the proofreading activity of POLD1 restores TLS activity in POLD3-deficient
chicken DT40 cells, suggesting that the PolD1–PolD2 complex can also mediate TLS [51]. Along
this line, yeast cells lacking Pol32 show cold-, HU- and MMS sensitivity, which can be suppressed
by a point mutation in Pol31 that enhances the Pol31–Pol3 interaction [41].

POLD2 and POLD3 are also subunits of the human TLS polymerase ζ (Pol ζ) [52], which carries
out both error-prone and error-free TLS in mammalian cells [53]. These pathways allow cells to
tolerate damaged DNA without repair. For instance, Pol ζ in cooperation with the TLS polymerase

Trends in Genetics, May 2021, Vol. 37, No. 5 479


Trends in Genetics
OPEN ACCESS

Pol η even allows bypass of cisplatin-induced GG adducts [52,54]. Consistently, a recent study
showed that overexpression of the catalytic subunit of Pol ζ (REV3L) conferred resistance of
H1299 cells to cisplatin, whereas its knockdown increased cisplatin sensitivity in the same cells
[55].

Although the loss of either of the accessory subunits POLD2 and POLD3 is lethal in mice and
humans, only the POL31 deletion has a similarly fatal effect in yeast, at least in the absence of
DNA damage. The essential role of Pol31 is unclear; yet, it may mainly have a structural role, en-
suring that the catalytic subunit is stable, to facilitate both lagging strand synthesis and long-
patch DNA repair [56]. Pol32 is implicated specifically in break-induced fork restart, although
Pol31, being essential, is also likely involved in this pathway.

Recent studies have begun to address the role of Pol δ and its subunits in human cells undergo-
ing replication stress. For instance, under replication stress, the structure-specific MUS81-EME1
endonuclease is recruited to chromosomal fragile sites (CFS) in early mitosis, generating a free
end and a gap that requires DNA synthesis for the completion of chromosomal replication.
After MUS81 cleavage, POLD3, and with it POLD1, are recruited to allow early mitotic filling in
of these gaps by DNA synthesis, reducing chromosome mis-segregation [57,58]. Given that
POLD3-dependent mitotic DNA synthesis is enhanced in aneuploid cancer cells, it was sug-
gested that targeting this pathway could represent a new therapeutic approach against cancer.
Interestingly, the overexpression of POLD3 and MUS81 are correlated in human cancer lines [59].
In addition to the correlation between MUS81 and POLD3 overexpression, we note a positive
correlation of POLD3 with elevated levels of other structure-specific nucleases, GEN1, RAD1
and MLH1, not to mention DNA2 and FEN1, which are involved in Okazaki fragment maturation
(Cancer Cell Line Encyclopedia RNA-sequencing data retrieved from cBioPortal). Elevated GEN1,
RAD1 and MLH1, together with MUS81 and POLD3, are negative prognostic factors for liver,
breast and endometrial cancers. Several of these endonucleases are necessary for resolving
structures that arise from strand invasion or other replication-associated repair pathways. Con-
sistently, in U2OS cells, it was shown that the POLD3 and POLD4 subunits of Pol δ are required
for DNA synthesis and cell cycle progression in the presence of oncogene-induced replication
stress, accounting for a significant portion of the genomic copy number alterations (CNAs) gen-
erated under these conditions [60]. On the basis of these findings, the authors proposed that BIR
mediated by POLD3 at compromised replication forks accounts in part for short-range genomic
duplications in human cancers. Finally, Pol δ can promote intra- and interchromosomal transloca-
tions through alternative NHEJ (nonhomologous end joining), thereby further contributing to po-
tentially oncogenic genome alterations [61].

Oncogene-Induced Replication Stress in Cancer Development


The crosstalk between repair pathways in cancer is widespread. Oncogene-induced replication
stress triggers the DNA damage response, which first arrests the cell cycle to enable repair and
prevent preneoplastic lesions from driving malignant transformation [5]. However, the loss of
tumor suppressors and perturbations of DNA damage response pathways [e.g., BRCA1 (breast
cancer type 1 susceptibility protein), BRCA2 and ATR] allow tumor cells to accumulate further
mutations and drive tumorigenesis. Such perturbations render tumors even more dependent
on other DNA damage repair pathways, including, inevitably, DNA Pol δ (Figure 2). Thus, tumor
cells are often ‘addicted to’, or dependent on, specific repair functions.

In addition to oncogene-linked replication stress, endogenous genotoxic stresses (e.g., metabolism-


derived oxygen radicals or the collision of transcription and replication) augment DNA damage [62].
Most endogenous damage reflects ssDNA (single-stranded DNA) nicks or breaks or stems from

480 Trends in Genetics, May 2021, Vol. 37, No. 5


Trends in Genetics
OPEN ACCESS

(A) (B) Loss of a repair pathway (C) Inhibion of the survival ensuring
e.g., HR in BRCA1 -/- pathways (genec or chemical)
NHEJ BER HR TLS BIR
Lesion tolerance Template switch/
or bypass fork restart
NHEJ BER HR TLS BIR NHEJ BER HR TLS BIR

DDR DDR DDR

Cancer CancerHR CancerHR

Oncogenic mutaons Cancer cells become


generate replicaon dependent on TLS/BIR for
stress; mutaons Synthetic
survival, possibly driving
promote survival Pol δ amplificaon lethality

Trends in Genetics

Figure 2. Illustration of the Synthetic Lethality Concept in Cancer Therapy. (A) Multiple repair pathways protect the cellular genome from damage through
conserved mechanisms that collectively activate the DNA damage response (DDR). In normally dividing cells, this prevents mutations that could lead to cancer; many
components of this pathway are in fact tumor suppressors. Key survival pathways include nonhomologous end joining (NHEJ); base excision repair (BER); homologous
recombination (HR); and then, in response to replication stress, break-induced replication (BIR) or translesion synthesis (TLS). (B) If precancerous lesions lead to
mutations that compromise one pathway of repair (for instance, compromised HR due to BRCA1 or BRCA2 mutation), the cell may become dependent on another
repair pathway, such as TLS and BIR, that could trigger upregulation of DNA polymerase (Pol) δ. (C) In summary, inhibition of the survival pathway to which cancer
cells are ‘addicted’ to, or dependent on, leads to synthetic cell death, particularly if challenged with hydroxyurea (HU) or another agent that aggravates replication stress.

ribonucleoside triphosphate (rNTP) misincorporation, but, depending on the efficiency and integrity of
repair, these errors can be converted into far more deleterious dsDNA (double-stranded DNA) breaks
[63]. Various repair mechanisms exist to protect against DNA lesions: base excision repair (BER); mis-
match repair (MMR); nucleotide excision repair (NER); and finally, if DSBs occur, either homologous
recombination (HR) or more error-prone NHEJ usually takes care of them. Remarkably, Pol δ serves
in all of these repair mechanisms, except canonical NHEJ and short-patch BER [30,64,65]. Thus, it
is reasonable to propose that the elevated protein levels of POLD2 and/or POLD3 in oncogene-
transformed cells may stem from their ability to protect cells from otherwise lethal levels of DNA
damage. However, by promoting transformed cell survival, elevated levels of Pol δ may also favor
late-stage oncogenesis. Intriguingly, many cancers accumulate base substitutions in late replicating
regions, with relatively fewer substitutions in early replicating domains [66], probably as a result of al-
tered repair rather than targeted mutation [67].

Effects of Hydroxyurea on Pol δ: A Role for [4Fe-4S] Chelation


To mimic the state of replication stress arising in oncogene-transformed cells, where indiscriminate
origin firing can lead to deoxynucleotide triphosphate (dNTP) restriction and replication fork arrest
[6], scientists often use HU. However, HU is also a human therapeutic, used particularly for acute T-
cell leukemia. HU inhibits the enzymatic activity of the key ribonucleotide reductase subunit RRM2
[68,69]. The HU-induced scavenging of the tyrosyl radical in RRM2 prevents the conversion of ri-
bonucleoside diphosphates to dNTPs, leading to dNTP shortage and, as a consequence, reduced
replication fork progression, ssDNA accumulation and replication checkpoint activation. The ques-
tion arises whether HU has other replication-relevant effects than the reduction of dNTP levels.

Recent findings have uncovered a second effect of HU in that it specifically affects Fe-S cluster co-
ordinating proteins, apparently by generating an oxidative stress that targets redox-sensitive iron
[70]. In this study, the authors showed that HU treatment strongly compromised the enzymatic

Trends in Genetics, May 2021, Vol. 37, No. 5 481


Trends in Genetics
OPEN ACCESS

activity of the Fe-S protein LEU1, an isopropylmalate dehydratase. Although LEU1 is not involved in
repair, the finding was striking because many proteins involved in DNA replication and repair con-
tain an Fe-S cluster, including all four B-family DNA polymerases: Pol α, δ, ε, and ƺ [45]. Further
strengthening the importance of Fe-S coordination is the observation that MMS19, which associ-
ates with the cytoplasmic Fe-S assembly complex CIA (CIAO1, IOP1, MIP18), directly recognizes
and binds client proteins such as POLD1 [71–73]. The loss of MMS19 prevents Fe-S attachment to
target proteins, provokes target protein instability and reduces levels of DNA replication/repair
proteins, such as levels of XPD (xeroderma pigmentosum group D protein), FANCJ (Fanconi
anemia complementation group J), not to mention POLD1 [72].

Oxidation of [4Fe-4S] in yeast and human Pol δ, such as that occurring upon exposure to HU, de-
creases Pol δ processivity at the level of both its exonuclease and polymerase activities, despite
an enhanced DNA-binding affinity and a lower rate of PCNA-Pol δ sliding along DNA [46,74]. This
reduction in Pol δ processivity drives ssDNA accumulation, accelerates RPA (replication protein
A) binding and leads to checkpoint activation, which feeds back to arrest fork progression. More-
over, the replication fork restart that depends on ssDNA invasion of an intact template requires
Pol δ, being strongly compromised by the loss of Pol32 in yeast [75]. Consistently, the lethal
C1059S and C1074S mutations, which compromise the CysB motif in budding yeast Pol3, ab-
late interaction with Pol31 and thus with Pol32. Although the adjacent CysA motif is indeed
thought to chelate Zn and mediate binding to PCNA, the CysB structure and [4Fe-4S] chelation
are crucial for Pol δ complex formation and activity [45,76] (see Outstanding Questions).

In humans, the situation may be slightly different: an artificial mutational disruption of the human
POLD1 CysB by the C1076F substitution seemed to affect POLD2 and POLD4 interaction only
mildly and had no overt effect on POLD3 binding [46]. Still, samples from syndromic immunode-
ficiency patients harboring POLD1 R1060C and R1074W mutations in the CysB domain showed
a substantial loss of interaction with POLD2 and POLD3 [77,78]. The attenuation of CysB muta-
tions in humans may be due to stabilizing effects of PCNA through the fourth Pol δ subunit,
POLD4, which is missing in yeast [39,46].

Interestingly, perturbations of the accessory Pol δ subunits can similarly destabilize the complex.
For instance, Pol31 RNAi reduces Pol3 and Pol32 levels in Drosophila [79] and patient-derived
PBMCs (peripheral blood mononuclear cells) carrying POLD2 D293N have barely detectable
levels of POLD1, POLD2, and POLD3 [77]. Similarly, the loss of POLD3 reduces protein levels
of POLD1 and POLD2 in mice [80], in human HeLa and Jurkat cells [81], and in flies and in
budding yeast [82]. Taken together, it is clear that the loss of a single noncatalytic Pol δ subunit
(excepting POLD4) provokes complex instability and reduced protein levels of the other Pol δ
subunits across many species. Consistently, budding yeast strains lacking POL32 (pol32 ) are
viable under normal conditions because they retain sufficient residual Pol δ complex for normal
replication but display enhanced sensitivity toward HU-derived DNA damage or replication stress
induced by MMS [33]. Both translesion synthesis and the restart at collapsed replication forks
mediated by BIR are almost entirely dependent on Pol32. However, this may also reflect the sta-
bility of the holocomplex: Similar results were obtained by deleting the last four amino acids in
Pol3, which causes lethality only upon HU or in combination with pol32 . This deficiency can be
suppressed by a gain-of-function mutation in Pol31 (K358E or I), which is thought to strengthen
the Pol3–Pol31 interaction and increase catalytic activity [56,76].

Pol δ in Human Diseases


Each individual Pol δ component has been implicated in various human diseases. Patients with
MDPL syndrome (mandibular hypoplasia, deafness, progeroid features, and lipodystrophy),

482 Trends in Genetics, May 2021, Vol. 37, No. 5


Trends in Genetics
OPEN ACCESS

which closely resembles the progeroid phenotype of patients with Werner syndrome (mutations
in RecQ helicase WRN) have been found to harbor germline mutations in POLD1 perturbing its
polymerase and exonuclease activity [83–85], suggesting that WRN may work together
with DNA Pol δ. Other POLD1 mutations cause a combined immunodeficiency syndrome
[77,78]. Yet, most mutations in Pol δ subunits were identified as predisposing to a range of
cancers, predominantly colorectal cancer but also endometrial and skin cancers [86–89]
(Figure 3, Key Figure).

Importantly, POLD1-altered human cancers often display a hypermutator phenotype [90,91].


This includes mutations that specifically affect the proofreading activity or dNTP selectivity of
POLD1 and also drive hypermutation in mice [92,93]. This defective proofreading activity of Pol
δ is associated with increased tumor incidence and decreased survival in mice [93], and germline
and sporadic mutations in POLD1 proofreading exonuclease domain have been identified in sev-
eral human cancers [87]. Interestingly, the spectrum of tumors emerging in POLD1 proofreading–
defective mice overlaps with that in humans (skin, colon, lung) [94]. We find that most oncogenic
genetic alterations in POLD1 are mutations, whereas the loci encoding subunits POLD2, POLD3,
and POLD4 are rarely mutated but rather show frequent amplification (Figure 3). This is true not
only genetically but also with respect to expression levels [95–101]. Indeed, based on the
Human Protein Atlas, all Pol δ proteins can be found at elevated levels in human cancers, and
the POLD3 gene is often amplified [29]. Interestingly, POLD4, which is also on chromosome
11, is often coamplified.

The elevated levels of POLD1–POLD4 subunits may ensure survival because a recent study in U2OS
cells demonstrated that the POLD3 and POLD4 subunits of Pol δ are required for DNA synthesis and
cell cycle progression in the presence of oncogene-induced replication stress, a hallmark of cancer
cells. However, the study also demonstrated that POLD3/POLD4 accounted for a significant portion

Key Figure
The Distribution of Mutations and Copy Number Alterations (CNAs) in
DNA Polymerase (Pol) δ Subunits

(A) (B)
Liver
15 Breast
Alteration frequency (%)

Multiple Unaltered
Deletion Ovary POLD1
10 Amplification Colorectal POLD2
Mutation Glioma POLD3
Melanoma POLD4
5
Stomach
Upper aerodigestive
lymphoma
0 (Burkitt, Hodgkin, DLBCL)
AML
1

4
LD

LD

LD

LD
PO

PO

PO

PO

0 20 40 60 80
Total numbers of alterations

Trends in Genetics

Figure 3. (A) Alteration frequencies of Pol δ components in human cancers. Mutational and copy number alteration (CAN)
data are based on the 961 cell lines in the Cancer Cell Line Encyclopedia (Broad Institute, 2019, from cBioPortal). (B) The
total number of alterations of Pol δ components in selected tumor types (overlapping samples included). We note that
POLD1 accumulates point and missense mutations, whereas POLD2 and POLD3 primarily have CMAs. Abbreviations:
AML, acute myeloid leukemia; DLBCL, diffuse large B-cell lymphoma.

Trends in Genetics, May 2021, Vol. 37, No. 5 483


Trends in Genetics
OPEN ACCESS

of genomic CNA generated under prolonged conditions of oncogene-induced replication stress, Outstanding Questions
arguing that BIR repair by POLD3 and POLD4 at compromised replication forks could be responsible What is the mechanism driving DNA
for genomic duplications, which also characterize human cancers [60]. Intriguingly, however, the Pol δ amplification in cancer cells?
hypermutational phenotype in developing tumors caused by loss of polymerase fidelity appears to
Is the amplification of the accessory
arise exclusively from POLD1 mutations. By contrast, elevated levels of the accessory subunits subunits of DNA Pol δ necessary for
promote oncogenesis by bolstering Pol δ activity sufficiently to cope with oncogenic replication stress survival of the oncogene-induced repli-
(that is, by supporting DNA repair and replication fork restart activity). Thus, POLD1 mutations may be cation stress found in cancer cells?
tumor drivers, whereas amplification of the other subunits may initially occur to suppress damage,
Is there a compensatory pathway for
later driving survival of the mutated genome. DNA replication stress survival in
POLD2/3-overexpressing cells? If so,
Targeted Synergistic Lethality in Pol δ–Overexpressing Cells: Implications for which pathway ?

Cancer Therapies Do the phosphoacceptor sites for


As mentioned earlier, the elevated levels of amplified Pol δ subunits, such as POLD2 and/or checkpoint kinases on DNA Pol δ sub-
POLD3, in oncogene-transformed cells likely reflect their ability to protect from lethal levels of units regulate repair or fork recovery
pathways?
DNA damage. One might use this observation to understand which other genes or repair
pathways become essential for growth under conditions of POLD2 or POLD3 overexpression To what extent can DNA Pol ε replace
(see Outstanding Questions). Conditional synthetic lethal screens followed by hit validation in DNA Pol δ in fork recovery, and if not,
why not?
human cancer cell lines may identify useful targets for therapeutic intervention for the large
number of cancers in which Pol δ subunits are overexpressed or in other cancers with compro- How does the [Fe-S] cluster regulate
mised replication fork recovery. protein–protein interactions during
HU-induced replication stress?

The general concept behind such an approach is to exploit conditional synthetic sensitivity, based
on genetic changes that are likely to arise in cancer cells. Exploiting this concept with respect to
DNA repair pathways implies inhibiting compensatory pathways of DNA repair or replication
stress survival to provoke synergistic lethality in damage pathway–addicted cancer cells
(Figure 2). This strategy not only is useful for identifying parallel or compensatory repair and
replication checkpoint pathways under stress but also may be exploited for highly efficient com-
bination with targeted therapies or chemotherapy. To this end, a mutational analysis of tumors
sensitive to DNA Pol δ inhibition would be a useful approach.

On the other hand, tumors with compromised POLD1 activity seem to be particularly vulnerable
to cancer therapies using ATR or CHK1 inhibitors [102]. If DNA replication is impaired, as is the
case when POLD1 function is lost, cells would usually undergo cell cycle arrest in order to
allow DNA repair. However, if the cell cycle checkpoint is suspended by inhibition of ATR or
CHK1, cells are engulfed in mitotic crisis and eventually cell death. Furthermore, cancer cells
with a hypermutator phenotype, as observed upon impaired exonuclease function of POLD1, ex-
hibit high levels of cancer cell–specific neoantigens that can be recognized and targeted by im-
mune cells [103]. Such elevated immunogenicity could be exploited for therapies based on
immune checkpoint inhibitors.

Concluding Remarks
Due to oncogenic mutations and replication stress, cancer cells often become dependent on re-
pair pathways, many of which make use of DNA Pol δ. The fact that DNA Pol δ subunit levels are
upregulated in human cancers argues for a dependence on this polymerase, particularly in repli-
cation stress survival (see Outstanding Questions). In this review, we highlight the well-
characterized roles of DNA Pol δ in DNA replication, as well as in DNA repair. We conclude
from the current literature that a strict regulation of DNA Pol δ subunit expression is crucial to
maintain genomic integrity and to deal with oncogene-induced stress. We propose that
POLD2/POLD3 overexpression may be an indicator for specific therapeutic sensitivities, and
therefore the identification of the repair pathways that these cancer cells rely on may open the

484 Trends in Genetics, May 2021, Vol. 37, No. 5


Trends in Genetics
OPEN ACCESS

door for more efficient targeted combinational cancer therapy. Furthermore, POLD1 function is
found to be mutated in many cancers, where it drives hypermutation.

Resources
i
www.proteinatlas.org/ENSG00000062822-POLD1/cancer
ii
www.rcsb.org/3d-view/6TNY
iii
https://diglib.eg.org/handle/10.2312/molva20181103

References
1. Bartkova, J. et al. (2005) DNA damage response as a candi- 21. Bennetzen, M.V. et al. (2010) Site-specific phosphorylation
date anti-cancer barrier in early human tumorigenesis. Nature dynamics of the nuclear proteome during the DNA damage
434, 864–870 response. Mol. Cell. Proteomics 9, 1314–1323
2. Gorgoulis, V.G. et al. (2005) Activation of the DNA damage 22. Chen, S.H. et al. (2010) A proteome-wide analysis of kinase-
checkpoint and genomic instability in human precancerous substrate network in the DNA damage response. J. Biol.
lesions. Nature 434, 907–913 Chem. 285, 12803–12812
3. Bartkova, J. et al. (2006) Oncogene-induced senescence is 23. Blasius, M. et al. (2011) A phospho-proteomic screen identifies
part of the tumorigenesis barrier imposed by DNA damage substrates of the checkpoint kinase Chk1. Genome Biol. 12,
checkpoints. Nature 444, 633–637 R78
4. Di Micco, R. et al. (2006) Oncogene-induced senescence is a 24. Michael, W.M. et al. (2000) Activation of the DNA replication
DNA damage response triggered by DNA hyper-replication. checkpoint through RNA synthesis by primase. Science 289,
Nature 444, 638–642 2133–2137
5. Halazonetis, T.D. et al. (2008) An oncogene-induced DNA 25. Pellicioli, A. et al. (1999) Activation of Rad53 kinase in response
damage model for cancer development. Science 319, to DNA damage and its effect in modulating phosphorylation of
1352–1355 the lagging strand DNA polymerase. EMBO J. 18, 6561–6572
6. Macheret, M. and Halazonetis, T.D. (2018) Intragenic origins due 26. Holt, L.J. et al. (2009) Global analysis of cdk1 substrate
to short G1 phases underlie oncogene-induced DNA replication phosphorylation sites provides insights into evolution. Science
stress. Nature 555, 112–116 325, 1682–1686
7. Kastan, M.B. and Bartek, J. (2004) Cell-cycle checkpoints and 27. Soufi, B. et al. (2009) Global analysis of the yeast osmotic
cancer. Nature 432, 316–323 stress response by quantitative proteomics. Mol. BioSyst. 5,
8. Sidi, S. et al. (2008) Chk1 suppresses a caspase-2 apoptotic 1337–1346
response to DNA damage that bypasses p53, Bcl-2, and 28. Costes, A. and Lambert, S.A.E. (2013) Homologous recombina-
caspase-3. Cell 133, 864–877 tion as a replication fork escort: fork-protection and recovery.
9. Murga, M. et al. (2009) A mouse model of ATR-Seckel shows Biomolecules 3, 39–71
embryonic replicative stress and accelerated aging. Nat. 29. Beroukhim, R. et al. (2010) The landscape of somatic copy-
Genet. 41, 891–898 number alteration across human cancers. Nature 463,
10. Myers, K. et al. (2009) ATR and Chk1 Suppress a caspase-3– 899–905
dependent apoptotic response following DNA replication 30. Donnianni, R.A. et al. (2019) DNA polymerase delta synthe-
stress. PLoS Genet. 5, e1000324 sizes both strands during break-induced replication in brief.
11. Toledo, L.I. et al. (2011) A cell-based screen identifies ATR in- Mol. Cell 76, 371–381
hibitors with synthetic lethal properties for cancer-associated 31. Miyabe, I. et al. (2015) Polymerase δ replicates both strands
mutations. Nat. Struct. Mol. Biol. 18, 721–727 after homologous recombination-dependent fork restart. Nat.
12. Cobb, J.A. et al. (2005) Replisome instability, fork collapse, and Struct. Mol. Biol. 22, 932–938
gross chromosomal rearrangements arise synergistically from 32. Sitney, K.C. et al. (1989) DNA polymerase III, a second essen-
Mec1 kinase and RecQ helicase mutations. Genes Dev. 19, tial DNA polymerase, is encoded by the S. cerevisiae CDC2
3055–3069 gene. Cell 56, 599–605
13. Aguilera, A. and Gómez-González, B. (2008) Genome instability: 33. Gerik, K.J. et al. (1998) Characterization of the two small subunits
a mechanistic view of its causes and consequences. Nat. Rev. of Saccharomyces cerevisiae DNA polymerase δ. J. Biol. Chem.
Genet. 9, 204–217 273, 19747–19755
14. Murga, M. et al. (2011) Exploiting oncogene-induced replicative 34. Jain, R. et al. (2019) Cryo-EM structure and dynamics of eu-
stress for the selective killing of Myc-driven tumors. Nat. Struct. karyotic DNA polymerase δ holoenzyme. Nat. Struct. Mol.
Mol. Biol. 18, 1331–1335 Biol. 26, 955–962
15. Toledo, L.I. et al. (2011) Targeting ATR and Chk1 kinases for 35. Zhang, S. et al. (2007) A novel DNA damage response: rapid
cancer treatment: a new model for new (and old) drugs. Mol. degradation of the p12 subunit of DNA polymerase δ. J. Biol.
Oncol. 5, 368–373 Chem. 282, 15330–15340
16. Hustedt, N. et al. (2015) Yeast PP4 interacts with ATR homo- 36. Zhang, S. et al. (2013) A novel function of CRL4Cdt2: regula-
log Ddc2-Mec1 and regulates checkpoint signaling. Mol. Cell tion of the subunit structure of DNA polymerase δ in response
57, 273–289 to DNA damage and during the S phase. J. Biol. Chem. 288,
17. Matsuoka, S. et al. (2007) ATM and ATR substrate analysis re- 29550–29561
veals extensive protein networks responsive to DNA damage. 37. Meng, X. et al. (2008) DNA damage alters DNA polymerase d
Science 316, 1160–1166 to a form that exhibits increased discrimination against modi-
18. Mu, J.J. et al. (2007) A proteomic analysis of ataxia telangiectasia- fied template bases and mismatched primers. Nucleic Acids
mutated (ATM)/ATM-Rad3-related (ATR) substrates identifies the Res. 37, 647–657
ubiquitin-proteasome system as a regulator for DNA damage 38. Zhang, S. et al. (2019) Loss of the p12 subunit of DNA poly-
checkpoints. J. Biol. Chem. 282, 17330–17334 merase delta leads to a defect in HR and sensitization to
19. Smolka, M.B. et al. (2007) Proteome-wide identification of PARP inhibitors. DNA Repair (Amst) 73, 64–70
in vivo targets of DNA damage checkpoint kinases. Proc. 39. Lancey, C. et al. (2020) Structure of the processive human Pol
Natl. Acad. Sci. U. S. A. 104, 10364–10369 δ holoenzyme. Nat. Commun. 11, 1109
20. Stokes, M.P. et al. (2007) Profiling of UV-induced ATM/ATR 40. Khandagale, P. et al. (2020) Identification of PCNA-interacting
signaling pathways. Proc. Natl. Acad. Sci. U. S. A. 104, protein motifs in human DNA polymerase δ. Biosci. Rep. 40,
19855–19860 BSR20200602

Trends in Genetics, May 2021, Vol. 37, No. 5 485


Trends in Genetics
OPEN ACCESS

41. Shimada, K. et al. (2021) Stabilized interaction between DNA 66. Schuster-Böckler, B. and Lehner, B. (2012) Chromatin organi-
polymerases and the second subunit of DNA Pol d bypasses zation is a major influence on regional mutation rates in human
the requirement for Pol32 under replication stress. bioRxiv cancer cells. Nature 488, 504–507
Published online February 11, 2021. https://doi.org/10.1101/ 67. Supek, F. and Lehner, B. (2015) Differential DNA mismatch
2021.02.11.430699 repair underlies mutation rate variation across the human
42. Johansson, E. et al. (2004) The Pol32 subunit of DNA polymer- genome. Nature 521, 81–84
ase δ contains separable domains for processive replication 68. Lassmann, G. et al. (1992) EPR stopped-flow studies of the re-
and proliferating cell nuclear antigen (PCNA) binding. J. Biol. action of the tyrosyl radical of protein R2 from ribonucleotide
Chem. 279, 1907–1915 reductase with hydroxyurea. Biochem. Biophys. Res.
43. Mondol, T. et al. (2019) PCNA accelerates the nucleotide incor- Commun. 188, 879–887
poration rate by DNA polymerase. Nucleic Acids Res. 47, 69. Thelander, M. et al. (1985) Subunit M2 of mammalian ribonu-
1977–1986 cleotide reductase: characterization of a homogeneous protein
44. Acharya, N. et al. (2011) PCNA binding domains in all three sub- isolated from M2-overproducing mouse cells. J. Biol. Chem.
units of yeast DNA polymerase δ modulate its function in DNA 260, 2737–2741
replication. Proc. Natl. Acad. Sci. U. S. A. 108, 17927–17932 70. Huang, M.E. et al. (2016) DNA replication inhibitor hydroxyurea
45. Netz, D.J.A. et al. (2012) Eukaryotic DNA polymerases require alters Fe-S centers by producing reactive oxygen species
an iron-sulfur cluster for the formation of active complexes. in vivo. Sci. Rep. 6, 29361
Nat. Chem. Biol. 8, 125–132 71. Stehling, O. et al. (2012) MMS19 assembles iron-sulfur
46. Jozwiakowski, S.K. et al. (2019) Human DNA polymerase delta proteins required for DNA metabolism and genomic integrity.
requires an iron-sulfur cluster for high-fidelity DNA synthesis. Science 337, 195–199
Life Sci. Alliance 2, e201900321 72. Gari, K. et al. (2012) MMS19 links cytoplasmic iron-sulfur clus-
47. Johnson, R.E. et al. (2012) Pol31 and Pol32 subunits of yeast ter assembly to DNA metabolism. Science 337, 243–245
DNA polymerase δ are also essential subunits of DNA polymer- 73. Kassube, S.A. and Thomä, N.H. (2020) Structural insights into
ase ζ. Proc. Natl. Acad. Sci. U. S. A. 109, 12455–12460 Fe–S protein biogenesis by the CIA targeting complex. Nat.
48. Makarova, A.V. et al. (2012) A four-subunit DNA polymerase ζ com- Struct. Mol. Biol. 27, 735–742
plex containing Pol δ accessory subunits is essential for PCNA- 74. Bartels, P.L. et al. (2017) A redox role for the [4Fe4S] cluster of
mediated mutagenesis. Nucleic Acids Res. 40, 11618–11626 yeast DNA polymerase δ. J. Am. Chem. Soc. 139, 18339–18348
49. Hirota, K. et al. (2015) The POLD3 subunit of DNA polymerase 75. Lydeard, J.R. et al. (2007) Break-induced replication and
δ can promote translesion synthesis independently of DNA po- telomerase-independent telomere maintenance require
lymerase ζ. Nucleic Acids Res. 43, 1671–1683 Pol32. Nature 448, 820–823
50. Siebler, H.M. et al. (2014) A novel variant of DNA polymerase ζ, 76. Garcia, J.S. et al. (2004) The C-terminal zinc finger of the cata-
Rev3δC, highlights differential regulation of Pol32 as a subunit lytic subunit of DNA polymerase δ is responsible for direct inter-
of polymerase δ versus ζ in Saccharomyces cerevisiae. DNA action with the B-subunit. Nucleic Acids Res. 32, 3005–3016
Repair (Amst) 24, 138–149 77. Conde, C.D. et al. (2019) Polymerase δ deficiency causes
51. Hirota, K. et al. (2016) In vivo evidence for translesion synthesis syndromic immunodeficiency with replicative stress. J. Clin.
by the replicative DNA polymerase δ. Nucleic Acids Res. 44, Invest. 129, 4194–4206
7242–7250 78. Cui, Y. et al. (2020) Combined immunodeficiency caused by a
52. Lee, Y.S. et al. (2014) Human Pol ζ purified with accessory sub- loss-of-function mutation in DNA polymerase delta 1. J. Allergy
units is active in translesion DNA synthesis and complements Pol Clin. Immunol. 145, 391–401 e8
η in cisplatin bypass. Proc. Natl. Acad. Sci. U. S. A. 111, 79. Ji, J. et al. (2019) The processivity factor Pol32 mediates nu-
2954–2959 clear localization of DNA polymerase delta and prevents chro-
53. Shachar, S. et al. (2009) Two-polymerase mechanisms dictate mosomal fragile site formation in Drosophila development.
error-free and error-prone translesion DNA synthesis in PLoS Genet. 15, e1008169
mammals. EMBO J. 28, 383–393 80. Murga, M. et al. (2016) POLD3 is haploinsufficient for DNA rep-
54. Sharma, S. et al. (2012) DNA polymerase ζ is a major determi- lication in mice. Mol. Cell 63, 877–883
nant of resistance to platinum-based chemotherapeutic 81. Tumini, E. et al. (2016) Roles of human POLD1 and POLD3 in
agents. Mol. Pharmacol. 81, 778–787 genome stability. Sci. Rep. 6, 38873
55. Wang, W. et al. (2015) REV3L modulates cisplatin sensitivity of non- 82. Shimada, K. et al. (2020) Uncoordinated long-patch base excision
small cell lung cancer H1299 cells. Oncol. Rep. 34, 1460–1468 repair at juxtaposed DNA lesions generates a lethal accumulation
56. Brocas, C. et al. (2010) Stable interactions between DNA of double-strand breaks. bioRxiv Published online November 15,
polymerase δ catalytic and structural subunits are essential 2020. https://doi.org/10.1101/2020.11.15.383513
for efficient DNA repair. DNA Repair (Amst) 9, 1098–1111 83. Weedon, M.N. et al. (2013) An in-frame deletion at the poly-
57. Minocherhomji, S. et al. (2015) Replication stress activates merase active site of POLD1 causes a multisystem disorder
DNA repair synthesis in mitosis. Nature 528, 286–290 with lipodystrophy. Nat. Genet. 45, 947–950
58. Bhowmick, R. et al. (2016) RAD52 facilitates mitotic DNA syn- 84. Pelosini, C. et al. (2014) Identification of a novel mutation in the
thesis following replication stress. Mol. Cell 64, 1117–1126 polymerase delta 1 (POLD1) gene in a lipodystrophic patient af-
59. Reinhold, W.C. et al. (2012) CellMiner: a web-based suite of ge- fected by mandibular hypoplasia, deafness, progeroid features
nomic and pharmacologic tools to explore transcript and drug (MDPL) syndrome. Metabolism. 63, 1385–1389
patterns in the NCI-60 cell line set. Cancer Res. 72, 3499–3511 85. Lessel, D. et al. (2015) POLD1 germline mutations in patients
60. Costantino, L. et al. (2014) Break-induced replication repair of initially diagnosed with Werner syndrome. Hum. Mutat. 36,
damaged forks induces genomic duplications in human cells. 1070–1079
Science 343, 88–91 86. Palles, C. et al. (2013) Germline mutations affecting the proof-
61. Layer, J.V. et al. (2020) Polymerase δ promotes chromosomal reading domains of POLE and POLD1 predispose to colorectal
rearrangements and imprecise double-strand break repair. adenomas and carcinomas. Nat. Genet. 45, 136–143
117 pp. 27566–27577 87. Rayner, E. et al. (2016) A panoply of errors: polymerase proof-
62. Lindahl, T. (1993) Instability and decay of the primary structure reading domain mutations in cancer. Nat. Rev. Cancer 16,
of DNA. Nature 362, 709–715 71–81
63. Lindahl, T. and Barnes, D.E. (2000) Repair of endogenous DNA 88. Dunlop, M.G. et al. (2012) Common variation near CDKN1A,
damage. Cold Spring Harb. Symp. Quant. Biol. 65, 127–133 POLD3 and SHROOM2 influences colorectal cancer risk.
64. Prindle, M.J. and Loeb, L.A. (2012) DNA polymerase delta in DNA Nat. Genet. 44, 770–776
replication and genome maintenance. Environ. Mol. Mutagen. 53, 89. Chae, Y.K. et al. (2016) Genomic landscape of DNA repair
666–682 genes in cancer. Oncotarget 7, 23312–23321
65. Tubbs, A. and Nussenzweig, A. (2017) Endogenous DNA 90. Nicolas, E. et al. (2016) POLD1: central mediator of DNA
damage as a source of genomic instability in cancer. Cell replication and repair, and implication in cancer and other
168, 644–656 pathologies. Gene 590, 128–141

486 Trends in Genetics, May 2021, Vol. 37, No. 5


Trends in Genetics
OPEN ACCESS

91. Zhang, J. et al. (2020) Inactivating mutations in exonuclease the Sp1-induced DNMT1 activities in breast cancer. Onco Targets
and polymerase domains in DNA polymerase delta alter sensi- Ther. 9, 1351–1360
tivities to inhibitors of dNTP synthesis. DNA Cell Biol. 39, 50–56 99. Qin, Q. et al. (2018) Elevated expression of POLD1 is associ-
92. Venkatesan, R.N. et al. (2007) Mutation at the polymerase active ated with poor prognosis in breast cancer. Oncol. Lett. 16,
site of mouse DNA polymerase δ increases genomic instability 5591–5598
and accelerates tumorigenesis. Mol. Cell. Biol. 27, 7669–7682 100. Kadota, M. et al. (2009) Identification of novel gene amplifica-
93. Goldsby, R.E. et al. (2001) Defective DNA polymerase-δ proof- tions in breast cancer and coexistence of gene amplification
reading causes cancer susceptibility in mice. Nat. Med. 7, with an activating mutation of PIK3CA. Cancer Res. 69,
638–639 7357–7365
94. Albertson, T.M. et al. (2009) DNA polymerase ε and δ proof- 101. Ren, L. et al. (2017) Potential biomarkers of DNA replication
reading suppress discrete mutator and cancer phenotypes in stress in cancer. Oncotarget 8, 36996–37008
mice. Proc. Natl. Acad. Sci. U. S. A. 106, 17101–17104 102. Hocke, S. et al. (2016) A synthetic lethal screen identifies ATR-
95. Elgaaen, B.V. et al. (2010) POLD2 and KSP37 (FGFBP2) inhibition as a novel therapeutic approach for POLD1-deficient
correlate strongly with histology, stage and outcome in ovarian cancers. Oncotarget 7, 7080–7095
carcinomas. PLoS One 5, e13837 103. Barbari, S.R. and Shcherbakova, P.V. (2017) Replicative DNA
96. Olstad, O.K. et al. (2003) Molecular heterogeneity in human polymerase defects in human cancers: consequences, mecha-
osteosarcoma demonstrated by enriched mRNAs isolated by nisms, and implications for therapy. DNA Repair 56, 16–25
directional tag PCR subtraction cloning. Anticancer Res. 23, 104. Berman, H.M. et al. (2000) The Protein Data Bank. Nucleic
2201–2216 Acids Res. 28, 235–242
97. Sanefuji, K. et al. (2010) Significance of DNA polymerase delta 105. Hughes, P. et al. (1999) Isolation and identification of the third
catalytic subunit p125 induced by mutant p53 in the invasive subunit of mammalian DNA polymerase δ by PCNA-affinity
potential of human hepatocellular carcinoma. Oncology 79, chromatography of mouse FM3A cell extracts. Nucleic Acids
229–237 Res. 27, 2108–2114
98. Zhang, L. et al. (2016) p53 inhibits the expression of p125 and 106. Liu, L. et al. (2000) Identification of a fourth subunit of mamma-
the methylation of POLD1 gene promoter by downregulating lian DNA polymerase δ. J. Biol. Chem. 275, 18739–18744

Trends in Genetics, May 2021, Vol. 37, No. 5 487

You might also like