*For correspondence: cejka@
imcr.uzh.ch
Competing interests: The
authors declare that no
competing interests exist.
Funding: See page 21
Received: 07 June 2016
Accepted: 08 September 2016
Published: 09 September 2016
Reviewing editor: Antoine M
van Oijen, University of
Wollongong, Australia
Copyright Pinto et al. This
article is distributed under the
terms of the
Creative Commons
Attribution License,
which
permits unrestricted use and
redistribution provided that the
original author and source are
credited.
Human DNA2 possesses a cryptic DNA
unwinding activity that functionally
integrates with BLM or WRN helicases
Cosimo Pinto
1
, Kristina Kasaciunaite
2
, Ralf Seidel
2
, Petr Cejka
1
*
1
Institute of Molecular Cancer Research, University of Zurich, Zurich, Switzerland;
2
Institute of Experimental Physics I, University of Leipzig, Leipzig, Germany
Abstract Human DNA2 (hDNA2) contains both a helicase and a nuclease domain within the
same polypeptide. The nuclease of hDNA2 is involved in a variety of DNA metabolic processes.
Little is known about the role of the hDNA2 helicase. Using bulk and single-molecule approaches,
we show that hDNA2 is a processive helicase capable of unwinding kilobases of dsDNA in length.
The nuclease activity prevents the engagement of the helicase by competing for the same
substrate, hence prominent DNA unwinding by hDNA2 alone can only be observed using the
nuclease-deficient variant. We show that the helicase of hDNA2 functionally integrates with BLM or
WRN helicases to promote dsDNA degradation by forming a heterodimeric molecular machine.
This collectively suggests that the hDNA2 motor promotes the enzyme’s capacity to degrade
dsDNA in conjunction with BLM or WRN and thus promote the repair of broken DNA.
DOI: 10.7554/eLife.18574.001
Introduction
DNA replication, repair and recombination require the function of multiple DNA helicases and nucle-
ases (
Tsutakawa et al., 2014; Wu and Hickson, 2006). The DNA replication ATP-dependent heli-
case/nuclease 2 (DNA2) is an enzyme that contains both helicase and nuclease domains within the
same polypeptide (
Bae et al., 1998), and has important functions in a variety of DNA metabolic pro-
cesses. Dna2 was first described in Saccharomyces cerevisiae where it is required for DNA replica-
tion under unperturbed conditions (
Budd and Campbell, 1995; Kuo et al., 1983). Specifically,
during Okazaki fragment processing, yeast Dna2 (yDna2) cleaves long 5’-flaps that are coated by the
Replication Protein A (RPA) and are therefore refractory to cleavage by Rad27 (FEN1)
(
Bae et al., 2001; Levikova and Cejka, 2015). Moreover, yDna2 is one of the nucleases that resect
5’-terminated strands of DNA double-strand breaks (DSBs) (
Cejka et al., 2010; Niu et al., 2010;
Zhu et al., 2008). This process leads to the formation of 3’-tailed DNA, which becomes a substrate
for the strand exchange protein Rad51 to initiate homology search and accurate DSB repair by the
recombination machinery (
Cejka, 2015; Heyer et al., 2010; Symington, 2014). Yeast Dna2 also
functions upon replication stress to degrades structures such as reversed replication forks (
Hu et al.,
2012
; Thangavel et al., 2015) and has a structural role in DNA damage signaling, where it is a com-
ponent in one out of three signaling branches that activate the Mec1 kinase in response to ssDNA in
S-phase (
Kumar and Burgers, 2013). Additionally, yDna2 was described to be required for the
proper function of telomeres (
Choe et al., 2002). In contrast to Okazaki fragment processing and
DNA end resection, the involvement of yDna2 in these latter DNA metabolic processes is poorly
understood. The yeast Dna2 protein contains a large unstructured N-terminal domain, which medi-
ates a physical interaction with yRPA (
Bae et al., 2003), is required for Dna2’s checkpoint function
(Kumar and Burgers, 2013) and its capacity to melt secondary structures within 5’ DNA flaps
(
Lee et al., 2013). The N-terminal domain is followed by a RecB-like nuclease domain
Pinto et al. eLife 2016;5:e18574. DOI: 10.7554/eLife.18574 1 of 24
RESEARCH ARTICLE
(Budd et al., 2000) and a Superfamily I helicase domain in the C-terminal part of the polypeptide
(
Budd and Campbell, 1995). With the exception of checkpoint signaling, all Dna2 functions are
exclusively dependent on its nuclease activity (
Sturzenegger et al., 2014; Thangavel et al., 2015;
Wanrooij and Burgers, 2015; Zhu et al., 2008). Dna2 homologs are present in all eukaryotic organ-
isms including human cells (
Budd and Campbell, 1995; Eki et al., 1996; Gould et al., 1998). Both
helicase and nuclease domains are well conserved in evolution, but the unstructured N-terminal
domain is only present in lower eukaryotes (
Bae et al., 1998; Kang et al., 2010; Wanrooij and Bur-
gers, 2015
).
Human DNA2 (hDNA2) also functions in DNA end resection (
Gravel et al., 2008;
Nimonkar et al., 2011; Sturzenegger et al., 2014) and in the processing of non-canonical DNA rep-
lication structures, such as reversed replication forks upon replication stress (
Duxin et al., 2012;
Thangavel et al., 2015). In contrast to yeast, however, hDNA2 appears to be dispensable for the
processing of most Okazaki fragments (
Duxin et al., 2012). Specific inactivation of the nuclease, as
well as the depletion or knockout of the protein/gene, result in lethal phenotypes in all organisms
tested to date (
Budd et al., 2000; Duxin et al., 2012; Kang et al., 2000; Lin et al., 2013). In yeast,
this has been ascribed to yDna2’s role in Okazaki fragment processing (
Kang et al., 2010). Human
DNA2-depleted cells arrest at late S/G2 phase of the cell cycle (
Duxin et al., 2012). The nature of
DNA intermediates that require the processing by hDNA2 is still rather elusive. It is conceivable that
the lethality of hDNA2-depleted cells results from the failure to process reversed replication forks or
other aberrant structures that arise during replication stress even in the absence of treatment with
genotoxic drugs (
Duxin et al., 2012; Thangavel et al., 2015). The role of hDNA2 in DSB end resec-
tion in contrast does not appear to be essential for viability as it functions redundantly with another
nuclease, Exonuclease 1 (EXO1) (
Gravel et al., 2008; Nimonkar et al., 2011; Tomimatsu et al.,
2012
). EXO1 is not involved in the processing of reversed replication forks, pointing towards an
essential function of hDNA2 in the response to intermediates arising during DNA replication
(
Thangavel et al., 2015).
The nuclease of hDNA2 is specific for ssDNA (
Kim et al., 2006; Masuda-Sasa et al., 2006) and
therefore requires an associated helicase activity to resect/degrade dsDNA. This was shown to be
either BLM or WRN during DSB end resection (
Gravel et al., 2008; Nimonkar et al., 2011;
Sturzenegger et al., 2014), or primarily WRN to degrade non-canonical DNA structures arising dur-
ing DNA replication (
Thangavel et al., 2015). Interestingly, the inherent helicase of hDNA2 was not
required for these processes (
Sturzenegger et al., 2014; Thangavel et al., 2015), and the function
of the hDNA2 motor activity remains unclear. The helicase function is not essential for viability in
yeast (
Bae et al., 2002), where it was proposed to unwind secondary structures forming on long
flaps at the 5’ ends of Okazaki fragments (
Lee et al., 2013). Yeast dna2 cells lacking the helicase
activity are dramatically sensitive to alkylating agents such as methyl methanesulfonate (MMS)
(
Budd and Campbell, 1995), suggesting that the yDna2 helicase might also play a role in the
response to replication stress. In contrast to yeast, both helicase and nuclease functions are essential
for viability in human cells (
Duxin et al., 2012). Similarly to hDNA2 nuclease-deficient cells, hDNA2
helicase-deficient cells also exhibit a terminal S/G2 cell cycle arrest, most likely due to the inability to
resolve structures arising in S-phase (
Duxin et al., 2012). Furthermore, hDNA2 nuclease-deficient
cells displayed cell cycle defects that were even more severe than upon depletion of hDNA2; inter-
estingly, this phenotype was dependent on the integrity of the Walker A motif within the helicase
domain (
Duxin et al., 2012). This suggested that the hDNA2 helicase performs essential functions
during DNA replication, yet it becomes toxic in the absence of the nuclease (
Duxin et al., 2012),
although mechanistic insights into the interplay between both activities have been lacking. There-
fore, it remains to be determined how the hDNA2 helicase contributes to the overall function of the
polypeptide.
The clear requirement for the helicase of hDNA2 for the viability of human cells (
Duxin et al.,
2012
) stands in contrast to the inconclusive reports regarding the capacity of the human recombi-
nant hDNA2 polypeptide to unwind dsDNA. One work concluded that hDNA2 lacks a helicase activ-
ity (
Kim et al., 2006 ), whereas another study could detect DNA unwinding, albeit very weak and
distributive (
Masuda-Sasa et al., 2006). It has been also proposed that the helicase domain may be
more responsible for DNA binding rather than as a motor activity per se (
Zhou et al., 2015). Here
we present that hDNA2 possesses a processive helicase activity capable of unwinding dsDNA of sev-
eral kilobases in length. Paradoxically, the helicase is cryptic and becomes detectable only upon
Pinto et al. eLife 2016;5:e18574. DOI: 10.7554/eLife.18574 2 of 24
Research article Biochemistry Genes and Chromosomes
inactivation of the nuclease. This explains the more pronounced phenotypes of the hDNA2 nuclease-
deficient cells as opposed to double nuclease- and helicase-deficient cells or depletions of the poly-
peptide (
Duxin et al., 2012). Finally, we show that the helicase of hDNA2 contributes to dsDNA
degradation in complex with Bloom syndrome protein (BLM) or Werner syndrome protein (WRN)
helicases, and may play a supporting role in the resection of DSBs or other aberrant structures aris-
ing during DNA replication. The motor activities within hDNA2 and BLM/WRN function in a synergis-
tic manner, and the stimulatory effect observed with the hDNA2-WRN and hDNA2-BLM pairs is
highly specific. This shows that the hDNA2-BLM and hDNA2-WRN complexes are functionally more
integrated molecular machines than previously thought.
Results
Expression and purification of human DNA2
Human DNA2 was prepared using a construct, which contained an N-terminal 6x-histidine and a
C-terminal FLAG affinity tags (
Figure 1A). The sequence of hDNA2 was codon-optimized
(
Supplementary file 1A) for the expression in Spodoptera frugiperda 9 (Sf9) cells, which improved
the yield ~2–3 fold (data not shown). Considering that hDNA2 contains an iron-sulfur cluster
(
Pokharel and Campbell, 2012; Yeeles et al., 2009), all buffers were degassed and contained
reducing agents throughout the preparation procedure to prevent oxidation of the cluster, as
described previously for S. cerevisiae Dna2 (
Levikova et al., 2013). Wild type hDNA2, nuclease-defi-
cient D277A, helicase-deficient K654R as well as nuclease- and helicase-deficient D277A K654R var-
iants were purified in the same manner to near homogeneity (
Figure 1B and Figure 1—figure
supplement 1A–C
). The yield of the recombinant proteins was ~330–390 mg from 3 liters of Sf9 cell
culture except for the variant containing the K654R mutation, which yielded only ~27 mg.
hDNA2 preferentially degrades 5’-tailed DNA in the presence of RPA
Human DNA2 is known to possess ssDNA-specific nuclease activity (
Kim et al., 2006; Masuda-
Sasa et al., 2006
). Considering that hDNA2 performs multiple functions during DNA metabolism,
we set out to analyze the preference of its nuclease activity using various oligonucleotide-based
DNA structures. Without the human Replication Protein A (hRPA), hDNA2 most efficiently degraded
ssDNA, while 5’-overhanged, 3’-overhanged and Y-structured DNA were degraded ~7–20-fold less
efficiently, based on the hDNA2 concentration required for the degradation of 50% DNA substrate
(
Figure 1C and Figure 1—figure supplement 1D). In contrast, dsDNA was largely refractory to
cleavage (
Figure 1C and Figure 1—figure supplement 1D), in agreement with the observations
that hDNA2 needs a helicase partner in DNA end resection to initiate homologous recombination
(
Cejka et al., 2010; Gravel et al., 2008; Nimonkar et al., 2011; Sturzenegger et al., 2014;
Zhu et al., 2008). As reported previously (Masuda-Sasa et al., 2006), the helicase-deficient hDNA2
(K654R) variant displayed a nuclease activity indistinguishable from that of the wild type enzyme on
a 5’-tailed DNA substrate (
Figure 1—figure supplement 1D–F). Using a 3’-end labeled ssDNA, we
observed that hRPA directs the nuclease of hDNA2 towards the 5’ terminus; while at the same time
inhibits the 3’-5’ nuclease activity (
Figure 1D). This is in agreement with previous observations in var-
ious organisms (
Cejka et al., 2010 ; Nimonkar et al., 2011; Zhou et al., 2015 ) and explains how
hRPA enforces the correct polarity of DNA degradation during DNA end resection. Interestingly, in
the presence of hRPA, hDNA2 most efficiently cleaved Y-structured and 5’-tailed DNA substrates,
which were degraded ~5–10-fold more efficiently than ssDNA (
Figure 1E and Figure 1—figure sup-
plement 1G
). In summary, the nuclease activities of yeast Dna2 and human DNA2 are very similar
qualitatively, but human DNA2 appears somewhat less active (~2-fold in degradation of 5’-tailed
DNA) than its yeast homologue (
Levikova et al., 2013).
The nuclease-deficient hDNA2 D277A was subsequently used to determine DNA binding prefer-
ence. hDNA2 D277A strongly bound ssDNA, with K
D
~2 nM for ssDNA of 50 nucleotides in length
(
Figure 1F,G). Similar binding affinity was observed for Y-structured DNA, while the apparent DNA
binding to 5’ and 3’-tailed structures was reduced ~8–12-fold, respectively, compared to ssDNA. In
contrast, dsDNA was bound very poorly (
Figure 1G and Figure 1—figure supplement 2A–D). Fur-
ther experiments revealed that the DNA binding affinity was determined by the length of ssDNA
rather than the specific structure (
Figure 1H,I and Figure 1—figure supplement 2A–F).
Pinto et al. eLife 2016;5:e18574. DOI: 10.7554/eLife.18574 3 of 24
Research article Biochemistry Genes and Chromosomes
Figure 1. Human DNA2 preferentially binds and degrades 5’ terminated ssDNA. (A) A schematic representation of
the recombinant hDNA2 protein used in this study. The polypeptide contains an N-terminal 6xHis- and a
C-terminal FLAG affinity tag. The positions of the mutations inactivating the nuclease (D277A) activity or the
helicase (K654R) activity are indicated. (B) A 10% polyacrylamide gel stained with Coomassie blue showing
fractions from a representative purification of hDNA2 D277A. (C) Quantitation of hDNA2 nuclease activity on
various DNA substrates in the absence of hRPA from experiments such as shown in
Figure 1—figure supplement
1D. Averages shown, n = 2; error bars, SEM. (D) Human DNA2 (0.2 nM) was incubated with ssDNA
32
P-labeled at
Figure 1 continued on next page
Pinto et al. eLife 2016;5:e18574. DOI: 10.7554/eLife.18574 4 of 24
Research article Biochemistry Genes and Chromosomes
Interestingly, the hDNA2-bound DNA species either entered the polyacrylamide gels during electro-
phoresis, or remained stuck in the wells, indicative of a multiprotein-DNA complex and most likely a
non-specific aggregate. Remarkably, the distinct DNA-protein species that entered the polyacryl-
amide gel were only observed with substrates containing a free 5’ end such as Y-structured, 5’ tailed
or ssDNA substrates (Figure 1F and Figure 1—figure supplement 2A,B,F), suggesting that hDNA2
exhibits a preference for this structure even in the absence of hRPA. This likely reflects its role in 5’
DNA end degradation in various metabolic processes (
Kang et al., 2010; Nimonkar et al., 2011;
Sturzenegger et al., 2014; Zheng et al., 2008).
hDNA2 shows DNA structure-dependent ATPase activity
Previous reports concluded that hDNA2 hydrolyses ATP, as expected from a protein containing an
SFI helicase domain (
Budd and Campbell, 1995). We next determined the ATP hydrolysis rate of
nuclease-deficient hDNA2 D277A in the presence of various DNA structures. The ATPase activity
was strongly enhanced in the presence the DNA cofactors. The greatest stimulation, ~13-fold, was
observed with 5’-tailed DNA (
Figure 2A), in agreement with the 5’-3’ polarity of the hDNA2 helicase
(
Balakrishnan et al., 2010a; Balakrishnan et al., 2010b; Masuda-Sasa et al., 2006). The apparent
turnover rate (k
cat
) of the ATP hydrolysis in the presence of 5’-tailed substrates of different lengths
was 6.9 ± 1.1 s
1
and 6.2 ± 0.9 s
1
. In contrast, dsDNA stimulated the hDNA2 ATPase to the lowest
extent, ~four-fold, compared to reactions without DNA. Next we performed the ATPase assays in
the presence of various amounts of poly(dT) DNA, which is a ssDNA devoid of any secondary struc-
ture. As expected, the ATP hydrolysis rate increased with poly(dT) concentration. The measured
reaction rate values were fitted into a Michaelis-Menten curve with V
max
= 3.1 ± 0.3 mMmin
1
and
K
M
= 115 ± 45 nM (in nucleotides, Figure 2B), which corresponds to k
cat
= 4.3 ± 0.4 s
1
. The nucle-
ase-deficient DNA2 D277A variant was used for the above assays, as the nuclease of wild type
DNA2 interferes with its capacity to hydrolyze ATP by degrading DNA that serves as a co-factor of
the ATPase activity. As demonstrated in
Figure 2C, the rate of ATP hydrolysis by the nuclease-defi-
cient D277A variant incubated with 5’-tailed substrate was constant over time. In contrast the rate of
ATP consumption decreased quickly in case of wild type DNA2 (
Figure 2C). We believe that the
nuclease activity of hDNA2 rapidly degrades the 5’ ssDNA overhang, producing a substrate that is
less efficient as a cofactor for the ATPase activity. Very similar behavior was previously observed
with yeast Dna2 (
Levikova et al., 2013). Collectively, these experiments establish that the ATPase
activity of hDNA2 qualitatively resembles that of the yeast Dna2 homologue in terms of DNA sub-
strate preference and interplay with the nuclease activity, but it is ~10-fold less active in quantitative
terms (
Levikova et al., 2013).
Figure 1 continued
its 3’ end and various concentrations of hRPA. The panel shows a representative denaturing 20% polyacrylamide
gel. The blue triangle indicates a truncation of the substrate. (E) Quantitation of hDNA2 nuclease activity on
various DNA substrates in the presence of hRPA (15 nM) from experiments such as shown in
Figure 1—figure
supplement 1G
. Averages shown, n = 2; error bars, SEM. (F) A representative 6% polyacrylamide gel showing the
binding of hDNA2 D277A to ssDNA of 50 nt in length. The blue triangle indicates the position of the wells. (G)
Quantitation of DNA binding from experiments such as shown in
Figure 1F and Figure 1—figure supplement
2A–D
. Averages shown, n = 2–3; error bars, SEM. (H) DNA binding and its dependence on the length of ssDNA.
Quantitation is based on experiments such as shown in
Figure 1F and Figure 1—figure supplement 2E. Long
ssDNA was more efficiently bound by hDNA2. Averages shown, n = 2–3, error bars, SEM. (I) DNA binding and its
dependence on the length of 5’ single-stranded DNA overhang. Quantitation is based on experiments such as
shown in Figure 1—figure supplement 2B,F. Averages shown, n = 3; error bars, SEM.
DOI: 10.7554/eLife.18574.002
The following figure supplements are available for figure 1:
Figure supplement 1. Human RPA guides the hDNA2 nuclease to 5’ terminated ssDNA.
DOI: 10.7554/eLife.18574.003
Figure supplement 2. hDNA2 binds ssDNA.
DOI: 10.7554/eLife.18574.004
Pinto et al. eLife 2016;5:e18574. DOI: 10.7554/eLife.18574 5 of 24
Research article Biochemistry Genes and Chromosomes