University of Southern California Dissertations and Theses

Mechanisms of nucleases in non-homologous DNA end joining

(USC Thesis Other)

Mechanisms of nucleases in non-homologous DNA end joining

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Mechanisms of Nucleases in Non-Homologous DNA End Joining

By
Sicong Li
______________________________________________
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Genetics, Molecular, and Cellular Biology)

Table of Contents
List of Figures
Abstract 1
Chapter 1 General Introduction 4
1.1 Double –Stranded DNA breaks and Non-homologous DNA end joining. 4
1.2 DNA nucleases and repair 7
1.3 DNA DSB and Lymphoma, perspectives on inhibiting NHEJ. 12
Chapter 2 PNK, Aprataxin Like Factor 14
2.1 Abstract 14
2.2 Introduction 14
2.3 Method 15
2.4 Results 18
2.5 Discussion 29

Chapter 3 Single Stranded Endonuclease Activity of Artemis 31
3.1 Abstract 31
3.2 Introduction 31
3.3 Method 33
3.4 Results 35
3.5 Discussion 47

Ch ap te r 4 5 ’ e x o n u cleas e a ctiv it y o f Art e m is 51
4.1 Abstract 51
4.2 Introduction 51
4.3 Method 54
4.4 Results 56
4.5 Discussion 65

Chapter 5 High-throughput Screen for Artemis Inhibitor 69
5.1 Abstract 69
5.2 Introduction 69
5.3 Method 71
5.4 Results 73
5.5 Discussion 80

Chapter 6 Concluding Comments 83

Acknowledgements 84
References 85

List of Figures

Figure 1.1 Schematic of Artemis Protein 10
Figure 1.2 PALF and PALF Binding Partners 12
Figure 2.1 Purification of PALF 19
Figure 2.2 3 ’ t o 5 ’ E xo n u cl ease A ctiv i t y o f P A L F o n Sin g le -Stranded DNA 21
Figure 2.3 Endonuclease Activity of PALF on Overhangs 23
Figure 2.4 Lack of Endonuclease Activity of PALF on Hairpin Substrates 25
Figure 2.5 Effect of Ku, XRCC4-DNA Ligase IV, and DNA-PKcs on PALF Nuclease Activity 26
Figure 2.6 PALF Can Cooperate with Ku and XRCC4-DNA Ligase IV in Double-Stranded 28
DNA End Ligation
Figure 3.1 Size Exclusion Chromatography of Purified Artemis 37
Figure 3.2 Endonuclease Activity of Artemis on Single-Stranded DNA 39
Figure 3.3 Time Course of endonuclease activity of Artemis and DNA-PKcs on 40
single-stranded DNA
Figure 3.4 Autophosphorylation of DNA-PKcs Regulates the Single-stranded DNA 43
Endonuclease Activity of Artemis
Figure 3.5 Anti-Artemis antibody down-regulates the single-stranded DNA 45
endonuclease activity of Artemis in the presence of DNA-PKcs.
Figure 3.6 Artemis single-stranded DNA endonuclease activities with different 47
divalent cations.
Figure 3.7 Artemis Endonuclease Activity on Single-stranded DNA and 49
Double-stranded DNA.
Figure 4.1 Size exclusion chromatography of a single amino acid point mutant of 57
Artemis, the ARM14 mutant
Figure 4.2 End o n u clease and 5 ’ e x o n u clease acti v ity o f t h e Ar te m is point mutant, 60
ARM14.
Figure 4.3 Small molecule inhibitors block both the 5' exonuclease and the 62

endonucleolytic activities of Artemis
Figure 5.1 The Human Acute Lymphoblastic Lymphoma Cell Line, Nalm-6, 70
is Hypersensitive to Etoposides When Artemis is Absent.
Figure 5.2 Inhibitors of Artemis Would Result in Double-Strand DNA Breaks in 71
Early Lymphoid Cells, in Acute Lymphoblastic Lymphoma (ALL) Cells,
and in any Somatic Cells Challenged with Double-Strand Break Inducing Agents.
Figure 5.3 Fluorescent DNA Substrate Design 74
Figure 5.4 Kinetics of Artemis Activity in 1536 Well Plate Format. 75
Figure 5.5 Hit Result, Z score, and Signal to Noise Ratio of MLPCN Screen. 76
Figure 5.6 Artemis Nuclease Activity In the Presence of Small Molecules identified 77
by MLPCN Screen
Figure 5.7 Effect of inhibitors on Artemis when using hairpin DNA and Single-stranded 78
DNA.
Figure 5.8 Effects of Artemis Specific Inhibitors on Mung Bean Nuclease. 79

1

Abstract
DNA double-stranded breaks (DSB) can occur through programmed mechanisms such as V(D)J
recombination and class switch recombination or pathological mechanisms such as ionizing radiation.
To maintain genomic stability and integrity it is essential for DSB to be repaired properly and efficiently.
Repair of DSB is predominantly carried out by the non-homologous end joining (NHEJ) pathway. Broken
DNA ends are recognized and bound by the Ku70/80 heterodimer, followed by DNA-PKcs binding,
processing by nucleases and polymerases, before being joined by the XRCC4-Ligase IV complex.
Unrepaired or improperly repaired DSB results in apoptosis, slowed cell growth, or translocation. Here
we characterize the nucleases that are involved in a subset of NHEJ as well as explore the implications of
inhibiting these nucleases for therapeutic purposes.
PNK Aprataxin like Factor (PALF also referred to as APLF) was first identified by searching
through protein libraries for proteins containing the FHA domain. PALF is able to bind to NHEJ factors,
Ku80 and XRCC4, as well as to ssDNA repair proteins, PARP1 and PARP3. It has been proposed that PALF
can function as a nuclease during DNA repair. We explore this possibility in a biochemical system using
p u rif ied P ALF pr o t ein . T h e d ata presen te d her e d e m o n strate s t h at P ALF ha s 3 ’ e xo n u cleas e ac tiv i ty independent of binding interaction with other NHEJ components. PALF also has endonuclease activity at
5 ’ and 3 ’ o v erha n g s. Ad d iti o n ally, in a reconstituted biochemical joining system, PALF is able to
participate in ligation by functioning to process incompatible DNA ends. In a cellular system, small
interfering RNA targeted against PALF slows repair kinetics after treatment with ionizing radiation.
Artemis is the only known mammalian nuclease which can open hairpin DNAs. This essential
feature of Artemis makes it indispensable for processing coding ends during V(D)J recombination.
Human and mice lacking Artemis protein suffer from immunodeficiency due to lack of T and B cells.
Data presented here explores the nucleolytic capabilities of Artemis on a single-stranded DNA. We

2

show that Artemis has low levels of nuclease activity on single-stranded DNA. Artemis preferentially
processes single-stranded pyrimidines and not purines. The single-stranded endonuclease activity of
Artemis is further stimulated by the addition of DNA-PKcs.
Previous data suggests tha t Arte m is ha s in trin sic 5 ’ e xo n u clease ac tiv i ty which elutes at the
same position as the intrinsic endonuclease activity across several different purification columns. We
wanted t o fur ther ev alu at e w h et h e r t h e 5 ’ e x o n u clase is in trin sic t o Ar te m is by creating a point mutant
of Artemis. The point mutant of Artemis, ARM14, has a single amino acid substitution at a conserved
histidine residue (H115). Purified ARM14 mutant, unlike wild type Artemis protein, does not have
endonuclease activity or 5 ’ exonuclease activity. CD spectrum demonstrates that purified ARM14
retains similar structural profile to wild-type Artemis in solution. Manipulation of the divalent cation
co n cent rati o n affec ts b o th the 5 ’ e x o n u clea se an d the end o n u cleas e wi th si m il ar b eh avior.
Furthermore, we show that several different selective chemicals are able to inhibit both the
endonuclease as well as the exonuclease activity of Artemis. We conclude that wild-type Artemis has
in trin sic 5 ’ ex o n u cleas e ac ti v ity w h ich utilizes the same catalytic site as its endonuclease activity.
To explore the possibility of inhibiting Artemis as a potential treatment for acute lymphoblastic
leukemia (ALL), we utilized a high-throughput screen to search for small molecule inhibitors of Artemis.
In a biochemical reaction, purified Artemis is added to a fluorescent substrate and the reaction is
miniaturized to a 4 to 6ul volume format to be compatible with conventional HTS screening. In
preliminary assay optimization, we were able to achieve a robust HTS, based on standard HTS criteria (Z
scores and signal to noise ratios in a 1536 well assay plate format). We carried out a small scale HTS
using 31,000 compounds from an NIH library. Of the 31,000 compounds screened, we observed a hit
rate of 0.59% when using a threshold of inhibition greater than 30%. Three of the molecules with the
mosts potent IC50 and greatest efficacy were further tested in a secondary gel based assay and counter-

3

screened against mung bean nuclease. From this initial screen we have identified 2 compounds that are
able to inhibit Artemis in our testing scheme.

4

Chapter 1 General Introduction
1.1 Double-Stranded DNA Breaks and Non-homologous DNA End Joining.
DNA double stranded breaks (DSBs) occur naturally in cells through programmed mechanisms such
as V(D)J recombination or class switch recombination (CS) in lymphocytes. Double-strand breaks can
also occur pathologically by topoisomerase failure, reactive oxygen species (ROS), ionizing radiation, or
nicks in the DNA. In cell culture, 5-10% of dividing mammalian cells appears to have chromosomal
breaks at any given time. Failure to properly repair DSBs and maintain chromosomal integrity results in
loss of genetic information, cell death, or chromosomal translocation leading to cancer. Two distinct
pathways have evolved in eukaryotes to repair DSBs. The predominant pathway, nonhomologous DNA
end joining (NHEJ), is active during all phases of the cell cycle while homologous recombination (HR) is
only active during the late S phase and G2 phase of the cell cycle when a sister chromatid is present.
Unlike HR, NHEJ is an imprecise joining mechanism and does not maintain 100% sequence fidelity at the
point of the break. Local addition of nucleotides or loss of nucleotides may occur at the site of repaired
breaks. This imprecise joining is responsible for generating the vast array of antigen receptors. At the
same time, accumulation of local mutation over time may result in deleterious effects such as cancer
(Lieber, 2008). Homologous recombination retains perfect fidelity by using a sister chromatid as the
template DNA for repair. However, due to the requirement of a sister chromatid, HR is only significantly
active during late S and G2 phase; recent studies have shown that even during G2 phase, HR may only
account for 15% of DSB repairs(Beucher et al., 2009).
NHEJ joining of DSBs occurs through several distinct steps. The first step occurs when the DSBs
are recognized, followed by recruitment of core NHEJ proteins. The second step is an iterative process
involving the processing of broken DNA ends. At this step nucleotides can be added by polymerases or
deleted by nucleases prior to joining. Lastly, when the DNA ends are in a configuration amenable to
ligation, the XRCC4-Ligase IV complex is able to ligate the two broken DNA ends to resolve the break.

5

The broken DNA ends are first recognized by the Ku heterodimer, Ku70 and Ku80. The Ku
heterodimer is able to act as a scaffold to which other NHEJ protein can bind to. The initial Ku-DNA
complex is able to increase the binding affinity of subsequent proteins such as DNA-PKcs (West et al.,
1998).
During the iterative processing stage, the DNA ends are edited by nucleases such as the
Artemis/DNA-PKcs complex or by polymerases: mu, lambda, and TdT (lymphocyte specific). DNA-PKcs is
a protein kinase of the PIKK family. DNA-PKcs is able to phosphorylate other NHEJ proteins in vitro,
histone variant H2AX, as well as other DNA-PKcs molecules. Auto-phosphorylation of DNA-PKcs can
occur in cis or in trans. Persistence of DNA-PKcs at broken DNA ends is known to inhibit DNA joining in
vitro; it is thought that auto-phosphorylation in trans within the SQ/TQ residues of DNA-PKcs facilitates
the disassociation of DNA-PKcs from broken DNA ends (Helmink and Sleckman, 2012) . DNA-PKcs is able
to bind to DNA end with an affinity of 3x10
-9
M, which is increased 100 fold when Ku-DNA is present.
The binding of DNA-PKcs causes the Ku molecules to translocate inwards on the DNA end to
accommodate binding by additional NHEJ components.
Artemis is able to function as an endonuclease during DNA repair as well as a structure specific
nuclease to resolve hairpin structures. The nuclease activities of Artemis are dependent upon the
presence of DNA-PKcs. In addition to its role in processing broken DNA ends, Artemis is crucial in
opening hairpin ends generated by the RAG complex during V(D)J recombination in vertebrate B and T
early lymphocytes.
Three of the four Pol X family DNA polymerases participate in NHEJ and in V(D)J recombination.
Polymerase (pol) lambda, pol mu, and TdT are able to add additional nucleotides to broken DNA ends
before joining in a template-independent manner. Pol lambda and pol mu both have template-
dependent polymerase activity as well. TdT is purely template independent. TdT is present only in

6

developing lymphocytes and is responsible for the addition of N nucleotides to DNA ends; the addition
of N nucleotides helps to generate additional junctional diversity during V(D)J recombination. The
ability of polymerases to add additional nucleotides not only plays an important role in generating
antigen receptor diversity, the addition of nucleotides may also create microhomology at an otherwise
incompatible pair of DNA ends, allowing for more efficient joining by XRCC4-Ligase IV in the subsequent
step.
Mammalians have three distinct DNA Ligases: Ligase I, III, and IV. Ligase IV is the predominant
Ligase responsible for annealing broken DNA ends during NHEJ. NHEJ can occur without Ligase IV, albeit
at a much lower efficiency, indi cating t h at L ig as e II I m a y fun cti o n as a “ b ac k u p ” L i g ase in NH EJ in t h e absence of Ligase IV. In vivo, Ligase IV exists in a complex with the scaffold protein XRCC4. XRCC4 is
thought to aid in promoting the stability of Ligase IV. XRCC4-Ligase IV is able to ligate not only across
compatible ends with microhomology but also incompatible ends as well as across 1nt gaps (Gu et al.,
2007a). Interestingly, XRCC4-LigaseIV is also able to ligate single-stranded homopolymers of poly dT.
When annealing incompatible ends, the activity of XRCC4-LigIV is further stimulated by the addition of
the protein XLF (cernunnos) (Gu et al., 2007b). While XLF (cernunnos) is not part of the core NHEJ
machinery, it has been shown to stimulate ligation in vitro.
Inherited defects in NHEJ proteins are responsible for a variety of clinical states. Defects in NHEJ
account for roughly 15% of human severe combined immunodeficiency (SCID). No human patient
without Ku has currently been found, perhaps indicating the necessity of Ku. Mice deficient in Ku
heterodimer proteins are small in size, sterile, and lack lymphocytes. In contrast, mice without DNA-
PKcs develop normally but also lack lymphocytes (Rooney et al., 2004). Patients without DNA-PKcs or
Artemis accumulate unprocessed coding ends and suffer from ionizing radiation sensitivity as well as
SCID. Mice which lack XRCC4-Lig IV suffer from increased neuronal death. No human patient lacking

7

XRCC4-Lig IV has currently been identified, though hypomorphic (low expression) variants have been
identified. Patients without XLF suffer from a less severe form of radiation-sensitivity and SCID (RS-SCID)
than Artemis null or DNA-PKcs null patients; however, XLF null mice exhibit radiation-sensitivity and a
mild reduction in lymphocyte numbers but is able to carry out V(D)J recombination at wild-type levels.
(Li et al., 2008)

1.2 DNA Nucleases and Repair
Nucleolytic resection is often required to process broken DNA ends before polymerase filling and
ligation can occur. DNA nucleases function by cleaving the phosphodiester bond between a deoxyribose
residue an d t h e ph o sph at e b ackbo n e l ea v in g a 5 ’ ph o s p h ate g ro u p and a 3 ’ O H gr o u p as th e resul tin g products, in contrast to AP lyases, which leaves a 3’-physphoglycolyate group in the resulting product
(Marti and Fleck, 2004). Nucleases often require a set of divalent cations in the catalytic site to initiate
the cleavage reaction. Magnesium is found to be the most widely prevalent divalent cation
physiologically; however, zinc, manganese, and calcium are also known to be used under various
reaction conditions. DNA nucleases can be categorized into several broad categories based on their
mechanism of action. These categories include: 5’--> 3 ’ e xo n u cl ease s, 3 ’ --> 5 ’ e x o n u cleas es,
endonucleases, and structure specific nucleases. The exonucleases catalyze the cleavage of ssDNA in a
directional manner, starting with the terminal nucleotide at one end of the substrate, while
endonucleases are able to initiate DNA cleavage using any internal site. Structure-specific nucleases
such as FEN1 and Artemis are able to target specific DNA structures, such as flaps, cruciformss, hairpins,
bubble structures, or stem loops, depending on the specificity of each nuclease.
In double-stranded DNA break (DSB) repair, distinct sets of nucleases are used in homologous
recombination and non-homologous end joining. The HR pathway utilizes MRE11 and Sae2/CtIP for end

8

rese cti o n to g enerat e a 3 ’ t ail c ap ab le o f re co m b in ati o n (Mimitou and Symington, 2009). In NHEJ,
Artemis is required for processing a subset of DSB following ionizing radiation (IR) and resolving all of
the the DNA hairpins formed during V(D)J recombination. Other nucleases such as PALF may play an
important role during NHEJ to resolve a subset of DSBs. It is conceivable that there exist undiscovered
nucleases which may also participate in the NHEJ pathway, but given the broad range of nuclease
activities covered by Artemis and by PALF, any additional nucleases would seem to serve only a further
back-up role.
Artemis belongs to the metallo- β-lactamase family of DNA nucleases. The metallo- β-lactamase
family of proteins is cha rac te riz ed by a c o n ser v ed β - la ctam a se d o m ain and β -caspase domain. Members
of the family include nucleases known to be involved in DNA and RNA processing such as CPSF-73,
SNM1A, and Apollo. Artemis in complex with DNA-PKcs is essential in processing hairpins generated by
the RAG complex during VDJ recombination (Ma et al., 2002a). Patients without Artemis suffer from
severe combined immunodeficiency (SCID) and radiation sensitivity (RS). Artemis null patients show
complete B-cell differentiation arrest at the pre B-cell receptor check point (Noordzij et al., 2003).
Patients who are hypomorphic for Artemis expression have decreased level of circulating lymphocytes
with varying severity of SCID. The majority of Artemis mutations, which results in RS-SCID phenotype,
are locat ed w ith in t h e h ig h ly c o n ser v ed β -la cta m ase a n d β -caspase domain (Huang et al., 2009). These
mutations include point mutations as well as exon deletions caused by faulty homologous
recombination between the wild type Artemis gene (DCLRE1C ) and a “ p seud o - Ar te m i s” gene l o cat ed
61.2kb upstream of the wild type translation start site (Pannicke et al., 2010).
In cellular assays, it has been found that Artemis deficient cells are defective in V(D)J recombination
when assayed for coding joint formation (Rooney et al., 2003b). Signal joints occur at near normal levels
while coding joint formation is reduced 10 to 100-fold; this phenotype is recapitulated in DNA-PKcs null

9

cells (Harrington et al., 1992). In addition to coding joint defects, Artemis deficient cells are defective in
IR induced DNA repair as evidenced by the persistence of y-H2AX foci in IR treated cells that lack Artemis
(Riballo et al., 2004a). During homologous recombination in G2 phase of the cell cycle, Artemis deficient
cells are also compromised for RPA, Rad51, and ssDNA foci formation. This indicates that in addition to
NHEJ, Artemis may play a crucial role in a subset of HR during the G2 phase of the cell cycle (Beucher et
al., 2009).
Mechanistically, Artemis has been found to be able to process a variety of DNA substrate
configurations in vitro that have a single- to double-strand transition, specifically 5 ’ o v erha n g s, 3 ’ o v erha n g s, hai rp in st ru c tures, 5 ’ flap s, s y m m et ric al b u b b les , and stem loop structures (Ma et al., 2005c).
DSBs cau s ed by I R o fte n r e sul t in bro k en e n d s with a 3 ’ ph o sph o g lyc o lat e g ro u p which m u s t b e r e m o v ed
in order for polymerase filling and ligation. Artemis in complex with DNA-PKcs is able to satisfy this
requirement by removing t h e 3 ’ -PG group while leaving a 3 ’ –OH (Povirk et al., 2007). The nuclease
activities of Artemis are dependent upon Artemis forming an active complex with DNA-PKcs (Ma et al.,
2002a). The amino acids 398-403 of Artemis are critical for interaction with DNA-PKcs. V(D)J
recombination is drastically reduced in cells where Artemis has reduced DNA-PKcs binding affinity
(Niewolik et al., 2006a). Artemis contains 3 basal phosphorylation sites and 11 DNA-PKcs
phosphorylation sites within its C- t erm in al “ tail ” (Figure 1.1). While the basal phosphorylation sites are
dispensable for Artemis activation by DNA-PKcs, it is thought that phosphorylation by DNA-PKcs at the
non-basal sites are required (Ma et al., 2005b). Phosphorylation by DNA-PKcs results in a
conformational change where the inhibitory tail of Artemis is moved to a position that does not block
the catalytic site of Artemis (Ma et al., 2005a). However, other studies have shown that
autophosphorylation of DNA-PKcs in the cluster T2609-T2647 is sufficient to confer activity to Artemis
(Goodarzi et al, 2006) and phosphorylation of the C-terminal Artemis tail is dispensable for its
endonuclease activity. The conserved a m in o a cid s w i t h in t h e β - lac tam a se / β-caspase domain are critical

10

for the endonuclease activity of Artemis. Point mutations in several residues including: D136, D165,
H33, H115, H319 abolish V(D)J recombination in vivo (Pannicke et al., 2004). In addition to its nuclease
functions, Artemis has also been implicated in cell cycle arrest pathways. Artemis is a direct
phosphorylation target of ATM and ATR after exposure to IR or UV radiation and is thought to be
involved in the G2/M transition cell cycle arrest checkpoint pathway (Zhang et al., 2004b).

Figure 1.1 Schematic of Artemis Protein: Artemis is a member of Metallo-B-lactamase nucleases which
contains the two conserved motifs, B-lactamase, and B-Caspase. Artemis is 692 amino acids long and
contains a non-conserved C-terminal tail.
Polynucleotide Kinase and Aprataxin-like Forkhead-associated protein (PALF) also designated
“Aprataxin PNK-Like Factor ” (APLF) was first identified through a sequence homology search identifying
proteins with a FHA domain similar to polynucleotide kinase (PNK) and aprataxin (Kanno et al., 2007).
PALF protein is 511 amino acids long and contains several distinct domains. PALF has a FHA domain
near the N-terminus and two tandem zinc finger domains (PBZ) near the C-terminus as well as an
additional acidic C-terminal tail.
In cellular assays, the depletion of PALF is associated with radiosensitivity as well as persistence of y-
H2AX foci and 53BP1 foci, both of which are regarded by many as direct markers of double-stranded
DNA breaks (Mehrotra et al., 2011). These results indicate that PALF may play a role in DNA damage
repair.
Yeast-two-hybrid assay identified XRCC1, XRCC4, and Ku 80 as potential binding partners of PALF.
PALF interacts with XRCC1 and XRCC4 in vitro and in vivo; this interaction is dependent upon casein
kinase 2 (CK2) phosphorylation of XRCC1 and XRCC4 respectively (Iles et al., 2007). Additionally, PALF

11

also interacts with Ku80 in a manner independent of the FHA and PBZ regions within PALF (Macrae et
al., 2008). Recently, it has been shown that the vWA domain of Ku80 interacts with the amino acid
residues 100-263 of PALF (Shirodkar et al., 2013). In an electrophoretic shift assay (EMSA) it was found
that Ku, XRCC4/LigIV, and PALF were able to bind to DNA as a complex. Furthermore, an anti-PALF
antibody was able to super-shift the entire complex (Grundy et al., 2013). PALF does not contain a
nuclear localization signal (NLS). The PALF-Ku interaction was found to be crucial for nuclear retention
of PALF; however, PALF may also passively diffuse into the nucleus independently of binding to Ku 80
(Shirodkar et al., 2013). PALF is also able to bind to Poly-ADP-ribose as well as PARP-3 through its
tandem zinc finger domain (Ahel et al., 2008), implicating PALF in single-stranded-break (SSB) repair as
well as double-stranded-break (DSB) repair. In support of this hypothesis, PALF was found to rapidly
accumulate at sites of SSB via PARP-1 and sites of DSB via Ku80.
Poly (ADP) ribosylation is essential in signaling DNA damage and in the relaxation of the chromatin
for subsequent repair process. One stud y has sh o wn that P A L F’s a cid ic C-terminal region contains high
levels of homology to the nucleosome assembly protein 1 (NAP-1) and that PALF can bind to histone
H3/H4 tetramers acting as a histone chaperone (Mehrotra et al., 2011). Functionally, PALF has also
been implicated in the retention of the XRCC4-Ligase IV complex at sites of DSBs to promote the stability
of the NHEJ complex in vitro. In addition to the previously suggested function of PALF, purified PALF
protein has both exonuclease and endonuclease activity on DNA overhangs in vitro (Kanno et al., 2007).
This raises the possibility that PALF may function as a nuclease in a subset of NHEJ events in the cell.

12

Figure 1.2 PALF and PALF binding partners: PALF contains an N-terminal ForkHead Associated (FHA)
domain and a tandem C-terminal zinc finger repeat, CYR. The FHA domain is able to associate with
XRCC1 and XRCC4. The CYR domain of PALF interacts with Parp1 and Parp3. Ku80 binds to PALF
through an intermediate region.
1.3 DNA DSB and lymphoma, perspectives on inhibiting NHEJ.
Lymphomas are malignancies caused by uncontrolled proliferation of lymphoid cells. These cells are
often pre-B and pre-T in origin. Lymphoid neoplasms often arise due to chromosomal translocation
where RSS or RSS-like sequences are cleaved by the RAG complex, follow by subsequent joining to
another chromosomal break. Hence translocations can be thought of as a two step process where the
first step requires off target activity by RAG and/or AID and the second step requires joining by the NHEJ
proteins. Most (~85%) acute lymphoblastic leukemia cells (ALL) express RAG1 and/or 2, just as the pre-B
and pre-T cells from which the ALL cells derive. In normal pre-B and pre-T cells, RAG expression allows
for productive V(D)J recombination and subsequent generation of B and T cell receptors.
Traditional treatment of ALL has focused on using chemotherapeutic agents such as etoposide to
induce double-strand DNA breaks (DSBs) and ionizing radiation to kill the ALL cells. However, these
methods are nonspecific, causing toxicity in all rapidly dividing cells of the body. Etoposide targets
topoisomerase II, generating DSBs, leading to apoptosis or senescence. In addition to ALL cells, normal
cells also utilize topoisomerase II during DNA replication, and therefore etoposides cause toxity in all
cells of the body.

13

The rationale in inhibiting proteins involved in NHEJ is similar to the use of etoposide. If one can
inhibit the productive joining of broken DNA ends during V(D)J recombination, one can purposely create
DSBs resulting in apoptosis and senescence. This strategy takes advantage of the fact that ALL cells
naturally undergo V(D)J recombination and use the NHEJ machinery for joining the DNA ends.
Additionally, radiation treatment and chemotherapeutic agents often generate DSBs in target cancer
cells. Like normal cells, cancer cells will utilize NHEJ proteins in an attempt to repair the DSBs caused by
chemotherapeutic agents and radiation. If the NHEJ mechanism within the cancer cells were to be
blocked, this would make the cancer cells to be more vulnerable at a lower dose of radiation or
chemotherapy.
There have been numerous attempts in targeting the NHEJ machinery as a treatment for cancer.
The predominant ligase in NHEJ, Ligase IV, can be targeted, however it has been shown that mice lacking
ligase IV are embryonic lethal and that ligase IV appears to play a critical role in development (Frank et
al., 1998). Inhibitors for DNA-PKcs have been used in biochemical studies. While these inhibitors are
effective, the IC50's are much too high to be used in a clinical setting. DNA-PKcs also has additional
roles as a kinase outside of its function NHEJ, making it a less specific target than desirable. Polymerase
mu and lambda can be targeted; however both polymerases are only used in a subset of NHEJ during
V(D)J recombination and are not indispensable. The nuclease Artemis is critical during the hairpin
processing stage of V(D)J recombination and also plays a role in a subset of NHEJ following IR. Hence,
Artemis is an ideal target to inhibit in ALL, where hairpins are generated during V(D)J recombination. The
only off target effect of inhibiting Artemis is the minor loss of a small wave of normal developing B and T
cells during treatment. Small molecule inhibitor of Artemis, if found, would be able to take advantage
of the RAG-induced breaks to purposely create DSBs, causing slowed cell growth or apoptosis in ALL
cells. Additionally, a small molecule inhibitor of Artemis would be able to work in conjunction with
lower doses of radiation therapy and chemotherapy in treating ALL patients.

14

The following chapter is taken from previously published work by the same author and reprinted
here: Li et al “Polynucleotide Kinase and Aprataxin-like Forkhead-associated Protein (PALF) Acts as
Both a Single-stranded DNA Endonuclease and a Single-stranded DNA 3’ Exonuclease and Can
Participate in DNA End Joining in a Biochemical System.” (2011) JBC. 286 36368-36377
Chapter 2 PALF Nuclease Activities and their Relevance to NHEJ
2.1 Abstract:
Polynucleotide kinase and aprataxin-like forkhead-associated protein (PALF) also known as APLF
has previously been shown to have nuclease activity on single-stranded DNA. PALF is also known to
bind NHEJ factors XRCC4 and Ku80. Here, I verified that PALF has 3 ’ exonuclease activity. However, PALF
also possesses single-stranded DNA endonuclease activity. This single-stranded DNA endonuclease
activity can act at all single-stranded sites except those within four nucleotides 3 ’ of a double-stranded
DNA junction, suggesting that PALF minimally requires approximately four nucleotides of single-
strandedness. Ku, DNA-dependent protein kinase catalytic subunit, and XRCC4-DNA ligase IV do not
modulate PALF nuclease activity on single-stranded DNA or overhangs of duplex substrates. PALF does
not open DNA hairpins. However, in a reconstituted end joining assay that includes Ku, XRCC4-DNA
ligase IV, and PALF, PALF is able to resect 3 ’ overhanging nucleotides and permit XRCC4-DNA ligase IV to
complete the joining process in a manner that is as efficient as Artemis. Reduction of PALF in vivo
reduces the joining of incompatible DNA ends. Hence, PALF can function in concert with other NHEJ
proteins.

2.2 Introduction

Double stranded DNA breaks in cells are resolved either by non-homologous end joining or
homology directed repair. The choice between the two pathways depends on the stage of cell cycle at
which the break occurs in. Non-homologous end joining does not require the presence of a sister-
chromatid. Core NHEJ factors include Ku80/70, XRCC4/LigIV, XLF, Poly u, lambda, TdT, when addition of

15

nucleotides are required and DNA-PKcs/Artemis when end resection is required. It has been found that
PALF is able to bind to core NHEJ factor XRCC4 as well as Ku80. Additionally PALF is able to bind Parp
1and XRCC1 during single stranded DNA repair process. This raises the possibility that in addition to
Artemis another nuclease such as PALF may also be present to resect broken DNA ends before
annealing.
PALF contains a fork-head associated domain (FHA) which is known to bind XRCC4 and two
tandem zinc finger binding motifs (CYR) which binds to PARP1. Previous studies have shown that
depletion of PALF by siRNA is associated with impaired NHEJ (Koch 2008). Recent data has also shown
that PALF may be able to promote the retention of XRCC4/LigIV complex at broken DNA ends (Caldecott
2011). However what is most interesting is that purified PALF has been found to have endonuclease as
well as exonuclease activity in-vitro against a previously nicked DNA substrate. Here we explore this
mechanism further in relations to other NHEJ proteins and assess the ability of PALF to process broken
DNA ends before joining by XRCC4/LigIV.

2.3 Material and Methods:

Oligonucleotides and DNA Substrates
Oligonucleotides used in this study were synthesized by Operon Biotechnologies,
Inc. (Huntsville, AL) and Integrated DNA Technologies, Inc. (San Diego, CA). We purified the
oligonucleotides using 12% or 15% denaturing PAGE and determined the concentration
spectr o p h o t o m et ri call y . D N A su b strate 5 ’ end lab elin g w as d o n e with [ -32P]ATP (3000 Ci/mmol) and T4
p o lynu cle o tid e acc o rd in g t o t h e m an u facturer’s instru ctio n s. Su b stra te s we r e in c u b ate d w ith [ _ -32P]ATP
and T4 PNK for 30 min at 37 °C. T4 PNK was subsequently inactivated by incubating samples at 72 °C for
20min. Unincorporated radioisotope was removed by using G-25 Sephadex spin-column
chromatography. For the hairpin substrate, YM164-labeled oligonucleotide was diluted in a buffer
containing 10 mM Tris-hydrochloride (pH 8.0), 1 mM EDTA (pH 8.0), and 100 mM NaCl, heated at 100 °C

16

for 5 min, allowed to cool to room temperature for 3 h, and then incubated at 4 °C overnight. The
sequences of the oligonucleotides used in this study are as follows:
J G6 8 : 5 ’ -GAT CCT TCT GTA GGA CTC ATG-3’
J G1 6 9 : 5 ’ -TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT- 3 ’
YM 1 6 4 : 5 ’ -TTT TTG ATT ACT ACG GTA GTA GCT ACG TAG CTA CTA CCG TAG TAA T- 3 ’
YM 1 3 0 : 5 ’ -TTT TTT TTT TTT TTT ACT GAG TCC TAC AGA AGG ATC- 3 ’
YM 1 4 9 : 5 ’ -ACT GAG TCC TAC AGA AGG ATC TTT TTT TTT TTT TTT- 3 ’
YM 8 : 5 ’ -AGG CTG TGT TAA GTA TCT GCG CTC GCC CTC AGA GG- 3 ’
YM 9 : 5 ’ -CCTCTGAGGGCG AGC GCA GAT ACT TAA CAC AGC CT- 3 ’
J G2 5 8 : 5 ’ -CGA GCC CGA TCC GCT TGA CCA GTA GTC TAG CAC GTG ACG ATT GCA TCC GTC AAG TAA GAT
GCA GAT ACT TAA CGG GG- 3 ’
SL1 1 : 5 ’ -GTT AAG TAT CTG CAT CTT ACT TGA CGG ATG CAA TCG TCA CGT GCT AGA CTA CTG GTC AAG
CGG ATC GGG CTC GCC CCA AAA AA- 3 ’
SL1 5 : 5 ’ -ACT GAG TCC TAC AGA AGG ATC TTT TTT TTT TTS SSS- 3 ’
The sequence of siRNA for PALF, 5’-CCA GAU GAC UCC CAC AAA UAG, was synthesized, annealed with a
complementary strand, and used with a final concentration of 20 uM. Antibody against PALF was
prepared as reported previously.

Protein Expression and Purification
N-terminal His-tagged PALF cloned into a pET-16b (Novagen) vector has been described
previously. Soluble His-PALF was expressed and purified from pLysE BL21(DE3)-competent cells
(Invitrogen). Cells were precultured in ampicillin until A600 of 0.5. Cells were then induced with
isopropyl 1-thio-_-D-galactopyranoside (1mM) and cultured for an additional 3 h before harvesting.
Harvested BL21(DE3) cells expressing His-PALF were resuspended in Ni-NTA binding buffer (50 mM
NaH2PO4 (pH 7.8), 0.5 M KCl, 2 mM _-mercaptoethanol, 10% glycerol, 0.1% Triton X-100, and 20mM

17

imidazole (pH 7.8)) supplemented with protease inhibitors and lysed by sonication. The cell lysate was
applied to a Ni-NTA-agarose resin (Qiagen). Resin was washed with 35 mM imidazole. His-PALF was
eluted off with binding buffer plus 500 mM imidazole. Eluted fractions were dialyzed against Hi-Trap
heparin binding buffer (50 mM Tris-HCl (pH7.5), 10% glycerol, 2 mM EDTA, 1 mM DTT, 100 mM NaCl,
0.02% Nonidet P-40), loaded onto a pre-equilibrated Hi-Trap heparin column, and eluted with a linear
gradient to 1 M NaCl over 20 ml. We were able to obtain a yield of 200 _g from a 1-liter starting culture.
His-PALF-containing fractions were reconcentrated onto Ni-NTA-agarose and eluted in binding buffer
plus 500mM of imidazole to a final elution volume of 200_l. His-PALF was then applied to a Superose 12
gel filtration column (GE Healthcare) and eluted with 250 mM NaCl, 10% glycerol, 50 mM Tris-HCl (pH
7.5), and 1 mM DTT. PALF samples were stored at 4 °C and used within 2 weeks because the nuclease
activity of PALF is highly sensitive to freezing, as pointed out previously. The expression and purification
of DNA-PKcs from HeLa cells has been described previously.

In Vitro Nuclease and Ligation Assays
I extended the PALF nuclease optimizations performed previously by further testing activity in a
range of pH (6.5 to 8), divalent cation concentration (0 to 20 mM MgCl2), and monovalent salt
concentration (1 mM to 100 mM KCl). In vitro DNA nuclease assays were performed in a total volume of
10ul with a buffer composition of 25 mM Tris-HCl (pH 7.5), 10 mM KCl, 10 mM MgCl2, 1 mM DTT and 50
ng/ul BSA. In the reaction, 50 nM single-stranded DNA substrate with an overhang or 20nMhairpin
substrate were incubated with 125 nM PALF, and, in specified cases, one or more of the following: 126
nM DNA-PKcs, 100 nM Ku, 75 nM XRCC4/ligase IV, or a combination of the three proteins. In reactions
containing XRCC4/ligase IV, the XRCC4 and ligase IV were prephosphorylated by CK2 according to the
m an u facturer ’s instru c tio n s ( Si g m a -Aldrich). Unless specified otherwise, when DNA-PKcs was present,
0.5 mM ATP and 0.5 uM 35-bp blunt-end DNA (YM 8/9) were also included in the specified reactions.

18

Reactions were then incubated at 37 °C for 1 h. After incubation, reactions were stopped and analyzed
on 12% denaturing PAGE gels.

In vitro ligation assays were performed in a total volume of 10ul with a buffer composition of 25mMTris
(pH 7.5), 2mMDTT, 0.025% Triton X-100, 0.1 mM EDTA, 10% PEG, 50 ng/_l BSA, and 5% glycerol. In the
in vitro ligation assay, a two-step reaction was performed. In the reaction, 50 nM double- stranded
substrate (SL11/JG25) was incubated with 125 nM PALF, and, in specified cases, one or more of the
following: 126 nM DNA-PKcs, 100 nM Ku, 75 nM XRCC4/ligase IV, or a combination of the three proteins.
Unless specified otherwise, when DNA-PKcs was present, 0.5 mM ATP and 0.5 _M 35-bp blunt-end DNA
(YM 8/9) were also included in the specified reactions. A two-step ligation reaction was carried out in
the following order: PALF, substrate, and DNA-PKcs were first added to the reaction and incubated at 37
°C for 30 min. Ligase IV-XRCC4 and Ku were then added, and the reaction was incubated for another 30
min at 37 °C. The reactions were stopped, and DNA was phenol-extracted and analyzed on 8%
denaturing PAGE gel. Gels were dried, exposed in a phosphorimager cassette, and scanned. Bands were
cut out at the dimer position on ligation gels, and we used TOPO TA to clone this DNA into the
Invitrogen pCR2.1 TOPO vector. The insert was subsequently sequenced using a Li-Cor 4200 sequencer,
and the junctions were analyzed.

2.4 Results:

PALF has 3’ Single-stranded Exonuclease Activity
PALF was purified using Ni-NTA, HiTrap heparin, and Superose 12 columns (Fig. 2.1A) and found
it to have nuclease activity across an elution peak in proportion to the amount of protein visualized
using Coomassie Blue staining (Fig. 2.1, A and B).

19

FIGURE 2.1. Purification of PALF.
A, PAGE gel on Superose 12 fractions. After Ni-NTA and Hi-Trap heparin purification, Superose 12
fractions of PALF are shown on an 8% SDS-PAGE gel stained with Coomassie Blue on which PALF has a
gel mobility position at 81 kDa. Ladder designates the protein marker lane, and the fraction numbers are
above each lane.
B, nuclease activity of PALF corresponding to Superose 12 fractions. Fractions across the Superose 12
elution peak were assayed for nuclease activity using poly(dT) substrate (JG169). Each reaction consists
of 50 nM single-stranded DNA substrate (JG169) and 50 nM PALF. Reactions were incubated for 2 h at
37°C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE.

20

The activity of PALF was tested on a single-stranded DNA substrate to determine whether PALF
has nuclease capabilities beyond those described originally. A single stranded poly dT 30mer labeled at
the 5 ’ e n d was u sed t o inv e stig ate thi s ( Fi g . 2 . 2 A, lane 1). On this substrate, PALF generates a series of
cleavage products (Fig. 2.2A, lane 2). The addition of DNA-PKcs and ATP does not increase or decrease
the nuclease activity of PALF on single-stranded DNA (Fig. 2.2A, lane 3). Stimulation of DNA-PKcs
autophosphorylation by the addition of unlabeled double-stranded DNA, YM8/YM9, also did not affect
the activity of PALF (Fig. 2.2A, lane 4). Artemis was used to generate a ladder starting at the 1 nucleotide
position (Fig. 2.2A, lane 5). PALF does not have any 5_exonuclease activity. Unlike Artemis, no distinct
product is formed at the 1 nucleotide position (Fig. 2.2A, lane 2 versus lane 5). The nucleolytic cleavage
p ro d u cts t h a t w e d o se e ca n be ex p lai n ed b y a co m b in atio n o f 3 ’ e xo n u cl ease acti v ity and sing le -
stranded endonuclease activity. In the presence of Mn2
+
, Artemis does not require DNA-PKcs to function
endonucleolytically . MnCl2 does not affect the activity of PALF (Fig. 2.2A, lane 6).
To dif fe rentiat e b e tw e en t h ese t wo acti v itie s, we lab eled t h e sa m e sub s trate at the 3 ’ e n d ( Fi g .
2.2B, lanes 1 –3). Products generated by PALF from this substrate include a prominent band at the 1
nucleotide position (Fig. 2.2B, lanes 2 and 3 ), de m o n s tratin g defin itiv e 3 ’ e xo n u cl ease a ctiv i ty . H o w ev er,
a full range of larger products with a weaker profile is also present (Fig. 2.2B, lanes 2 and 3). The amount
of each species in this distribution of products increases relatively equally as the amount of enzyme was
increased (twice as much PALF in lane 2 than in lane 3 ). Giv en t h at the s u b str ate was lab el ed at the 3 ’
terminus, these weaker products can only be generated by the endonuclease activity of PALF acting on
the single-stranded substrate. Hence, we conclude that PALF has endonuclease activity on single-
strand ed D N A in ad d itio n t o 3 ’ e xo n u cle ase a cti v ity .

21

FIGURE 2.2. 3’ to 5 ’ exonuclease activity of PALF on single-stranded DNA.
A, exonuclease activity of PALF on single-stranded DNA. In the reaction, 50 nM single-stranded DNA
substrate (JG169) was incubated with the protein(s) indicated above the lane in a 10-ul reaction for 60
min at 37 °C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE. Protein
concentrations are as follows: PALF, 125 nM and DNA-PKcs, 126 nM. As specified, 0.5mM ATP and 0.5
uM YM8/9 was also included in designated reactions. YM8/9 is a 35-bp blunt-ended double-stranded
DNA that is used as DNA-PKcs cofactor.
B , 3 ’ ex o n u cleas e m o n it o r e d w ith 3 ’ -labeled substrate. In the reaction, 50 nM single-stranded DNA
sub strate (JG 1 6 9 ) lab eled a t its 3 ’ end w as in cub ate d with P A L F in a 1 0 -ul reaction for 60 min at 37 °C.
Concentrations are as follows: 250 nM PALF (lane 2) and 125 nM PALF (lane 3). After incubation,
reactions were stopped and analyzed by 12% denaturing PAGE. The asterisk represents the radiolabel in
all figures. The bold arrow on the DNA substrate diagram beside the gel represents the site of DNA
cleavage in all figures.

PALF Has Single-stranded DNA Endonuclease Activity
To test for endonuclease activity that does not require substrates with abasic sites, we designed
substrates containing either a 15- n u cle o tid e 3 ’ o v erha n g o r a 1 4 - n u cl eo tid e 5 ’ o v erha n g , bo th lab eled at the 5 ’ e n d ( Fi g . 2 .3 A, lane 1 for the 5_overhang substrate, and B, lane 1 fo r t h e 3 ’ o v erha n g sub stra te ).
This end o n u cl ease a cti v ity cleav es ran d o m l y within t h e 5 ’ o v erha n g . Unlike Artemis, the addition of

22

DNA-PKcs does not further stimulate the endonuclease activity (Fig. 2.3a,lane 2 versus lane 3). Addition
of unlabeled dsDNA to DNA-PKcs (to stimulate autophosphorylation) also has no effect on the activity of
P ALF a t the 5 ’ overhang (Fig. 3a, lane 2 versus lane 3). Hence, we conclude that unlike Artemis, PALF
does not require DNA- P K cs fo r end o n u cleas e ac tiv i ty a t a 5 ’ o v erha n g . On a 5 ’ -labeled duplex DNA
substrate with a 15- n t p o l y T 3 ’ o v erha n g , nin e t o t en d istin ct p ro d u c t s are seen (Fig. 2.3B lane 2). PALF
demonstrates endonuclease activity but does not cleave at positions within 4 or 5 nts from the double
stranded portion of the substrate (Fig. 2.3B, lane 2 ). T h e 3 ’ o v erha n g ac tiv it y is no t s tim u la te d o r bl o cke d by the addition of DNA-PKcs and ATP (Fig. 2.3B, lane 3). Further addition of unlabeled double-stranded
DNA, YM8/YM9, with DNA-PKcs also had no effect on the nuclease activity of PALF (Fig. 2.3B, lane 3).
The addition of Ku did not yield any changes when compared with the basal level with PALF alone (data
n o t sh o wn). L ike t h e 5 ’ o v e rh an g sub strate , n o o n e cle av ag e p r o d u ct is do m in an t o v er an y o ther. H enc e,
the nuclease activity on the single- strand ed 3 ’ o v erha n g app ears t o be r an d o m i n its p o si tio n . P ALF also h as 3 ’ ex o n u clease act i v it y . T h eref o re, t h e act i v ity w e see a t the 3 ’ o v erha n g is a co m b in atio n o f b o th
end o n u cleas e as we ll as 3 ’ ex o n u cl ease acti v ity o n sin g le -stranded regions. When we position four
p h o sph o thi o est er li n kage s at the 3 ’ end o f t h e sub stra te , m u c h le ss pro d u c t d eri v es fr o m t h e 3 ’
o v erha n g , ind icatin g dimini shed 3 ’ ex o n u cleas e ac tiv it y . H o we v er, so m e pro d u ct still ari ses, at trib u tabl e
to the e n d o n u cleas e acti v it y , w h ich c an nick up stre a m o f the ph o sph o thi o est er li n kages. F o r 3 ’
overhang substrates , P ALF see m s t o fun c tio n all y o ccu p y app ro xi m at ely fo u r o r fi v e n u cl eo tid es 3 ’ t o the
duplex portion of the substrate, as indicated by termination of nicking at length 26 nt (rather than 21 nt)
(Fig. 2.3B). The approximately four or five nucleotides may reflect the minimum size required for PALF to
bind to single-stranded DNA and act upon substrates.

23

FIGURE 2.3. Endonuclease activity of PALF on overhangs.
A, in specified reactions, 50 nM 5_-labeled double- str a n d ed D N A su b strat e, YM 1 3 0 / Y M 6 8 (5 ’ o v erha n g ),
was incubated with 125 nM PALF, 126 nM DNA-PKcs, 0.5mM ATP, and 0.5uM YM8/9 in a 10-ul reaction
for 60 min at 37 °C. After incubation, reactions were stopped and analyzed on 12% denaturing PAGE.
B , in sp ecifi ed reac tio n s, 5 0 nM 5 ’ -labeled double- stra n d ed D N A su b strat e, YM 1 4 9 / Y M 6 8 (3 ’ o v erha n g ),
was incubated with 125 nM PALF, 126 nM DNA-PKcs, 0.5 mM ATP, and 0.5 uM YM8/9 in a 10-ul reaction
for 60 min at 37 °C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE.

This is the sa m e b ehavi o r a s is se en f o r Ar t em is e n d o n u clease a cti o n o n 3 ’ o v erha n g s ( Fi g . 2 .4, lane 5).
Hence PALF has single-stranded endonuclease activity and that this i s distin c t fr o m its 3 ’ e x o n u cleas e

24

activity. Because of the fact that PALF has endonuclease activity, the possibility that PALF could
potentially open a hairpin substrate like Artemis was also investigated. A hairpin substrate with a 20-
nucleotide duplex portion and a 6- n u cle o tid e 5 ’ o v erh an g w as used fo r thi s e xper im ent ( Fi g . 2 .4 , lane 1).
When PALF alone is incubated with the substrate, four endonuclease products can be seen at the
bottom of the gel (Fig. 2.4, lane 2). This is consistent with the endonuclease activity we observed before
fo r t h e 5 ’ o v erha n g sub stra t e, Y M 1 3 0 / Y M 6 8 . Onc e ag ai n w e u s ed Ar te m is -DNA-PKcs to generate a ladder
to show the 1 nucleotide position and the hairpin opening product at the 28-nt position (Fig. 2.4, lane 5).
This product is absent in the reaction where PALF alone is added (Fig. 2.4, lane 2). The hairpin opening
activity of Artemis is dependent upon DNA-PKcs, and we explored this possibility with respect to PALF.
DNA-PKcs and ATP were added to the reaction in addition to PALF. This did not stimulate PALF to cleave
the hairpin, and no additional endonuclease activity is seen (Fig. 2.4, lane 3). Further stimulation of DNA-
PKcs by unlabeled DNA YM8/9 also did not stimulate PALF to cleave a hairpin substrate (Fig. 2.4, lane 4).
Therefore, unlike Artemis, PALF does not show DNA-PKcs-dependent ability in opening hairpin
substrates (Fig. 2.4, lane 4 versus lane 5). The endonuclease activity further supports our previous
fin d in g s o f P A L F ac tiv it y at a 5 ’ o v erha n g .

25

FIGURE 2.4. Lack of endonuclease activity of PALF on hairpin substrates.
In specified reactions, 20 nM hairpin DNA substrate, YM164, was incubated with
125 nM PALF, 50 nM Artemis, 126 nM DNA-PKcs, 0.5mM ATP, and 0.5uM YM8/9
in a 10-_l reaction for 60 min at 37 °C. After incubation, reactions were
stopped and analyzed by 12% denaturing PAGE.

The Nuclease Activity of PALF Is Not Affected by Ku or XRCC4-DNA Ligase IV—
Previous data have demonstrated binding of PALF to XRCC4 via the FHA domain of PALF and
binding to Ku through a region in PALF between the FHA domain and the two CYR (poly [ADP-ribose]
binding zinc finger) motifs. I wanted to investigate the possibility that the XRCC4-DNAligase IV complex
or Ku can enhance or block the exonuclease or endonuclease activity of PALF. As above, a single-
strand ed po l y T sub strat e l ab eled at the 5 ’ end w as us ed in incu b atio n s w ith P AL F fo r 1 h with o r w i tho u t
purified XRCC4-ligase IV or Ku. The basal activity of PALF generates a combination of endonuclease
p ro d u cts and 3 ’ ex o n u cl eas e products (Fig. 2.5, lane 2). XRCC4-DNA ligase IV, after being phosphorylated
by CK2, was run alone to rule out the possibility of contamination (Fig. 2.5, lane 3). XRCC4-DNA ligase IV

26

without phosphorylation was also used as a negative control (Fig. 2.5, lane 4). The addition of XRCC4-
ligase IV did not produce a significant change over basal PALF activity (Fig. 2.5, lane 2 versus lane5).
Finally, a combination of PALF, XRCC4-ligase IV, and DNA-PKcs failed to have any effect relative to PALF
alone (Fig. 2.5, lane 2 versus lane 6). This is noteworthy, given that both Ku and XRCC4-ligase IV are
known to bind to PALF and that DNA-PKcs is suggested to interact with PALF according to proteome
analysis. Therefore, the interactions do not affect the nuclease activity of PALF.

FIGURE 2.5. Effect of Ku, XRCC4-DNA Ligase IV, and DNA-PKcs on PALF
nuclease activity. In specifi ed reac tio n s, 5 0 nM 5 ’ -labeled single-stranded
DNA substrate (JG169) was incubated with 125 nM PALF, 126 nM DNA-PKcs,
0.5 mM ATP, 0.5 uM YM8/9, 100 nM Ku, and 75 nM XRCC4-DNA ligase IV in a
10-ul reaction for 60 min at 37 °C. After incubation, reactions were stopped
and analyzed by 12% denaturing PAGE.

27

PALF Is Able to Cooperate with Other NHEJ Factors to Promote Ligation in Vitro
Even though the core NHEJ factors did not stimulate the nuclease activity of PALF or permit it to
open hairpins, we were interested in whether PALF could cooperate with other NHEJ factors to
stimulate ligation in an in vitro assay. In particular, we wondered if the nuclease activities of PALF could
substitute for those of Artemis in resecting overhangs. We tested this by using an oligonucleotide
substrate (SL11/JG258) with four nucleotides of microhomology that had six additional A nucleotides
att ach ed to t h e 3 ’ e nd that would require nucleolytic resection before the 4 nts of microhomology could
be utilized for ligation. Successful ligation would result in the substrate ligated to another identical
molecule in a head-to-tail fashion. When the DNA substrate (Fig. 2.6A, lane 1) is incubated with XRCC4-
ligase IV complex and Ku, it results in a small basal amount of ligation versus substrate alone (Fig. 2.6A,
lane 2 versus lane 1). In the absence of known NHEJ factors, when PALF alone is added to the reaction
(Fig. 2.6A, lane 3), no ligation products are formed. However when PALF, Ku, and XRCC4-ligase IV are all
present, ligation products are formed (Fig. 2.6A, lane 4). PALF is able to significantly increase the amount
of product formed versus the basal level (Fig. 2.6A, lane 4 versus lane 2). The addition of DNA-PKcs to
the reaction hinders the ligation process (Fig. 2.6A, lane 5). It is thought that DNA-PKcs must dissociate
before ligation of DNA ends can occur. This process is thought to be facilitated by autophosphorylation
of DNA-PKcs. However, the precise mechanism is not well understood. In vitro, this process may be too
slow in our ligation reactions and hence may inhibit the overall ligation efficiency. The ligation product
was subsequently cut out from the gel, cloned into a TA cloning vector, and then sequenced. We found
that all ten of the joined product molecules had the six As resected, leaving only the fo u r Cs a t the 3 ’ overhang (Fig. 2.6C). This allows for efficient joining because of the microhomology between the 4 Cs on
the t o p st ran d a t th e 3 ’ t er m in u s of o n e d u p lex sub s tr ate and t h e 4 Gs o f the b o t to m st r an d o f a se co n d duplex substrate. This is also consistent with the results above showing that the PALF nuclease activity
terminates its cleavage 3 to 4 nucleotides away from the junction between the overhang and the double

28

stranded portion of a substrate. We also compared the efficiency of ligation of PALF versus that of
Artemis in an in vitro assay. Using the same substrate (SL11/JG258), we added Ku and XRCC4-DNA ligase
IV to Artemis or PALF. Artemis is able to resect the ends in a similar manner as PALF, resulting in ligation
using terminal microhomology (Fig. 2.6B, lane 1). Under the same conditions, PALF is also able to ligate
the substrate with a similar efficiency as Artemis (Fig. 2.6B, lane 3, 12%, versus lane 1, 16%).

FIGURE 2.6. PALF can cooperate with Ku and XRCC4-DNA ligase IV in double-strand DNA end ligation.

29

A, in specified reactions, 50 nM 5_-labeled doublestranded
DNA substrate (SL11/JG258) was incubated with 125 nM PALF, 126 nM DNA-PKcs, 0.5mM ATP, and 0.5
uM YM8/9 for 30 min at 37 °C and followed by the addition of 100 nM Ku and 75 nM XRCC4-DNA ligase
IV for 30 min at 37 °C. After incubation, reactions were stopped and analyzed by 8% denaturing PAGE.
Positions of the dimerized and trimerized DNA duplex products from the monomeric ligations were
determined on the basis of duplex DNA markers not shown on the gel (also see sequencing results in C,
which confirm the dimer junctions).
B , in sp ecifi ed reac tio n s, 5 0 nM 5 ’ -labeled double-stranded DNA substrate (SL11/JG258), was incubated
with 75 nM Artemis or 125 nM PALF, 126 nM DNA-PKcs, 0.5 mM ATP, and 0.5 uM YM8/9 for 30 min at
37 °C and followed by the addition of 100 nM Ku and 75 nM XRCC4-DNA ligase IV for 30 min at 37 °C.
After incubation, reactions were stopped and analyzed by 8% denaturing PAGE.
C, dimer products from A, lane 4, were cut out of the gel, extracted, PCR-amplified, TA-cloned, and
sequenced. The junctional sequences are sh o wn. P AL F rem o v ed th e AA AA AA ( in itali cs) fr o m e a ch 3 ’ o v erha n g , t h u s all o wing t h e 3 ’ - CC CC o v erha n g o f o n e d u p lex sub strate t o ann eal to the 3 ’ -GGGG on
another duplex substrate molecule. These proceeded to ligation by XRCC4-DNA ligase IV to yield the
dimer product.

2.5 Discussion:

PALF Is Both a Single-stranded DNA 3’ Exonuclease and a Single-stranded DNA Endonuclease
The Yasui laboratory (1 –3) was the first to demonstrate that PALF (also called APLF) has nuclease
activity. Here we confirm that PALF h as 3 ’ e xo n u cl eas e acti v it y o n sing le -stranded DNA. We have also
found that PALF has endonuclease activity on single-stranded DNA. This activity is capable of cleaving at
all site s o f a sin g le s tran d e xce p t f o r the las t fe w n u cle o tid es at the 5 ’ e n d . In ad dition, this activity is
ab le t o c lea v e at all po si tio n s o f 5 ’ o v erha n g s o f du p le x D N A ex c ept f o r t h e 5 ’ -most few nucleotides of
the ove rh an g . Fi n all y , t h is a ctiv it y c l eav es a t all p o siti o n s o f 3 ’ o v erha n g s e xce p t f o r t h e app ro xim a te ly fo u r nu cle o tid es 3 ’ to the boundary of the overhang with the duplex DNA.

Unification of the PALF Nuclease Activities
We believe that the double-stranded DNA endonuclease activity described previously might be
accounted for by the broader single-stranded DNA endonuclease activity described here. The transient
breathing of a double-stranded DNA end into single-stranded flaps, which is substantial , could serve as
a single-stranded substrate for this single-stranded DNA endonuclease activity of PALF. Given the action

30

of PALF on all single-stranded DNA, the endonuclease of PALF has a broader range of substrates than
appreciated previously, thereby increasing its potential importance within the cell.

In Vitro and in Vivo Function of PALF with Other NHEJ Proteins
We did not see an y st i m u la tio n o r m o d ifi c atio n o f P A L F 3 ’ e xo n u cl ease and sing le -stranded DNA
endonuclease activity by Ku, DNA-PKcs, or XRCC4-DNA ligase IV. Although studies indicate that Ku and
XRCC4-DNA ligase IV each bind to PALF, these interactions do not appear to affect PALF enzymatically.
Importantly, we find that in an in vitro system consisting of PALF, Ku, and XRCC4-DNA ligase IV is indeed
able to join incompatible DNA ends as efficiently as a system consisting of Artemis-DNA-PKcs, Ku, and
XRCC4-DNA ligase IV. Specifically, PALF is able to resect the incompatible portion of an overhang to a
point at which XRCC4-DNA ligase IV is able to support efficient ligation. Interestingly, at the DNA
sequence level, we did not observe any ligation products other than those with the four junctional
nucleotides that provide the maximal terminal microhomology. (This does not preclude the possibility
that some junctions do contain only three nucleotides and make up a small minority of the products
formed.) Use of maximal microhomol o g y c o u ld o ccur b y ite ra tiv e re m o v al o f t h e As b y t h e 3 ’ exonuclease activity of PALF or by the endonuclease PALF activity. We favor the latter possibility
because we did not see a more diverse set of products.

31

The following chapter is taken from previously published work by the same author and reprinted
here: Gu & Li “DNA-PKcs regulates a single-stranded DNA endonuclease activity of Artemis” (2010)
DNA Repair 9; 429-437
Chapter 3 Single Stranded Endonuclease Activity of Artemis
3.1 ABSTRACT
Human nuclease Artemis belongs to the metallo-beta-lactamase protein family. It acquires
double-stranded DNA endonuclease activity in the presence of DNA-PKcs. This double-stranded DNA
endonuclease activity is critical for opening DNA hairpins in V(D)J recombination and is thought to be
important for processing overhangs during the nonhomologous DNA end joining (NHEJ) process. Here
we show that purified human Artemis has a molecular weight that corresponds to a trimer on size
exclusion chromatography. Interestingly, Artemis exhibits single-stranded DNA endonuclease activity.
This activity is stimulated by DNA-PKcs and modulated by purified antibodies raised against Artemis.
Moreover, the single-stranded endonuclease activity is modulated by changes in divalent cations in the
same manner as the double-stranded DNA endonuclease activity of Artemis:DNA-PKcs. These findings
further expand the range of DNA substrates upon which Artemis and Artemis:DNA-PKcs can act. These
findings are discussed in the context of NHEJ and V(D)J recombination.
3.2 INTRODUCTION
Nonhomologous DNA end joining (NHEJ) is the primary DNA repair pathway for double-strand
breaks in multicellular eukaryotes. Like most DNA repair pathways, NHEJ includes a nuclease to resect
damaged DNA, polymerases to fill-in DNA, and a ligase to restore integrity of the DNA strands (Lieber,
2008). Among vertebrates, it appears that a complex of Artemis:DNA-PKcs provides endonuclease
activity (Ma et al., 2002b; Ma et al., 2005c). DNA-PKcs is a protein kinase which is only active when it is
bound to a DNA end (Anderson and Carter, 1996). DNA-PKcs phosphorylates itself as well as other
proteins (Meek et al., 2008). DNA-PKcs autophosphorylation appears to change its conformation (Meek

32

et al., 2004). Artemis is a nuclease which only acquires endonuclease activity on double-stranded (ds)
DNA when it is bound to an autophosphorylated DNA-PKcs (Ma et al., 2005b; Ma et al., 2002b; Niewolik
et al., 2006b; Yannone et al., 2008). DNA-PKcs also phosphorylates Artemis at 11 sites in the C-terminal
portion of Artemis (Ma et al., 2005b), and this region of Artemis is important for Artemis function in vivo
(Huang et al., 2009). In vitro, removal of this C-terminal region of Artemis permits it to function as an
endonuclease independent of DNA-PKcs on short hairpins (Ma et al., 2005b; Niewolik et al., 2006b).
The Artemis:DNA-PKcs complex endonuclease activities are interesting. The complex appears to
localize to the junction of dsDNA and ssDNA and nick the DNA. For 5' overhangs, the complex cuts
directly at the junction, generating a blunt duplex product (Ma et al., 2002b). For 3' overhangs, the
complex prefers to cut on the single-stranded DNA (ssDNA) overhang 4 nt out from the junction with the
duplex DNA. Interestingly, the Artemis:DNA-PKcs complex also cuts perfect hairpins 2 nt past the
hairpin tip and on the 3' side. All of these endonuclease activities are most consistent with a model in
which DNA-PKcs helps Artemis localize to the junction of dsDNA and ssDNA; but the Artemis:DNA-PKcs
complex appears to require 4 nt of ssDNA to bind, and then it cuts on the 3' side of that 4 nt stretch (Ma
et al., 2002b). This would explain cutting directly at the ss/ds DNA junction for 5' overhangs, but 4 nt
away from the ss/ds DNA junction on 3' overhangs (Lu et al., 2007; Ma et al., 2002b). Moreover, it
explains nicking of hairpins not at the very tip, but 2 nt past the tip on the 3' side; the last 2 bp of a
perfect hairpin are known to be largely unpaired, thus providing 4 nt of ssDNA at the hairpin tip where
Artemis:DNA-PKcs can bind (Blommers et al., 1989; Howard et al., 1991; Raghunathan et al., 1991).
Here we describe a new endonucleolytic property of Artemis. We find that Artemis alone has
weak endonucleolytic activity on ssDNA (homopolymers). When DNA-PKcs is added, the Artemis:DNA-
PKcs complex is markedly stimulated in this ssDNA endonuclease activity, and this is ATP-dependent and
requires duplex DNA to stimulate DNA-PKcs kinase activity. Immunoinhibition studies using antibodies

33

against Artemis confirm that the ssDNA endonuclease activity is intrinsic to Artemis. These findings
further expand the range of substrates that Artemis and Artemis:DNA-PKcs can act upon. The relevance
for NHEJ is discussed.
3.3 MATERIALS AND METHODS
Oligonucleotides
Oligonucleotides used in this study were synthesized by Operon Biotechnologies, Inc.
(Huntsville, AL, USA) and Integrated DNA Technologies, Inc. (San Diego, CA, USA). We purified the
oligonucleotides using

12% or 15% denaturing polyacrylamide gel electrophoresis (PAGE)

and
determined the concentration spectrophotometrically.
D N A su b strat e 5 ’ e n d lab elin g w as do n e wi th [g am m a -
32
P]ATP (3000

Ci/mmol) (PerkinElmer Life
Sciences, Boston, MA, USA) and T4

polynucleotide kinase (New England Biolabs, Beverly, MA, USA)

according to the manufacturer's instructions. Unincorporated

radioisotope was removed by using G-25
Sephadex (Amersham Biosciences,

Inc., Piscataway, NJ, USA) spin-column chromatography. For the
hairpin substrate, YM164, labeled oligonucleotide was diluted

in a buffer containing 10 mM Tris –
hydrochloride, pH 8.0,

1 mM EDTA, pH 8.0 and 100 mM NaCl and then heated at 100°C for 5 min,
allowed to cool to room temperature for

3 h, and then incubated at 4°C overnight.
D N A su b strat e 3 ’ e n d lab elin g w as do n e wi th [al p h a -
32
P]TTP (3000

Ci/mmol) (PerkinElmer Life
Sciences, Boston, MA, USA), ddTTP and terminal deoxynucleotidyl transferase (Promega, Madison, WI,
USA) according to the manufac turer’s in s tructi o n s. T h e m o lar r atio o f [alp h a -
32
P]TTP to ddTTP used was
1:5. Unincorporated

radioisotope was removed by using G-25 Sephadex (Amersham Biosciences,

Inc.,
Piscataway, NJ, USA) spin-column chromatography.
The sequences of the oligonucleotides used in this study are as follows:

34

J G1 6 7 : 5 ’ -AAA AAA AAA AAA AAA AAA AAA AAA AAA AAA-3’
J G1 6 8 : 5 ’ -CCC CCC CCC CCC CCC CCC CCC CCC CCC CCC-3’
J G1 6 9 : 5 ’ -TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT- 3’
J G2 8 2 : 5 ’ -[P]TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT-3’
YM 1 6 4 : 5 ’ -TTT TTG ATT ACT ACG GTA GTA GCT ACG TAG CTA CTA CCG TAG TAA T-3’
Protein expression and purification
Artemis recombinant baculovirus with a C-terminal His tag (and an intervening TEV site) was a
gift from Dr. John Harrington (Athersys, Ohio). Soluble Artemis-his is expressed and purified from Sf9
insect cells. Briefly, harvested Sf9 cells were resuspended in Ni-NTA binding buffer (50 mM NaH
2
PO
4
(pH
7.8), 0.5 M KCl, 2 mM beta-mercaptoethanol, 10% glycerol, 0.1% Triton X-100, and 20 mM imidazole (pH
7.8)) supplemented with protease inhibitors and lysed by sonication. Cell lysate was applied to Ni-NTA
agarose resin (Qiagen). Artemis-his was eluted off with binding buffer plus 250 mM imidazole. Eluted
fractions were dialyzed against DEAE binding buffer (50 mM Tris-HCl (pH 7.5), 10% glycerol, 2 mM EDTA,
1 mM DTT, 100 mM NaCl, and 0.02% NP-40), loaded onto preequilibrated DEAE-Sepharose column, and
eluted with a linear gradient to 1 M NaCl over 20 mL. Artemis-his containing fractions were
concentrated with Vivaspin 500 (Sartorius Stedim Biotech) and further purified by gel filtration on a
Superose 12 column using a SMART system (Amersham Pharmacia) in gel filtration buffer (25 mM HEPES
(pH 7.5), 0.5 M KCl, 10 mM MgCl
2
, and 1 mM DTT). Eluted Artemis-his from Superose 12 was aliquoted
and stored at -80
o
C.
The expression and purification of DNA-PKcs from Hela cells has been previously described (Ma
et al., 2002b).

35

Antibody production
Anti-Artemis antibody was produced as described (Zhang et al., 2004a). In brief, a fragment
encoding amino acid residues 347 to 692 of Artemis was fused to a hexahistidine tag by insertion into
pET28 (Novagen). Purified recombinant protein from E. coli was used to raise antiserum in rabbits using
standard protocols. Antisera were affinity purified using antigen that had been blotted and immobilized
on nitrocellulose paper or by affinity chromatography.
In vitro nuclease assay
In vitro DNA nuclease assays were performed in a total volume of 10 µl with a buffer
composition of 25 mM Tris-HCl (pH 8.0), 10 mM KCl, 10 mM MgCl
2
, 1 mM DTT and 100 ng/uL BSA. In the
reaction, 50 nM single-stranded DNA substrate or 20 nM hairpin substrate were incubated with 50 nM
Artemis and 50 nM DNA-PKcs unless otherwise specified. When DNA-PKcs was present, 0.25 mM of ATP
and 0.25 uM of 35bp blunt end DNA (YM 8/9) were also included in specified reactions. Reactions were
then incubated at 37°C for 30 min. After incubation, reactions were stopped and analyzed on 12%
denaturing PAGE gels. Gels were dried, exposed in a phosphorImager

cassette and scanned.
3.4 RESULTS
Size Exclusion Chromatography of Purified Artemis Indicates a Molecular Volume that Corresponds to
a Trimer
Human Artemis-His was over expressed with a baculovirus-insect cell system as described in the
Methods. Purified Artemis from Ni-NTA columns and DEAE-Sepharose columns was further fractionated
on Superose 12 gel filtration columns to estimate the stoichiometry of the active species. The
predominant absorbance material elutes off the column as a single peak, which corresponds to a MW
range of 239-292 kDa, based on the calibration curve generated with standard molecular weight
markers (Fig. 3.1). SDS-PAGE protein gels were run on all fractions and demonstrated that the only

36

protein bands visible across the entire Superose 12 elution was a band that ran at 100 kDa, which is
precisely the PAGE mobility of denatured Artemis (not shown). The identity of the band was further
confirmed on a linear ion trap LTQ (Thermo-Fisher) mass spectrometer. The only protein identified was
the recombinant human Artemis. The Superose 12 fractions (9 and 11) containing active Artemis were
concentrated and in -solution digested with trypsin. The digested peptide mixtures were analyzed by
LC/MS/MS on a linear ion trap LTQ (Thermo-Fisher). We searched two insect protein databases for any
proteins that might co-purify with the Artemis, given that it was purified from baculovirus-infected
insect cells. No insect proteins (or any other proteins) were identified. The molecular weight of
Artemis-his is about 80 kDa; therefore, it appears that purified Artemis exists primarily as a trimer.

37

Figure 3.1. Size exclusion chromatography of purified Artemis.
(A) Superose 12 gel filtration chromatogram of purified Artemis and corresponding SDS-PAGE gel stained
with Coomassie Blue on which Artemis has a gel mobility position at ∼ 100 kDa. The protein markers are
in the left-most lane, and the fraction numbers are below each lane.
(B) Superose 12 elution volume plot of Artemis and molecular weight markers. The Artemis identity was
further confirmed by mass spectrometry, and no other proteins were identified by mass spectrometry
(see text).
(C) Fractions across the Superose 12 elution peak were assayed for nuclease activity using the poly(dT)
substrate (JG169). In each reaction, 50nM single-stranded DNA substrate (JG169) was incubated with
the protein(s) indicated above the lane in a 10_l reaction for 30 min at 37 ◦C. After incubation, reactions

38

were stopped and analyzed by 12% denaturing PAGE. Concentrations are as follows: Artemis, 50 nM;
DNA-PKcs, 50 nM, and 0.25mM ATP.

Artemis has endonuclease activity on single-stranded DNA
After we obtained pure and homogeneous Artemis protein from the gel filtration Superose 12
column, we tested its activity on a variety of DNA substrates. Among these, we tested for activity on
ssDNA. We designed three different homopolymer substrates to investigate whether there is any
sequence preference for Artemis activity on ssDNA (Figure 3. 2 ). As p r ev i o u sly re p o rted, Ar te m is ha s 5 ’ exonuclease activity on all of the three substrates. Interestingly, Artemis also shows a pattern of
nucl ease ac tiv i ty o n p o ly (dC) and po l y ( d T) sub strat es that c an n o t b e ex p lai n ed b y 5 ’ e x o n u clea se
activity. This activity exhibits a substantial pyrimidine preference, as it does not act on poly (dA)
substrates (Fig. 3.2 lanes 1-4 versus 5-12). However, those products can either be explained by
sequence- d epend en t 3 ’ e x o n u cleas e acti v it y o n ss D N A o r s equ ence -dependent endonuclease activity on
single-stranded DNA.

39

Figure 3.2. Endonuclease activity of Artemis on single-stranded DNA. In each reaction, 50nM single-
stranded DNA substrates (JG167, JG168, JG169, JG175 or YM8) or 20nM hairpin opening substrate
(YM164) were incubated with the protein(s) indicated above the lane in a 10_l reaction for 30 min at 37
◦C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE. Protein
concentrations are as follows: Artemis, 50 nM; DNA-PKcs, 50 nM. As specified, 0.25mMof ATP and
0.25_Mof YM8/9 were also included in some of the reactions. YM8/9 is a 35 bp blunt end double-
stranded DNA, which is used as DNA-PKcs cofactor. The hairpin DNA substrate (YM164) is sufficient to
stimulate DNA-PKcs directly [22], but YM8/9 further stimulated the autophosphorylation level of DNA-
PKcs and the endonuclease activity of Artemis (lane 15 versus lane 16).

In order to differentiate these two hypotheses, we designed time course assays. We used two
p o ly (d T) sub stra te s : o n e w as lab eled o n t h e 5 ’ e n d , a n d t h e o ther o n t h e 3 ’ e n d . T o be co m p arab le, t h e
3 ’ e n d lab eled sub stra te ha s a c o ld ph o sph at e g ro u p o n t h e 5 ’ e n d . Then, Artemis and DNA substrates
along with DNA-PKcs are incubated for various times (Fig. 3.3). As we can see from both panels, the

40

products in 30 min reactions have the same pattern as in the 6 min, 10 min, and 20 min reactions, but
with increased intensity, which indicates that Artemis has sequence-dependent endonuclease activity
o n ss D N A rath er t h an 3 ’ e x o n u cleas e acti v it y . T h e ss D N A end o n u cleas e ac tiv it y o f Arte m is: D N A -PKcs
seems to have a slight preference for sites closer to the 5' portion of this 30 nt substrate, perhaps
reflecting a preference for how the Artemis:DNA-PKcs loads onto ssDNA substrates.

Fig. 3.3. Time course of endonuclease activity of Artemis and DNA-PKcs on single-stranded DNA.

41

(A) In each reaction, 50nM 5_ labeled single-stranded DNA substrate, JG169, was incubated with 45nM
Artemis, 50nM DNA-PKcs, 0.25mM ATP and 0.25_M YM8/9 in a 10_l reaction for indicated times at 37
◦C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE.
(B) In each reaction, 50nM 3_ labeled single-stranded DNA substrate, JG282, was incubated with 45nM
Artemis, 50nM DNA-PKcs, 0.25mM ATP and 0.25_MYM8/9 in a 10_l reaction for indicated times at 37
◦C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE.

Autophosphorylation of DNA-PKcs regulates the single-stranded DNA endonuclease activity of
Artemis
Artemis alone has a basal level of ssDNA endonuclease activity (Fig. 3.2, lane 6 and 10).
However, with activated DNA-PKcs present in the reaction, this ssDNA endonuclease activity of Artemis
is dramatically increased (Fig. 3.2, lane 8 and 12). To further understand the regulation of the ssDNA
endonuclease activity of Artemis by DNA-PKcs, we carried out additional experiments. It is interesting to
see this regulation is not only dependent on DNA-PKcs, but also on ATP and duplex DNA (Fig. 3.4A, lanes
6, 7, 9 versus lane 8), which suggests that protein phosphorylation plays an important role in this
regulation.
It is clear that the dsDNA endonuclease activity of Artemis requires DNA-PKcs and ATP.
However, the underlying mechanism for activating the endonuclease activity of Artemis by DNA-PKcs is
still a subject of investigation. Recently, a report from Jeggo and colleagues showed that
autophosphorylation of DNA-PKcs is necessary and sufficient to activate the dsDNA endonuclease
activity of Artemis (Goodarzi et al., 2006). In order to elucidate the mechanism of regulation of the
ssDNA endonuclease activity of Artemis by DNA-PKcs, we performed a corresponding experiment (Fig.
3.4B). We pre-autophoshorylated DNA-PKcs in the presence of ATP and unlabeled dsDNA. We then
added this preparation to Artemis (Fig. 3.4B, lane 9). For comparison, in order to prevent DNA-PKcs
from phosphorylating Artemis, we added wortmannin to the Artemis before adding the pre-
autophosphorylated DNA-PKcs (Fig. 3.4B, lane 8). The Artemis ssDNA endonuclease activity was similar

42

for both reactions. This shows that pre-autophosphorylation of DNA-PKcs is sufficient to stimulate the
ssDNA endonuclease activity of Artemis (Fig. 3.4B lane 4 versus lane 8). We have previously shown that
Artemis and DNA-PKcs form a complex (Ma et al., 2002b). The conformational change within DNA-PKcs
upon autophosphorylation must cause a conformational change within the bound Artemis, which
stimulates Artemis activity, similar to what we have previously described for duplex DNA with overhangs
or for hairpin DNA (Ma et al., 2002b). Based on this, we speculate that the DNA endonuclease activity
and dsDNA endonuclease activity of Artemis are properties of the same active site with Artemis.

43

Fig. 3.4. Autophosphorylation of DNA-PKcs regulates the single-stranded DNA endonuclease activity of
Artemis.
(A) In each reaction, 50nM single-stranded DNA substrate,
JG169, was incubated with the reagents indicated above the lane in a 10_l reaction for 30 min at 37 ◦C.
After incubation, reactions were stopped and analyzed using 12% denaturing PAGE. Reagent
concentrations were as follows: Artemis, 50 nM; DNA-PKcs, 50 nM; ATP, 0.25mM; YM8/9, 0.25_M.
(B) In each reaction, 50nM single-stranded DNA substrate JG169 was incubated with the reagents
indicated above the lane in a 10_l reaction for 30 min at 37 ◦C. After incubation, reactions were stopped
and analyzed by 12% denaturing PAGE. Reagents demarcated by dotted rectangles were preincubated
with nuclease buffer on ice for 30 min. Reagent concentrations were as follows: Artemis, 45 nM; DNA-

44

P Kcs , 5 0 n M ; A TP , 0 . 2 5 m M ; Y M 8 / 9 , 0 . 2 5 _M ; w o rt m an n in , 5 _ M . “ W” ind icates w o rtm an n in . “ D ” indicates DMSO. Wortmannin was dissolved in DMSO. An equal amount of DMSO was used as a control
in lanes 5 and 9.

Single-stranded DNA endonuclease activity is intrinsic to Artemis
To verify that this ssDNA endonuclease activity is intrinsic to Artemis, we performed the
immunoinhibition assay with purified anti-Artemis antibody (Figure 3.5). Anti-Artemis antibody
successful ly abo li sh ed the 5 ’ o v erha n g and hai rp in o p eni n g e n d o n u cleas e ac tiv it y o f Arte m is, w h il e heat -
inactivated anti-Artemis purified antibody did not have any effect (Fig. 3.5, lanes 11-14). When DNA-
PKcs is present, anti-Artemis antibody suppressed the ssDNA endonuclease activity of Artemis close to
the basal level (Fig. 3.5, lane 7 versus lane 2). This indicates that anti-Artemis antibody may disrupt a
productive conformational interaction between Artemis, DNA-PKcs and the DNA substrate. This may
explain why there is no effect of the antibody on the 5' exonuclease activity. It is interesting to note that
anti-Artemis antibody up-regulates the ssDNA endonuclease activity of Artemis in the absence of DNA-
PKcs (Fig. 3.5, lane 5 versus lane 2). Since the anti-Artemis antibody itself does not carry any nuclease
activity (Fig. 3.5, lane 10), this suggests that Artemis adopts a conformation with higher activity when
the antibody is bound, which also illustrates that the basal level of ssDNA endonuclease activity is also
intrinsic to Artemis protein.

45

Fig. 3.5. Anti-Artemis antibody down-regulates the single-stranded DNA endonuclease activity of
Artemis in the presence of DNA-PKcs. Artemis and purified anti-Artemis antibody was preincubated with
nuclease buffer on ice for 20 min. Double-underlined purified anti-Artemis antibody amounts designate
that this is heat-inactivated (100 ◦C×5min) and used as negative control. After preincubation, 50nM
single-stranded DNA substrate, JG169, or 20nM hairpin opening substrate, YM164, was added to each
10_l reaction along with other reagents indicated above the lane. Reactions were further incubated for
30 min at 37 ◦C. After incubation, reactions were stopped and analyzed using 12% denaturing PAGE.

46

Reagent concentrations were as follows: Artemis, 50 nM; DNA-PKcs, 50 nM; ATP, 0.25mM; YM8/9,
0.25_M. Amounts of purified anti-Artemis antibody are specified.

We and others have observed an interesting phenomenon that Artemis can acquire dsDNA
endonuclease activity in the absence of DNA-PKcs under Mn
2+
condition (Huang et al., 2009)(Lu, H &
MRL, unpublished). The single-stranded DNA endonuclease activity of Artemis was also tested under
Mn
2+
condition (Fig. 6). Interestingly, Mn
2+
up-regulates the ssDNA endonuclease activity of Artemis
relative to the level with Mg
2+
, similar to the effect by DNA-PKcs (Fig. 3.6, lane 3 versus lane 2). This
further demonstrates that the ssDNA endonuclease activity is intrinsic to Artemis, and it is closely
related to its dsDNA endonuclease activity.

47

Fig. 3.6. Artemis single-stranded DNA endonuclease activity with different divalent
cations. In each reaction, 50nM single-stranded DNA substrate, JG169, was incubated with the reagents
indicated above the lane with different divalent cations (MgCl2, MnCl2 and ZnCl2) in a 10_l reaction for
30 min at 37 ◦C. After incubation, reactions were stopped and analyzed by 12% denaturing PAGE.
Reagent concentrations were as follows: Artemis, 17.5 nM; DNA-PKcs, 50 nM; ATP, 0.25mM; YM8/9,
0.25_M; MgCl2, 10mM; MnCl2, 5mM; ZnCl2, 1mM. Artemis fractions used in this experiment were
purified using Ni-NTA followed by a DEAE-Sepharose column, without the gel filtration step.

3.5 DISCUSSION
Single-Stranded Endonuclease Activity of Artemis and Artemis:DNA-PKcs

48

Using Artemis purified on myc-antibody immunobeads, we previously showed that
Artemis:DNA-PKcs can cut 5' overhangs, 3' overhangs, and hairpins (Ma et al., 2002b). Artemis alone
also appears to have 5' exonuclease activity (Ma et al., 2002b; Ma et al., 2005c; Povirk et al., 2007;
Yannone et al., 2008). We have now further purified Artemis using high resolution size exclusion
chromatography. This more highly purified preparation has all of the nucleolytic properties that we
described previously. But in addition, we have been able to elucidate a ssDNA endonuclease activity of
Artemis alone. This activity is stimulated by DNA-PKcs and modulated by Mn
2+
in a manner that is quite
similar to modulation of Artemis:DNA-PKcs activity on dsDNA substrates. In addition, purified
antibodies raised against Artemis modulate this ssDNA endonuclease activity at concentrations that are
similar to those that modulate the previously described activities of Artemis:DNA-PKcs. Hence, the
ssDNA endonuclease activity is intrinsic to Artemis.
Could the single-stranded endonuclease activity account for the 5' exonuclease activity of
Artemis alone? This does not appear to be the case because the 5' exonuclease activity of Artemis is
not stimulated by DNA-PKcs. In fact, the 5' exonuclease activity of Artemis is the only activity of Artemis
that is not responsive to DNA-PKcs. This raises the question of whether the 5' exonuclease could be a
contaminant. In this regard, point mutants of Artemis that destroy the overhang and hairpin opening
activities of Artemis:DNA-PKcs do not affect the 5' exonuclease activity of Artemis (Pannicke et al.,
2004). However, the point mutants thus far may affect the binding of the DNA substrate rather than the
catalytic active site, which is still not fully defined (Pannicke et al., 2004). The binding of substrates for
the 5' exonuclease activity may be different than the binding for endonucleoytic cleavage. In addition,
the 5' exonucleolytic activity is present not only after the Ni-NTA affinity tag step and DEAE-Sepharose
chromatography step, but also after the additional high resolution gel filtration step, where Artemis runs
at a size that is much larger than common contaminating exonucleases. Finally, mass spectrometry on
the Artemis peak from the gel filtration column identify no other proteins than human Artemis (not

49

even insect cell proteins) Therefore, we still favor the view that a 5' exonuclease activity is intrinsic to
Artemis, but certainty on this point must await active center identification and further mutagenesis.
Model Unifying All of the Endonucleolytic Activities of Artemis
All of the endonucleolytic activities of Artemis appear to function similarly. These lead to a
model in which Artemis prefers ~4 nt of single-stranded DNA to bind any type of DNA substrate (Fig.
3.7). After binding, Artemis nicks on the 3' side of its binding site. When DNA-PKcs is present, Artemis
exists in an Artemis:DNA-PKcs complex. Upon autophosphorylation of DNA-PKcs, a conformational
change occurs which stimulates Artemis endonucleolytic activity on all single-stranded substrates. DNA-
PKcs causes the Artemis:DNA-PKcs complex to favor junctions of ssDNA and dsDNA. All of the
endonucleolytic properties of Artemis and Artemis:DNA-PKcs are consistent with this model.

Fig. 3.7. Artemis endonuclease activity on single-stranded DNA and double-stranded DNA. The model
shows Artemis as a grey-filled oval and DNA-PKcs as a rectangle. Artemis appears to prefer to bind to 4
nt of ssDNA and cut immediately 3_ . This behavior is most noticeable for duplex substrates with
overhangs. But for ssDNA substrates, the preferred 4 nt zone can be anywhere. We have previously
raised the possibility that DNA-PKcs might bend the single-stranded overhang into a near-hairpin
conformation, in order for the associated Artemis to cut [7,8]; but for simplicity, we have drawn the
overhangs and ssDNA as artificially straight (unbent).

50

Pyrimidine preference and relationship to such preferences by polymerase mu, ligase IV and Artemis
The endonucleolytic action of Artemis on ssDNA is much greater for poly(dC) and poly(dT)
substrates than a poly(dA) substrate. This preference also exists for Artemis:DNA-PKcs action on
overhangs. For example, a 3' overhang of four Gs followed by six Ts is not cut nearly as efficiently as one
with entirely consecutive Ts (data not shown). Proteins that bind ssDNA often show biases for either
purines or pyrimidines (Kim et al., 1992).
It is interesting that the XRCC4:DNA ligase IV complex can ligate ssDNA, but this is only efficient
for poly (dT) substrates (Gu et al., 2007c). Moreover, polymerase mu, one of the primary polymerases
for NHEJ, can add dC or dT in a template-independent manner much more efficiently than dA or dG (Gu
et al., 2007a). Hence, the nuclease, polymerase, and ligase of NHEJ all have pyrimidine preferences. It
will be interesting to see if there is some unifying reason for such consistent preferences by the various
NHEJ enzymatic activities.

51

Chapter 4 5’ Exonuclease Activity of Artemis
4.1 Abstract:
Artemis is a member of the metallo- -lactamase protein family. Artemis has endonuclease
activ it y at D N A h air p in s an d at 5 ’ an d 3 ’ D N A o v erha n g s, an d thi s end o n u cle o ly tic acti v it y is d epend en t upon DNA- P Kcs. There h a s b ee n u n cert ain ty ab o u t whet h er Ar t em is als o h as 5 ’ ex o n u clea se acti v ity o n single-stranded DNA and 5' overhangs, b ecaus e thi s 5 ’ ex o n u cleas e is n o t d epe n d ent u p o n D N A -PKcs.
Here, we show that a point mutant of Artemis at a putative active site residue abolishes both the
end o n u cleas e acti v ity a s we ll as the 5 ’ ex o n u clea s e activ i ty . B o th the end o n u clease acti v it y an d 5 ’ exonuclease activity can be blocked in parallel by three different specific small molecule inhibitors of
Ar te m i s. M o re o v er, d ival e n t catio n e ffect s o n the 5 ’ ex o n u cl ease an d the en d o n u clease p aral l el o n e another. We conclude that the 5'-exonuclease is intrinsic to Artemis, and this activity contributes to
Artemis action during NHEJ.

4.2 Introduction:
Artemis is a member of the metallo- -lactamase superfamily of nucleases. The metallo- -
lactamase family is characterized by a conserved metallo- -lactamase and -CASP domains, and their
ability to hydrolyze DNA or RNA. Two other mammalian members of the same family, SNM1A and
SNM1B (Apollo), function in DNA interstrand cross-link repair (Dronkert et al., 2000) and protection of
telomeres (van Overbeek and de Lange, 2006). Artemis, also known as SNM1C, functions in the
nonhomologous end joining (NHEJ) repair pathway. Artemis has endonuclease activity on 5' and 3' DNA
overhangs and at DNA hairpins, and these are all dependent upon DNA-PKcs (Ma et al., 2002b). In the
context of V(D)J recombination, the Artemis:DNA-PKcs complex is required for the opening of hairpins
generated by the RAG complex (Ma et al., 2002b). Patients lacking Artemis suffer from severe combined
immunedeficiency (SCID)(Moshous et al., 2001). Canonical NHEJ requires nucleases for resection of

52

broken DNA ends, polymerase to fill in the gaps, and the XLF:XRCC4:DNA ligase IV complex for ligation
(Lieber, 2010). Artemis is at least one of the nucleases participating in this process and is responsible for
a subset of end processing during NHEJ in response to ionizing radiation (Riballo et al., 2004b). Patients
or cells lacking Artemis are deficient for repair caused by ionizing radiation (IR) (Moshous et al., 2001;
Rooney et al., 2003a).
In a biochemical system consisting of purified proteins of the canonical NHEJ pathway, Artemis,
in complex with the serine/threonine kinase DNA-PKcs, removes 5' and 3' overhangs in order to join
duplex DNA ends (Lu et al., 2008). Artemis:DNA-PKcs is also able to open hairpins both in vivo and in
vitro. Artemis opens hairpins preferentially 2 nucleotides past the tip of a perfect hairpin (Ma et al.,
2002b) . At a 5 ’ o v erha n g , Ar te m i s: D N A -PKcs cuts directly at the double-strand/single-strand junction,
whereas o n a 3 ’ o v erha n g , p refere n tial c lea v ag e o c cur s 4 nu cle o tid es a way f ro m the do u b le strand/single-stranded DNA junction (Niewolik et al., 2006b). All of the endonuclease activities of
Artemis on duplex DNA ends can be unified by a model invoking binding to the single- to double-strand
transition for a length of 4 nt along the single-stranded portion, followed by nicking 3' of the transition.
In addition to its endonucleolytic action on the single-strand to double-strand transitions
(overhangs and hairpins) of duplex DNA, Artemis alone has a weaker single-stranded DNA endonuclease
activity (Gu et al., 2010). Like the endonuclease activity of Artemis on duplex DNA substrates, the action
of Artemis on entirely single-stranded DNA is stimulated by the addition of DNA-PKcs, which forms a
stable complex with Artemis. Our model for Artemis action on ssDNA proposes binding by Artemis,
followed by nicking, and these activities of Artemis are increased by DNA-PKcs.
P u rif ied Art em i s pro te in als o app ears t o hav e 5 ’ e xo n u clease acti v ity , w h ich i s D N A -PKcs
independent. The 5' exonuclease activity co-fractionates with the known endonucleoltyic activities (Gu
et al., 2010). Exonucleases are abundant in cells and are common contaminants. As other 5'
exonucleases are purified away, the apparent level of 5' exonuclease activity falls, but not to zero, and

53

the question arises whether the residual level of 5' exonuclease activity is intrinsic to Artemis or a
contaminating enzyme that is difficult to remove.
Using a three step purification process consisting of phosphocellulose, nickel-NTA agarose, and
hydroxyapatite, one group separated the usual bulk of other cellular 5 ’ ex o n u clea se acti v it y a way fr o m the endonuclease activity of Artemis (Pawelczak and Turchi, 2010). They concluded that the 5'
exonuclease activity was not intrinsic to Artemis; however, we would note that some 5' exonuclease
activity is still apparent from their assay gels (Pawelczak and Turchi, 2010). For example, they tested
their purified Artemis on a single-stranded 5' radiolabeled DNA substrate to assay for 5' exonuclease
activity (Fig. 5B of ref. 11). Though they could detect no 5' exonuclease activity, their assay was
insensitive to the extent that they could not detect any of the single-stranded endonuclease of Artemis
(Pawelczak and Turchi, 2010). This was due in part to exposure intensities of their gels that were 10 to
20-fold lower than what we use to detect both single-stranded endo- and 5' exonuclease activities (Gu
et al., 2010). Despite the fact that their gels show apparent residual 5' exonuclease activity at their
darker radiolabel exposures (Fig. 5C, lanes 10 & 11 in ref. 11), they nevertheless concluded that Artemis
has no intrinsic 5' exonuclease activity (Pawelczak and Turchi, 2010).
Like other members of the metallo- -lactamase family, it is thought that Artemis requires the
binding of divalent cations to two co-catalytic portions of the active site for catalysis (Pannicke et al.,
2004). Each catalytic site may hold one Mg
2+
, Mn
2+
, or Zn
2+
ion. In this family, the ligands for the
divalent cations are generally conserved, and consist of a combination of histidine and aspartic acid
residues (Aravind, 1999). Our previous studies have demonstrated that by mutating these conserved
residues, the endonuclease activity of Artemis:DNA- P K cs can be abo li shed . Ho we v er, the 5 ’ ex o n u cl ease activity appeared to remain (Pannicke et al., 2004). T h is raised t h e po s sib il ity that t h e 5 ’ ex o n u cleas e
activity and the endonuclease activity of Artemis use two independent catalytic sites. However, an
alternative explanation is that previous attempts at testing point mutants was carried out using 293T

54

cells, a system known for high levels of background nucleases, and this failed to remove a contaminating
5' exonuclease.
Here, we present a point mutant of Artemis created using a baculovirus expression system. This
mutant abolishes both the endonuclease as well as the 5' exonuclease activity of Artemis. We have also
identified inhibitors that affect the Artemis endonuclease and the 5' exonuclease activities similarly.
These and other findings shed light on the catalytic sites involved in Artemis activity as well as expands
the range of substrates that Artemis can act upon. We also discuss these findings in the context of
NHEJ.
4.3 Material and Methods:
Oligonucleotides & DNA Substrates - Oligonucleotides used in this study were synthesized by Operon
Biotechnologies,Inc. (Huntsville, AL) and Integrated DNA Technologies, Inc. (San Diego, CA). We purified
the oligonucleotides using 12% or 15% denaturing PAGE and determined the concentration
spectrophotometrically. D N A su b strate 5 ’ end lab elin g w as d o n e with [ -32P]ATP (3000 Ci/mmol)
(PerkinElmer Life Sciences, Boston, MA) and T4 polynucleotide kinase (New England Biolabs) according
to the m an u facturer ’s instr u ctio n s. Su b stra te s we r e in cub ate d w i th [ -32P]-ATP and T4 PNK for 30 min
at 37 °C. T4 PNK was subsequently inactivated by incubating samples at 72 °C for 20 min.
Unincorporated radioisotope was removed by using G-25 Sephadex (Amersham Biosciences, Inc.) spin-
column chromatography. For the hairpin substrate, YM164-labeled oligonucleotide was diluted in a
buffer containing 10 mM Tris-hydrochloride (pH 8.0), 1 mM EDTA (pH 8.0), and 100 mM NaCl, heated at
100 °C for 5 min, allowed to cool to room temperature for 3 h, and then incubated at 4 °C overnight. To
create the 3 ’ end lab eled J G2 7 7 / J G 2 2 6 ddG, we first incubated JG226 with ddGTP and TdT () for 1h at 37
°C to add an un reacti v e dd G t o t h e 3 ’ end . TdT was in activ a te d by incubating samples at 72 °C for 20
min. Unincorporated ddGTP was removed by using G-25 Sephadex spin-column chromatography.

55

JG226 ddG was then annealed to JG277 by boiling at 100 °C for 5 min, allowed to cool to room
temperature for 3 h, and then incubated at 4 °C overnight. This su b strat e was t h en in cub ate d with [α -
32P]-dTTP (3000 Ci/mmol) and Kle n o w ( 3 ’ ex o m in u s) p o ly m eras e (N e w Eng lan d Bio lab s) f o r 30 min at
37 °C to perform fill-in synthesis on JG277 to create a blunt-end. Unincorporated radioisotope was
removed by using G-25 Sephadex spin-column chromatography. Klenow polymerase was subsequently
inactivated by incubating samples at 72 °C for 20 min. The sequences of the oligonucleotides used in
thi s st u d y a re as fo ll o w s: J G 1 6 9 5 ’ -TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT- 3 ’. J G 1 6 7 5 ’ – AAA AAA
AAA AAA AAA AAA AAA AAA AAA AAA - 3 ’. YM164, 5_-TTT TTG ATT ACT ACG GTA GTA GCT ACG TAG CTA
CTA CCG TAG TAA T- 3 ’. J G 2 2 6 5 ’ -ACG AGC CCG ATC CGC TTG ACC AGT AGT CTA GCA CGT GAC GAT TGC
ATC CGT CAA GTA AGA TGC AGA TA CTT AC- 3 ’ JG 2 7 7 5 ’ -TTT TTT CCC CGT TAA GTA TCT GCA TCT TAC
TTG ACG GAT GCA ATC GTC ACG TGC TAG ACT ACT GGT CAA GCG GAT CGG GCT CG-3’

Protein expression and purification - Arm14 recombinant baculovirus with a C-terminal 8x histidine tag
was cloned into baculodirect vector and transfected into SF21 according to protocol (Invitrogen). Virus
was harvested after 48hrs and used to infect SF9 cells in log phase growth. Wild type Artemis was
purified as previously described (Gu et al., 2010). Harvested Arm14 infected SF9 cells were resuspended
in Ni-NTA binding buffer (50mM NaH
2
PO
4
(pH 7.8), 0.5M KCl, 2mM beta-mercaptoethanol, 10% glycerol,
0.1% Triton X-100, and 20mM imidazole (pH 7.8)) supplemented with protease inhibitors. Cells were
lysed by sonication and applied to Ni-NTA agarose resin (Qiagen). Bound Arm14 protein was eluted off
using 250mM imidazole. Eluted protein was directly applied to DEAE-sepharose resin (GE Healthcare).
Unbound soluble f r ac o n w as c o ll ect ed and dia ly z ed ag ain s t Su p er o se 1 2 b u er ( 5 0 0 m M K Cl , 25 m M T ris- H Cl , 1 m M D TT ). Ar m 1 4 fr ac o n s e lu t ed fr o m Su p er o se 1 2 w e r e f r o z en d o wn and s t o r ed a t - 8 0 C.
Expression and purification of DNA-PKcs from Hela cells has been previously described. The expression
and purification of Artemis and DNA-PKcs has been previously described.

56

Circular Dichroism - All CD studies were conducted on Jasco-810 Spectro Polarimeter. Wave length of
260nm was used as the start and 190nm used as the end reading. A data pitch of 0.5nm is used with
continuous reading and accumulation of 20.

Biochemical nuclease assay - The in vitro DNA nuclease assay were performed in a volume of 10ul with
a buffer composition of 25mM Tris-HCl (pH 7.5), 10mM KCl, 10mM MgCl
2
, and 1mM DTT. In the
reaction, 50nM of single-stranded DNA or 20nM of hairpin DNA was mixed with 50nM of Arm14 or
50nM of wild-type Artemis. In specified cases, 42nM of DNA-PKcs was added with 0.5mM ATP.
Reactions were then incubated at 37°C for 1h. After incubation, reactions were stopped, and analyzed
on a 12% denaturing PAGE.

4.4 Results:
Co-Fractionation of WT Artemis with the 5' Exonuclease Activity and Comparison with an Active Site
Point Mutant - Artemis has a well-documented set of endonuclease activities, all of which are
stimulated by DNA-PKcs. The 5' exonuclease activity of Artemis co-fractionates with the Artemis
endonucleolytic activities. The 5' exonuclease activity is not stimulated by DNA-PKcs, and for this
reason, we wondered whether the 5' exonuclease is intrinsic to Artemis or not (see Introduction).
We created a point mutant of Artemis by mutating the conserved histidine residue at amino acid
115 to an alanine (Fig. 4.1A). This was subsequently cloned into a baculodirect vector (Invitrogen) to
generate ARM 14 baculovirus. Human Artemis-His and Arm14 mutant were both expressed in SF9 insect
cells using a baculovirus expression system. Both wild-type Artemis and the Arm14 mutant were
purified using Ni-NTA and Mono-Q, followed by a Superose 12 gel filtration column. Purified Arm14

57

mutant elutes as a relatively sharp peak off of the Superose 12 column and runs at 100kDa on a 8% SDS-
PAGE (Fig. 4.1 B). The Arm14 mutant runs at the same position (~100kDa) as the wild-type Artemis on
8% SDS-PAGE.

Figure 4.1A. Size exclusion chromatography of a single amino acid point mutant of Artemis, the
ARM14 mutant. Arm14 mutant was constructed by creating a point mutation in the putative active site
of Artemis, specifically histidine 115 to alanine, followed by cloning into a baculodirect vector.

Figure 4.1B. Size exclusion chromatography of a single amino acid point mutant of Artemis, the
ARM14 mutant. SDS-PAGE gel stained with Coomassie Blue on which ARM14 mutant protein has a gel
mobility of ~100kDa

58

To correlate enzyme activity with the active fractions, we tested the activity of relevant Arm14
fractions collected from the Superose 12 column on a DNA hairpin substrate to determine whether the
protein was active. The wildtype Artemis endonuclease activity also elutes as a symmetrical and defined
peak. Unlike the wild type Artemis, the single point mutation at histidine 115 abolished all
endonuclease activity as well as hairpin opening activity. The peak ARM14 showed no detectable
endonuclease activity (Fig. 4.1C, lane 2- 5 ). Fu rther m o re, n o 5 ’ e x o n u cleas e ac t ivity was detected, as
demonstrated by the lack of a band at the 1 nucleotide position at the bottom of the gel.

Figure 4.1C. Size exclusion chromatography of a single amino acid point mutant of Artemis, the
ARM14 mutant. Fractions across the Superpose 12 peak were collected and assayed for nuclease
activity. In each reaction 50nM of ARM14 mutant was incubated with 50nM of ssDNA polyT substrate
(JG169) in a 10ul reaction. Reactions were incubated for 60 min at 37°C. The single-nucleotide
(Klenow) ladder was generated by adding one unit of Klenow with 50nM of substrate (JG169). Klenow
reactions were incubated for 5 min, 10 min, 30 min, 60 min, and 90 min, and the reactions were stopped
and analyzed on 12% denaturing PAGE. The 1 nt position in the Klenow ladder is weaker than the larger
products in the ladder.

59

To differentiate between the loss of activity due to inability to retain divalent cations at the
catalytic site versus protein instability/unfolding, we conducted circular dichromism (CD) on the wt
Artemis, on ARM14, and on completely heat-denatured Artemis. We found that the wt Artemis and
ARM14 had similar CD profiles. Hence, purified Arm14 mutant has a conformation that is
indistinguishable from wt Artemis by CD.

Arm14 mutant does not retain hairpin opening, endonuclease, or 5’ exonuclease activity - We
compared the activity of Arm14 with wild-type Artemis on a hairpin substrate. We designed a single-
stranded DNA substrate that is able to fold back on to itself upon annealing to form a hairpin substrate
(YM164). The substrate itself is devoid of any nuclease contamination (Fig. 4.2A, lane 1). The addition
o f Ar m 1 4 t o the s u b stra te d id no t g ener ate any h air p i n o p eni n g , e n d o n u clease , o r 5 ’ -exonuclease
products (Fig. 4.2A,lane 3). Wt Artemis is inactive in Mg
2+
solutions lacking DNA-PKcs. We added DNA-
PKcs to the reaction containing Arm14 to see whether its nuclease activity can be stimulated. Unlike wt
Artemis, we found that the addition of DNA-PKcs did not result in any additional nuclease activity by
Arm14 (Fig. 4.2A, lane 3). As an added precaution we tested whether our DNA-PKcs protein was free of
nuclease contamination by running DNA-PKcs alone with the substrate (Fig. 4.2A, lane 5). Our DNA-PKcs
prep is devoid of any nuclease contamination; therefore, any nucleolytic product formed must be
generated by Arm14 or wt Artemis. As previously reported, wt Artemis has hairpin opening,
end o n u cleas e acti v it y , as w ell as 5 ’ e xo n u cl ease act i v it y ( Fi g . 4.2A, lane 4).

60

Figure 4.2A. Endonuclease and 5’ exonuclease activity in WT Artemis and the Artemis point mutant,
ARM14. 50nM of hairpin DNA substrate (YM164) was incubated with 50nM Artemis or 50nM of ARM14
in a 10ul reaction. Reactions were incubated at 37°C for 60min, and the reaction buffer contains 25mM
Tris pH 7.8, 10 mM KCl, 10 mM MgCl2, and 1 mM DTT. In lanes where DNA-PKcs is added, 50nM of
DNA-PKcs, 0.5mM of ATP, and 0.25uM of cold DNA YM8/9 were also added. Lane 1 contains only the
hairpin substrate. Lane 2 contains Artemis, which does not have endonucleolytic activity without DNA-
PKcs being present, but it ha s some 5 ’ e x o n u cle o ly tic a ctiv it y . L an e 3 is the p o in t m u tant o f Arr te m is. Lane 4 wt Artemis plus DNA- P Kcs, w h ich g enerat es a h air p in o p ened p r o d u ct, an d 5 ’ o v erha n g c lea v ag e
products. Lane 5 is the control lane containing DNA-PKcs, but no Artemis.

We also tested the activity of ARM14 on a single-stranded DNA substrate, specifically a single-
stranded poly dT substrate (Fig. 4.2B lane 3). Consistent with our previous work, wt Artemis, with the
addition of DNA-PKcs, generates a ladder of products on this substrate ending at the 1 nucleotide
position (Fig. 4.2B, lane 2). The addition of DNA-PKcs to the point mutant, ARM14, in an identical
reaction buffer was unable to generate any cleavage products at any positions including the 1
nucleotide position (Fig. 4.2B, lane 3). Hence, the ARM14 mutant does not have any endonuclease
activ it y o r 5 ’ ex o n u cl ease a ctiv it y .

61

Figure 4.2B. Endonuclease and 5’ exonuclease activity in WT Artemis and the Artemis point mutant,
ARM14. 50nM of single stranded DNA substrate (JG169) was incubated with 50nM Artemis or 50nM of
ARM14 in a 10ul reaction. Reactions were incubated at 37C for 60min. In lanes were DNA-PKcs is
added, 50nM of DNA-PKcs, 0.5mM of ATP, and 0.25uM of cold DNA YM8/9 were also added. Both sets
of reactions were stopped and analyzed by 12% denaturing PAGE.
The endonuclease and 5’ exonuclease activity of Artemis can be inhibited in parallel by selective
compounds
We further tested the ability of selective compounds to concurrently inhibit the hairpin opening,
end o n u cleas e, an d 5 ’ e xo n u clease a ctiv i ty o f Ar te m is. T h e t w o c o m p o u n d s w ere id entifi ed by can d id ate approaches using a high-through put screen. In the absence of any inhibitory agent, Artemis has
end o n u cleas e acti v it y as w ell as 5 ’ e xo n u cl ease activity on a single-stranded homopolymer substrate
(Fig. 4.3A, lane 2). However, at 20 uM of compound 0104713, the endonuclease activity of Artemis is
ab o li shed , and t h e 5 ’ e xo n u clease a ctiv i ty o f Ar te m is i s greatl y dimini sh ed (Fig . 4.3A, lane 4). (Fig. 4.3A,
lane 4 ) Additionally we test the effect of the 2 inhibitors on a ssDNA substrate composed of
homopolymers of dT. On the ssDNA substrate wild type Artemis is able to generate a ladder of
endonuclease product as well as an exonuclease product at the 1nt position (figure 4.3B, lane 2). The

62

addition of 20uM of compound 0104713 was able to completely abolish the endonuclease and the
exonuclease activity of Artemis on a ssDNA (Figure 4.3B, lane 3). The addition of 20uM of compound
0095116 was able to partially inhibit the endonuclease and the exonuclease activity of Artemis).

Figure 4.3A. Small molecule inhibitors block both the 5' exonuclease and the endonucleolytic
activities of Artemis on hairpin DNA. Addition of compound 0104713 inhibits th e end o n u cle ase an d 5 ’
exonuclease activity of Artemis. 50nM of hairpin DNA substrate (YM164) was incubated with 60nM
Artemis in a 10ul reaction. Reactions were incubated at 37C for 60min. 20uM of compound 0095116 or
20uM of compound 0104713 were added in indicated reactions.

63

Figure 4.3B. Small molecule inhibitors block both the 5' exonuclease and the endonucleolytic
activities of Artemis on ssDNA. Ad d itio n o f co m p o u n d 0 1 0 4 7 1 3 inh ib its t h e en d o n u clease and 5 ’ exonuclease activity of Artemis. 50nM of ssDNA substrate (JG169) was incubated with 60nM Artemis in
a 10ul reaction. Reactions were incubated at 37C for 60min. 20uM of compound 0095116 or 20uM of
compound 0104713 were added in indicated reactions.
To separate the effect of endonuclease act i v ity o f Art em is act in g o n t h e t er m in al 5 ’ nu cle o tid e
versus that of the exonuclease activity of Artemis, we utilized a linear poly dA substrate which cannot be
processed by the single-stranded endonuclease activity of Artemis. The exonuclease activity of Artemis
is thought to be DNA-PKcs independent. We tested the exonuclease activity of Artemis in the presence
and absence of DNA-PKcs (Fig. 4.3C, lane 2 versus lane 3). Consistent with previous findings, we do not
see any difference in the exonuclease activity of purified Artemis with or without DNA-PKcs. The
addition of 20uM of the inhibitor 0104713 was able to inhibit the exonuclease activity of Artemis as
demonstrated by the lack of 1nt product exonuclease product (Fig. 4.3C, lane 4).

64

Figure 4.3C. Addition of a small molecule inhibitor of Artemis inhibits the 5’ exonuclease activity of
Artemis on ssDNA (Poly A substrate). The endonuclease activity of Artemis does not act on a single-
stranded poly A substrate. 60nM of single-stranded DNA substrate (JG167) was incubated with 60nM of
Artemis in addition to 75nM of DNA-PKcs, 0.5mM ATP where applicable. 20uM of the inhibitor 0104713
was added to the indicated reaction. Reaction s w er e in cub a t ed f o r 1 h r a t 3 7 °C.

We also tested a previously known inhibitor of Artemis, ampicillin, across a range of
concentrations (Supplemental Figure 3). At 10mM, ampicillin was also able to completely abolish both
the endonuclease and the exonuclease activity of purified Artemis. If the 5' exonuclease were a
contaminant, we would not expect both Artemis endonuclease and the contaminating exonuclease to
be inhibited by the same drugs, both ampicillin and compound 0104713. Neither of these drugs inhibits
mung bean nuclease to a substantial level.
In related studies, in 10 mM MgCl
2
reaction buffers, addition of ZnCl
2
above 10 uM inhibit
Artemis endonuclease activity (Suppl. Fig. 4). It has been shown previously in certain cases that excess
zinc can inhibit the enzymatic activity of zinc dependent enzymes. This is thought to be caused by
excess zinc-monohydroxide binding to the active form of the zinc dependent enzyme preventing the
catalytic site from binding active zinc (Larsen and Auld, 1989). This same phenomenon is observed for

65

the e n d o n u clease act i v it y a s w ell as t h e 5 ’ e x o n u clea s e acti v it y o f Arte m is. T h is i s c o n sis te n t with the 5 ’
exonuclease activity of Artemis requiring the same catalytic site as the endonuclease activity of Artemis.
4.5 Discussion:
Though a 5' DNA exonuclease activity co-purifies with Artemis, the 5' exonuclease is not
dependent on DNA-PKcs, whereas the Artemis DNA endonuclease activites (5' and 3' overhang cleavage
and hairpin opening). This has resulted in uncertainty about contamination.
In this study, we have compared a baculovirus-generated point mutant of human Artemis to the
wild type protein t o anal y z e the na ture o f t h e 5 ’ e x onuclease activity of Artemis. Our study
d em o n strat es tha t the po i n t m u tan t o f Arte m is is d e v o id o f 5 ’ e x o n u cleas e ac tiv i ty in th e p res ence an d absence of DNA-PKcs. A point mutant of Artemis is very unlikely to affect a n y c o n ta m in atin g 5 ’ exonuclease. Additionally the point mutant of Artemis has a similar CD profile as wild type Artemis,
indicating that the point mutant did not lose activity due to denaturation, precipitation, or improper
folding.
We also clearly sho w tha t t h e end o n u cle ase an d the 5 ’ e x o n u cleas e ac tiv it y o f Ar te m i s c an be inhibited by four selective non-chelating chemical inhibitors. It is unlikely that one inhibitor is able to
in h ib it bo th Ar t em is and an un id entifi ed c o n ta m in atin g 5 ’ e xo n u cl ease . I t is even more unlikely that
fo u r di fferen t sel ect i v e inh ib ito rs wo u ld aff ect a c o n ta m in atin g 5 ’ e x o n u clea se a t the s a m e
concentrations as they affect Artemis. The chemicals tested here do not substantially inhibit other
nucleases (i.e. MBN), demonstrating their specificity for Artemis.
Lastly, disruption of the catalytic sites by point mutation or divalent replacement is known to
alter the activity of enzymes in the metallo- -lactamase family of nucleases. We found that the addition

66

of excess zinc to th e rea cti o n bu ffer de creas es th e en d o n u clease as we ll as t h e 5 ’ e x o n u cleas e ac tiv it y o f
Ar te m i s. It i s i m p ro b ab le t h at a c o n ta m in atin g 5 ’ e x o n u clease wo u ld als o ha v e t h is feat u re.
We als o ex p lo r ed the po ssi b il ity t h at t h e 5 ’ ex o n u clea se acti v it y s ee n i n our gels is actually the
resul t o f t h e e n d o n u cl ease activ it y o f Art e m is ac tin g o n t h e 5 ’ t erm in al n u cl eo tid e. W e cr eat ed a po l y dA ssDNA substrate that is not able to be endonuleolyticly processed by Artemis. Despite this, we
noticed a persistence of 5 ’ ex o n u cl ease acti v ity w h en u sin g t h is sub strate , ind icat in g t h at the acti v ity see n at t h e 5 ’ end o n a n o r m al p o ly p y rimidi n e sub str ate is att rib u tabl e t o 5 ’ e x o n u clease ac tiv i ty and not solely due to the endonuclease activity of Artemis.
From the eviden ce p r ese n t ed here we c o n clu d e that t h e 5 ’ ex o n u cleas e ac tiv it y s ee n in p u rif ied Artemis is intrinsic to the Artemis protein and not due to a co-purified contaminating protein.
The metallo- -lactamase family of nucleases uses several conserved catalytic residues to
coordinate divalent cations. The two most conserved residues within the catalytic site are histidine and
aspartic acid (Aravind, 1999). We generated a point mutant, ARM14, which changed a conserved
h istid in e resid u e to an alan in e resid u e. If bo th t h e en d o n u clease act i v it y o f Arte m is an d 5 ’ ex o n u cl ease activity of Artemis shared a common catalytic site, by disrupting the divalent ligand in the site there
should be a c o rr espo n d in g d ecre as e in bo th end o n u cl ease a s w ell as the 5 ’ e xo n u clease a ctiv i ty o f
Artemis.
We have previously purified ARM14 mutant from 293T cells using a 2-step purification process.
While this point mutant is inactive for the hairpin opening and endonuclease activity on a hairpin
sub strate , thi s v ersi o n o f A RM 1 4 app ear ed t o s till re ta in 5 ’ e xo n u cl ease acti v ity (Pannicke et al., 2004).
We attribute the discrepancies in our findings with insect cell-purified ARM14 in the current
study with our previous data on ARM14 purified from 293T to the high levels of contaminating nucleases
in mammalian cells that are difficult to remove during purification. The baculoviral system for Artemis

67

production here contains much lower levels of background contaminating nucleases, which can be
subsequently removed in our 3-step purification.
Previous studies by others using a hydroxyapartite column and an insect cell expression system
ap p eared abl e to generat e n early 5 ’ e xo n u cl ease -free fractions of Artemis (Pawelczak and Turchi, 2010).
Upon closer examination of their gels, while the 5' exonuclease activity is diminished, there is still a
basal level of activity (Pawelczak and Turchi, 2010) . T h e d ecre a sed 5 ’ e xo n u cl ea se acti v it y o f w t Ar t em i s
in their preparations corresponds to a weaker hairpin opening and endonuclease activity compared to
our preparations of wt Artemis. This suggests that the sensitivity of their assays for detecting the endo-
or 5' exonuclease activities of Artemis is ~10 to 20-fold lower (Pawelczak and Turchi, 2010) than our
assays.
In vitro we find that the ARM14 mutant is catalytically dead for hairpin opening, overhang
cutting, a n d 5 ’ e x o n u clea se act ivi ty and is not stimulated by] the presense of DNA-PKcs. On a single-
stranded substrate, the ARM14 mutant is also inactive for endonuclease as well as 5 ’ ex o n u cleas e
activity (Fig.2B). Additionally using a plate reader and a fluorescent substrate we find that the ARM14
mutant gives no detectable signal (data not shown). This leads us to conclude that histidine 115 is not
o n ly a c ritical re sid u e fo r e n d o n u clease acti v ity bu t al so a critical resid u e f o r the 5 ’ e xo n u cl ease acti v ity of Artemis
Excess Zn
2+
is known to inhibit some enzymes that use zinc divalent cations in their active sites
by creating zinc monohydroxide. We find that addition of Zn
2+
above 10 M inhibits both the
endonuclease and 5' exonuclease activities of wild-type Artemis over the same concentration range, 10
– 100 uM (Suppl. Fig. 4).
Two small molecule inhibitors previously identified through a HTS were tested for their ability to
inhibit the endonuclease and the exonuclease activity of Artemis. These small molecules have only

68

minimal effects on other nucleases such as Mung Bean nuclease (data not shown) and they do not
inhibit micrococcal nuclease (SL and MRL, not shown). This strongly supports our conclusion from the
point
Similar to Artemis, other members of the metallo- -lactamase family also possess various
nuclease activities. CPSF-73 an enzyme involved in the 3' end processing of histone pre-mRNA has been
shown to possess endonuclease activity as well as 5 ’ exonuclease activity on single stranded RNA
substrates (Dominski, 2007; Mandel et al., 2006). Interestingly, like Artemis, the nuclease activities of
CPSF-73 can also be abolished by the addition of zinc specific chelators suggesting that members of this
family use the same divalent cations in the catalytic site (Ryan et al., 2004).
Biochemical studies have also demonstrated that human SMN1A and SMN1B proteins, also
members of the metallo- - lactam ase fa m il y , ar e cap a b le o f acting as a 5 ’ e x o n u c lease o n sing le -
stranded and double-stranded DNA substrates (Li et al., 2005), and this is its only biochemically defined
enzymatic activity. Mutations of key catalytic residues within the metallo- β-lactamase domain also
ab o li sh the 5 ’ - 3 ’ D N A p ro ce ssin g abi li ty o f hS M N 1 a an d hS M N 1 b . T h eref o re, 5 ’ e xo n u cleas e ac tiv i ty is intrinsic to many members of the metallo- β-lactamase nuclease family, and the catalytic site involved
during DNA processing resides in the metallo- β-lactamase domain. Our study now definitively adds the
5' exonuclease enzymatic activity to Artemis as well.

69

Chapter 5 High-throughput Screen for Artemis Inhibitor
5.1 Abstract:
Acute lymphoblastic leukemia (ALL) cells are pre-B or pre-T cells in origin. During lymphocyte
maturation, productive V(D)J recombination requires the enzymatic cutting by the RAG complex
followed by joining utilizing the components of the NHEJ machinery. DNA-PKcs and Artemis is required
to open the hairpin generated by the RAG complex and is essential for the formation of the coding ends.
If the joining of the RAG generated hairpins could be inhibited by a small molecule, such small molecules
would present a powerful and specific therapy for ALL. Such small molecules would not have toxicity for
any cells in the human body except for ALL cells and the normal early lymphoid cells in patients. The
short-term loss of early developing lymphoid cells for a period of a few months during treatment would
not be of critical concern because the immune system of most patients is already well-established. More
specifically, a direct inhibitor of the nuclease Artemis would have the least adverse impact to non-
lymphoid cells. Hence, we choose Artemis as the target of choice when developing a high-throughput
screen to generate small molecule inhibitors.
5.2 Introduction:
The value of Artemis inhibitors can be understood in terms of the three related endonuclease
activ iti es o f Ar t em i s, which are its ab il ity to clea v e 5 ’ o v erha n g s, 3 ’ o v erha n g s, a n d h air p in s. Ar te m is i s critical for the major pathway of double-strand break repair because its overhang cleavage activity is
needed to resect damaged DNA overhangs generated by chemicals (e.g., topoisomerase II inhibitors)
and ionizing radiation, such as X-rays. Therefore, Artemis inhibitors would be particularly useful for any
human cancers when used in combination with agents that create double-strand breaks (DSBs), such as
therapeutic radiation and type II topoisomerase inhibitors. Human tumor cells lacking Artemis are
highly sensitive to etoposides (Fig 5.1), one of the type II topoisomerase inhibitors, at clinically relevant
concentrations.

70

Figure 5.1 The Human Acute Lymphoblastic Lymphoma Cell Line, Nalm-6, is Hypersensitive to
Etoposides When Artemis is Absent. Wild type and Artemis-/- human ALL cells were compared for
survival in response to VP-16 treatment in culture. The Artemis-/- cells are 10-fold hypersensitive at VP-
16 concentrations (e.g., 300 nM) that are still 4-fold below blood concentrations seen in patients treated
with VP-16 in chemotherapy for ALL and other human cancers. The wild type and Artemis-/- cells are
identical except for the Artemis knockout (Kurosawa et al., 2008).
The hairpin opening activity of Artemis makes it a particularly valuable drug target for human
acute lymphoblastic leukemia (ALL). Hairpins are quite specific DNA structure, which only arise in early
lymphoid cells when RAG proteins create DSBs. Artemis is the only enzyme in vertebrate cells that can
efficiently open these hairpin structures. Hence, lack of Artemis endonuclease results in accumulation
of unresolved DSBs in cells expressing RAG proteins, and these cells appear to die or proliferate more
slowly as they try to repair broken chromosomes by alternative and slower repair pathways (Rooney et
al., 2003b). ALL tumor cells arise from the pre-B and pre-T stages of lymphoid development and
generally express RAG proteins and carry out V(D)J recombination (Fig. 5.2). Blockage of this step will
generate chromosome breaks in ALL cells. With sufficient specificity, such a drug would be highly
selective for the ALL tumor cells, with only a small amount of toxicity to a transient wave of new pre-B
and –T cells (which is almost negligible to the patient). Therefore, small molecule inhibitors of Artemis
would be extremely selective in this particular damage to ALL tumor cells. It is known that human ALL
tumor cells lacking Artemis have slower doubling times (Kurosawa et al., 2008).
0.01
0.1
1
10
100
0 100 200 300
100
Etoposide (nM)
% Survival
WT
ARTEMIS
-/-

71

Figure 5.2. Inhibitors of Artemis Would Result in Double-Strand DNA Breaks in Early Lymphoid Cells,
in Acute Lymphoblastic Lymphoma (ALL) Cells, and in any Somatic Cells Challenged with Double-
Strand Break Inducing Agents (e.g, ionizing radiation). Artemis is the nuclease in vertebrate cells that
opens the hairpins that are generated by the RAG proteins during V(D)J recombination. Inhibition of
Artemis would result in unrepaired hairpin ends. This would cause a double-strand break (DSB) in ALL
tumor cells (and, as a tolerable limited side-effect, in pre-B and pre-T cells). Artemis-/- cells that express
RAG proteins grow more slowly than the +/- or +/+ cells from which they were derived.
5.3 Material and Methods:
Oligonucleotides:
Oligonucleotide SL7, was synthesized by Integrated DNA Technology Inc (San Diego, CA). HPLC
purification on oligonucleotide was also performed by Integrated DNA Technology Inc. SL7
oligonucleotide was diluted in a buffer containing 10 mM Tris-hydrochloride (pH 8.0), 1 mM EDTA (pH
8.0), and 100 mM NaCl, heated at 100 °C for 5 min, allowed to cool to room temperature for 3 h, and
then incubated at 4 °C overnight. The oligonucleotide was stored using aluminum foil as covering to
prevent exposure to light. The sequences of the substrates used in this study are as follows:
SL7 , 5 ’ -/56-FAM/T*T*T*T*TT TTG CCA GCT GAC GCG CGT CAG CTG GC/3BHQ_1/- 3 ’

72

SL2 3 , 5 ’ -T*T*T*T*TT TTG CCA GCT GAC GCG CGT CAG CTG GC- 3 ’,
YM164, TTT TTG ATT ACT ACG GTA GTA GCT ACG TAG CTA CTA CCG TAG TAA T-3’
J G1 6 9 , 5 ’ -TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT – 3’

In vitro nuclease assay
In vitro DNA nuclease assays were performed in a total volume of 10 µl with a buffer
composition of 25 mM Tris-HCl (pH 8.0), 10 mM KCl, 10 mM MgCl
2
, 1 mM DTT and 100 ng/uL BSA. In the
reaction, 50 nM of single-stranded DNA substrate or 20 nM of hairpin substrate were incubated with
Artemis. In reactions were DNA-PKcs is present, 45 nM of purified DNA-PKcs and 0.25 mM of ATP were
added to the reaction. In reactions were lead compound 0095116 and 0104713 were used, 20 uM of
each were added to the reaction. Mung Bean Nuclease was purchased from New England Biolabs.
Where MBN is used, 0.5 units were added in addition with NEB buffer 2 to each reaction. In reactions
where compound 0095113 and 0104713 were added, a final concentration of 20uM of each compound
was used. Reactions were then incubated at 37°C for 60 min. After incubation, reactions were stopped
and analyzed on 12% denaturing PAGE gels. Gels were dried, exposed in a phosphorImager

cassette and
scanned.

Protein expression and purification
Artemis recombinant baculovirus with a C-terminal His tag (and an intervening TEV site) was a
gift from Dr. John Harrington (Athersys, Ohio). Soluble Artemis-his is expressed and purified from Sf9
insect cells. Briefly, harvested Sf9 cells were re-suspended in Ni-NTA binding buffer (50 mM NaH
2
PO
4

73

(pH 7.8), 0.5 M KCl, 2 mM beta-mercaptoethanol, 10% glycerol, 0.1% Triton X-100, and 20 mM imidazole
(pH 7.8)) supplemented with protease inhibitors and lysed by sonication. Cell lysate was applied to Ni-
NTA agarose resin (Qiagen). Artemis-his was eluted off with binding buffer plus 250 mM imidazole.
Eluted fractions were dialyzed against DEAE binding buffer (50 mM Tris-HCl (pH 7.5), 10% glycerol, 2 mM
EDTA, 1 mM DTT, 100 mM NaCl, and 0.02% NP-40), loaded onto pre equilibrated Mono-Q column, and
eluted with a linear gradient of 100mM to 1 M NaCl over 20 mL.
5.4 Results:
We have developed an HTS compatible assay designed to identify small-molecule inhibitors of
the Arte m is e n d o n u cl ease . The p ro p o sed pr i m ar y HTS assay f o r Ar te m is use s a q u ench ed flu o r esc en t 5 ’ -
DNA overhang substrate that generates a fluorescent signal following endonucleolytic cleavage by
Artemis at the overhang position (Fig 5.3 ). Arte m is cle av es the 5 ’ o v erha n g o f t h e lab eled D N A h air p in substrate resulting in the liberation of the 6FAM group from the proximity of the black-hole quencher
(BHQ). N and n nucleotide sequences can be variable as Artemis acts in a sequence-independent
m an n er. T h e s ind ica te s a p h o sph o thi o es te r lin kag e in the 5 ’ o v erha n g . The p r ese n ce o f phosphothioester linkages prevents a false signal due to p o ssib le co n t am in atin g 5 ’ e xo n u cl ease activity
in purified Artemis protein.
Ar te m i s c lea v e s t h e 5 ’ o v er h an g o f t h e lab eled D N A h a irp in sub strate r esultin g in the s epar a tio n of the 6FAM group from the region of the black-hole quencher (BHQ) within which it can absorb light
emitted from 6FAM. The reaction generates a robust, DNA and enzyme concentration-dependent,
fluorescence signal within a 6 uL assay volume in high-density 1536-well format.

74

Figure 5.3 Fluorescent DNA Substrate Design. The substrate contains a 12 nucleotide duplex portion
an d an 8 nu cle o tid e 5 ’ o v er h an g . T h e sub stra te is bo il ed and se lf an n e aled t o ge n erate the hair p in structur e. 4 ph o sph o thi o e ste r b o n d s at the 5 ’ most portion of the overhang prevent possible
contaminating 5 ’ e xo n u cl e ase ac tiv it y . “ N ” po si tio n s d esig n ate t h e an ticip at ed A rtem i s c u t tin g site s o n the substrate.
In collaboration with Drs. Chen-Ting Ma and Eduard Sergienko at the Sanford Burnham institute,
we successfully miniaturized the assay into 1536-well format. The reaction was carried out in 6uL
volume containing 30 nM Artemis, 125 nM of substrate, and 20 µM test compound dissolved in 0.2%
DMSO.
The assay worked properly in 1536-well plates with full utilization of liquid handling equipment
and plate readers available at the CPCCG. After validating the assay performance in 1536-well plates, we
proceeded to screen Library of Pharmacologically Active Compounds (LOPAC-1280; Sigma-Aldrich)
collection at CPCCG. We conclude that the assay is robust, demonstrates acceptable performance in
1536-well format and thus, it is HTS ready.
In collaboration with Dr. Michael Hedrick and Dr. Chris Hassig of the Sanford Burnham Institue,
we further optimized the assay for Artemis concentration and reaction concentration to generate a
linear kinetic curve for our HTS. Using purified Artemis, we tested the robustness of our assay and the
enzyme kinetics in a 1536 well plate format. 60nM and 30nM of purified Artemis was added to our
fluorescent substrate in a total volume of 6ul and the activity was measured over a 120 minutes period.

75

Figure 5.4 Kinetics of Artemis Activity in 1536 Well Plate Format.
Ni-NTA agarose and DEAE-Sepharose purified Artemis were incubated in a 6ul reaction in 10mM Tris-
HCl, 10mM KCl, 1mM DTT, and 10mM MnCl
2
solution. Reactions were conducted in Corning 1536 well
black polystyrene plate and signals were read at the indicated time points using Pherastar Plus
fluorescent reader.
In wells without Artemis, where only the substrate alone, there is a low level of background
fluorescence. When purified Artemis is added to the reaction, there is a 30 min of lag phase before any
fluorescence is detected. We were able to generate a linear curve for activity over time from 30 min to
120 min. As expected, a higher concentration of Artemis protein resulted in higher signal yield as
measured by relative fluorescent unit (RFU).
30,987 compounds were screened using twenty-two 1536 well plates. Compounds were tested
at a final concentration of 10 uM. All plates passed the quality control test with a Z factor greater than
0.5, where the Z factor is a measure of the robustness of an assay for HTS campaign. A Z factor between
0.5 and 1 is acceptable for HTS purposes. The screen achieved a signal to noise ratio (S/N) of 39.9 and
an average plate Z score of 0.83. The initial hit rate was lower than expected; using a cut off of 30%
inhibition, 235 hits were selected for further testing (Fig 5.5). The 235 initial hits were re-tested in
triplicates for confirmation purposes. Of the 235 initial candidate, 153 were reconfirmed as certified
hits. 133 of the compounds were available in sufficient quantity for additional testing to generate a

76

dose response curve. From the dose response curves generated, we picked the three leading candidates
with the lowest level of corresponding IC50 concentration to test in our secondary assays. The three
compounds chosen for further testing were determined to not be promiscuous compounds or possible
chelators.

Figure 5.5 Hit Result, Z score, and Signal to Noise Ratio of MLPCN Screen.
Potential hits were calculated at the 50%, 40%, 35% and 30% inhibition points. Due to the low hit rate, a
30% inhibition cut off was chosen for further analysis.

We tested the compounds for their ability to inhibit the endonuclease activity of Artemis using a
short hairpin substrate. While the compound 0019074 was found to have an effect in the plate reader
assay, we did not notice any inhibition of Artemis activity in our radioactive gel assay (Fig 5.6 lane 3).
We further tested the two remaining compounds on a hairpin substrate in the presence of DNA-PKcs
and 10mM magnesium. The sub strat e c o n tain s a 1 2 n u cleo tid e d u p lex p o rti o n , 8 nu cleo tid e 5 ’
o v erha n g and 4 ph o sph o th io est er b o n d s at the 5 ’ e n d o f t h e o v erha n g (Fig 5.3). The substrate used in
our radioactive gel assay is identical in sequence to the substrate used in the HTS except for the absence
of the 6- FA M at t h e 5 ’ and b lack ho l e q u ench er gro u p at the 3 ’. In ste ad o f the 6 ’ FA M gro u p , the 5 ’ e n d here is radioactively labeled. Without any inhibitor present, wild-type Artemis has strong
Corrected
S/N 39.9 ± 7.0
Z’ 0.82 ± 0.05
Hits (≥ 50%) 51 (0.13%)
H its ( ≥ 4 0 %) 98 (0.25%)
H its ( ≥ 3 5 %) 145 (0.36%)
H its ( ≥ 3 0 %) 235 (0.59%)

77

endonuclease activit y at t h e 5 ’ o v erha n g ( Fi g 5.6, Lane 6). 20uM of compound 095116 was able to
partially inhibit the endonuclease and hairpin opening activity of Artemis (Fig 5.6., Lane 4). Compound
0104713 at 20uM was also able to partially inhibit the hairpin opening as well as the overhang cutting
activities of Artemis (Fig 5.6, Lane 5). This demonstrated the value of the high throughput screen in
identifying physiological relevant candidate leads.

Figure 5.6 Artemis Nuclease Activity In the Presence of Small Molecules identified by MLPCN Screen.
50nM of hairpin DNA substrate (SL23) was incubated with 50nM Artemis in a 10ul reaction. 20uM of
0095116, 0104713, and 0019074 were added respectively. Reactions were incubated at 37°C for 60min.
Reactions were stopped and analyzed by 12% denaturing PAGE.
We then tested the effects of both inhibitors on a longer hairpin substrate (YM164) which
contains no phosphothioester bonds and a 20 nucleotide d u p lex p o r tio n with a 6 n u cleo tid e s 5 ’ overhang. In contrast to the shorter substrate, Artemis was not inhibited by compound 0095116 (Fig
5.7A, Lane 3). However, compound 0104713 retains its ability to block the hairpin opening and the
overhang cutting activities of Artemis (Fig 5.7A, Lane 4). The results indicate that compounds that are
identified through the HTS may function in a substrate specific manner. In vivo, broken chromosomes
ends are exponentially longer than our in vitro substrates. While we cannot recapitulate the size or

78

length of broken DNA ends in an in vitro biochemical system, the preferences of a candidate molecule to
inhibit Artemis in the presence of a longer substrate rather than a shorter one makes the molecule a
better candidate for tertiary screening assays. Additionally, we also tested the activity of both candidate
molecules on a single-stranded homopolymer (poly dT) substrate. At 20uM, compound 0104713 is able
to completely inhibit the endonuclease activity of Artemis (Fig 5.7B Lane 3 versus Lane 2) on an ssDNA.
At the same concentration, compound 0095116 is only able to partially inhibit the endonuclease activity
of Artemis (Fig 5.7B Lane 4 versus Lane 2) on an ssDNA.

Figure 5.7 Effect of inhibitors on Artemis when using hairpin DNA and single-stranded DNA. A) 50nM
of hairpin DNA substrate (YM164) was incubated with 50nM Artemis in a 10ul reaction. 20uM of
0095116 and 0104713 were added respectively. Where DNA-PKcs is present, 50nM of purified DNA-PKcs
and 0.5mM ATP were added. Reactions were incubated at 37°C for 60min. Reactions were stopped
and analyzed by 12% denaturing PAGE. B) 50nM of single stranded DNA substrate (JG169) was
incubated with 50nM Artemis in a 10ul reaction. Where DNA-PKcs is present, 50nM of purified DNA-
PKcs and 0.5mM ATP were added. 20uM of 0095116 and 0104713 were added. Reactions were
incubated at 37°C for 60min. Reactions were stopped and analyzed by 12% denaturing PAGE.

79

To distinguish between Artemis specific inhibition and general inhibition through intercalating or
other DNA binding mechanisms we developed a counter screen using purified Mung Bean nuclease.
Like Artemis, Mung Bean nuclease has hairpin and endonuclease activity on a hairpin substrate and
overhangs (Figure 5.8 Lane 1). The addition of 20uM of compound 0095116 did not inhibit the
endonuclease activity of mung bean nuclease (Fig 5.8 Lane 2). While the addition of 0104713 did not
inhibit the activity of mung bean nuclease, interestingly, it was able to modify the activity of mung bean
nuclease presumably by altering the structure of the DNA substrate resulting in generation of a wider
variety of products than expected (Fig 5.8 Lane 3). The result of the counter screen is encouraging and
indicates that both compounds are specific for Artemis inhibition rather than promiscuous for general
nuclease inhibition or DNA interaction.

Figure 5.8 Effects of Artemis Specific Inhibitors on Mung Bean Nuclease. 50nM of hairpin DNA
substrate (SL23) was incubated with 0.5 units of Mung Bean Nuclease in a 10ul reaction. 20uM of
inhibitor 0095116 and 0104713 were added respectively. Reactions were incubated at 37°C for 60min.
Reactions were stopped and analyzed by 12% denaturing PAGE.

80

5.5 Discussion:
We have been able to adapt our radioactive gel based assay into an in vitro biochemical system
that is suitable for high-throughput screening. Physiologically, Artemis is active and has endonuclease
and hairpin opening activity in the presence of magnesium and DNA-PKcs. However the requirement for
DNA-PKcs is abolished by substituting manganese for magnesium in the reaction buffer. At 10mM of
Mn
2+
, Artemis has equivalent amount of endonuclease and hairpin opening activity as Artemis in the
presence of DNA-PKcs and magnesium. Our assay is able to generate a robust fluorescent signal using
manganese instead of magnesium. This allowed us to circumvent the need to purify large quantities of
DNA-PKcs to use in the HTS. However we acknowledge that in vivo DNA-PKcs is critical to the enzyme
activity of Artemis; therefore, we utilize a secondary radioactive gel based assay to screen potential
inhibitors in the presence of DNA-PKcs and magnesium ions.
Our preliminary HTS screen indicates that there is a lower hit rate than anticipated in a
conventional HTS campaign. By adjusting the criteria of inhibition to 30%, we were able to achieve a hit
rate of 0.59%. A hit rate of 2% is traditionally seen in most high-throughput screens, though nucleases
have not been subject to HTS previously to our knowledge. A low hit rate confers several advantages.
Having a low hit rate reduces the absolute number of false positives generated through the screen as
well as saving time when conducting secondary screens and counter screens. However, having a low hit
rate decreases the odds of finding an ideal compound due to the lower absolute number of potential
candidates that passes any given stage of the screening process. Lastly, despite the low initial hit rate,
we still encountered many promiscuous compounds.
We were able to demonstrate that our primary screen is able to identify viable targets in the
presence of DNA-PKcs and magnesium by conducting secondary gel based assays for verification. In a
fluorescent based gel assay as well as in a radioactive gel based assay, we were able to verify two

81

specific compounds previously identified through the HTS. In the radioactive gel based assay, we were
able to verify the inhibition of Artemis in the presence of DNA-PKcs and magnesium using a variety of
substrates including single-stranded DNA, short hairpin substrates, and long hairpin substrates. Of the
two compounds assayed, one of the compounds was able to inhibit all three substrates, while the 2
nd

compound was only able to inhibit two out of the three substrates. This leads us to conclude that the
structure of the DNA substrate is important and some inhibitors are substrate specific. Furthermore, by
using a secondary gel based assay, we were able to confirm the specificity of the inhibitors for Artemis
by counter screening the identified molecules against other hairpin nucleases such as mung bean
nuclease. Given the low initial hit rate, it is conceivable to conduct our counter screen using a gel based
method. Alternatively, given a higher hit rate, we can adapt our counter screen to a plate based method
by using mung bean nuclease with our fluorescent substrate.
One of the key challenges in designing a non-cellular based high-throughput screening method
is the demand for active purified protein. Using a baculovirus expression system we were able to
produce 800ug of Artemis per liter of SF9 insect cell culture. Due to the fact that proteins are lost during
each column purification step, we attempted to minimize the protein loss by using a two-step batch
purification process consisting of Ni-NTA agarose followed by DEAE Sepharose. While this method is
able to produce clean and active Artemis protein, the protein yield on a per cell basis is less than
optimal. Any screen conducted using this method would require the generation of large batches of
insect cell culture. By optimizing the codon of Artemis protein for insect cell expression, our
collaborator at Southern Research Institute, Dr. Rongbao Li, was able to increase the yield of Artemis
protein by five-fold while concurrently increasing the activity of any given purified batch. The new
expression and purification system using codon optimized Artemis is making it feasible to carry out a
large scale screening project (300,000 compounds) as originally envisioned.

82

If a given molecule were to pass the primary screen, the counter screen, and the secondary gel
assays, the molecule will be subsequently tested in a cellular V(D)J recombination assay. The molecule
will also be subjected to chemistry analysis and structural optimization.
We have developed a robust system to assay for potential inhibitor of Artemis. A candidate
molecule will not only show efficacy in the primary high-throughput screen but also in all secondary
assays as well as cellular assays. We are confident that the screen designed can be adapted to other
libraries to screen for additional compounds and generate more lead molecules. The work presented
here provides an alternative approach in blocking key molecules necessary for DNA repair and NHEJ. In
conjunction with radiotherapy and previously identified chemotherapeutic drugs, an inhibitor of Artemis
would offer a powerful tool to combat acute lymphoblastic leukemia.

83

Chapter 6 Concluding Comments
Nucleotide loss and addition occur at double-strand breaks during NHEJ. Nucleotide resection
results from nucleases operating at sites of double-stranded breaks. Based on biochemical studies, two
likely nucleases that contribute to this process are Artemis and PALF. Combined, these two nucleases
h av e 3 ’ ex o n u cl ease act i v it y , 5 ’ e xo n u cle ase a cti v ity , e n d o n u clease acti v ity , and h air p in o p eni n g act i v it y .
Hence, these two enzymes alone may be sufficient to process the majority of DNA end configurations
present after DSBs.
Here we present data demonstrating that Artemis has weak endonuclease activity independent
of DNA-PKcs and this activity is further stimulated by the addition of DNA-PKcs. We also demonstrate in
vitro, purified Artemis protein has intri n sic 5 ’ ex o n u clease a ctiv i ty ind epend ent o f D N A -Pkcs. Taken
together the data suggests that at DSBs Artemis utilizes a combination of both endonuclease activity
an d 5 ’ e xo n u cl ease acti v ity to pr o ce ss bro ke n D N A end s.
We also demonstrate that PALF conta in s 3 ’ ex o n u clea se acti v it y as w ell as end o n u clease a ctiv i ty . In a b io che m ical sy s te m , P ALF is abl e to pro c ess a 3 ’ o v erha n g t o all o w fo r e ffi ci e n t li g atio n . Our data
sug g est s t h at P A L F p lay a r o le in a sub set o f NH EJ rep air w h ere a 3 ’ o v erha n g e xi sts wh ich cannot be
processed by Artemis or other nucleases. Furthermore, PALF may act in conjunction with Artemis to
process DNA ends to expedite the repair process.
While we have studied the mechanistic process of nucleotide processing by PALF and Artemis it
should be noted that PALF and Artemis may play additional roles beyond that of a nuclease as other
groups have suggested. In addition to serving as a nuclease, Artemis may also function in cell cycle
arrest signaling pathway. PALF may also function as a scaffold protein for Ligase IV:XRCC4 complex
during NHEJ. One cannot exclude the possibility that other nucleases besides PALF and Artemis also

84

contribute to a subset of NHEJ. Possible undiscovered proteins may function in alternative end joining
as backup pathways or operate with slower kinetics than either Artemis or PALF during classical NHEJ.
Lastly, we have designed and demonstrated the feasibility of using purified Artemis protein to
create a high-throughput drug screen with the aim to find small molecules that inhibits the activity of
Artemis. By utilizing this approach we are able to accurately identify molecules that may be used
towards the treatment of acute lymphoblastic leukemia.

Acknowledgements:
I would like to acknowledge Dr. Chris Hassig, and Michael Hedrick of Sanford Burnham Medical Research
Institute for supplying figures 5.4 and 5.5.
Additionally I would like to thank Dr. Noritaka Adachi for providing figure 5.1.
Thank you to Dr. Jiafeng Gu, co-author of the paper for which chapter 3 is based upon for his
contribution to figures 3.1 -3.7.

85

References
Ahel, I., Ahel, D., Matsusaka, T., Clark, A.J., Pines, J., Boulton, S.J., and West, S.C. (2008). Poly(ADP-
ribose)-binding zinc finger motifs in DNA repair/checkpoint proteins. Nature 451, 81-85.
Anderson, C.W., and Carter, T.H. (1996). The DNA-Activated Protein Kinase-DNA-PK. In Molecular
Analysis of DNA Rearrangements in the Immune System, R. Jessberger, and M.R. Lieber, eds.
(Heidelberg: Springer-Verlag), pp. 91-112.
Aravind, L. (1999). An evolutionary classification of the metallo-beta-lactamase fold proteins. In Silico
Biol 1, 69-91.
Beucher, A., Birraux, J., Tchouandong, L., Barton, O., Shibata, A., Conrad, S., Goodarzi, A.A., Krempler, A.,
Jeggo, P.A., and Lobrich, M. (2009). ATM and Artemis promote homologous recombination of radiation-
induced DNA double-strand breaks in G2. The EMBO journal 28, 3413-3427.
Blommers, M.J., Walters, J.A., Haasnoot, C.A., Aelen, J.M., Marel, G.v., Boom, J.v., and Hilbers, C.W.
(1989). Effects of base sequencee on the loop folding in DNA hairpins. Biochemistry 28, 7491-7498.
Dominski, Z. (2007). Nucleases of the metallo-beta-lactamase family and their role in DNA and RNA
metabolism. Critical reviews in biochemistry and molecular biology 42, 67-93.
Dronkert, M.L.G., Wit, J.d., Boeve, M., Vasconcelos, M.L., Steeg, H.v., Tan, T.L., Hoeijmakers, J.H.J., and
Kanaar, R. (2000). Disruption of mouse SNM1 causes increased sensitivity to the DNA interstrand cross-
linking agent mitomycin C. Mol. Cell. Biol. 20, 4553-4561.
Frank, K.M., Sekiguchi, J.M., Seidl, K.J., Swat, W., Rathbun, G.A., Cheng, H.L., Davidson, L., Kangaloo, L.,
and Alt, F.W. (1998). Late embryonic lethality and impaired V(D)J recombination in mice lacking DNA
ligase IV. Nature 396, 173-177.
Goodarzi, A.A., Yu, Y., Riballo, E., Douglas, P., Walker, S.A., Ye, R., Harer, C., Marchetti, C., Morrice, N.,
Jeggo, P.A., et al. (2006). DNA-PK autophosphorylation facilitates Artemis endonuclease activity. The
EMBO journal 25, 3880-3889.
Grundy, G.J., Rulten, S.L., Zeng, Z., Arribas-Bosacoma, R., Iles, N., Manley, K., Oliver, A., and Caldecott,
K.W. (2013). APLF promotes the assembly and activity of non-homologous end joining protein
complexes. The EMBO journal 32, 112-125.
Gu, J., Li, S., Zhang, X., Wang, L.C., Niewolik, D., Schwarz, K., Legerski, R.J., Zandi, E., and Lieber, M.R.
(2010). DNA-PKcs regulates a single-stranded DNA endonuclease activity of Artemis. DNA Repair (Amst)
9, 429-437.
Gu, J., Lu, H., Tippin, B., Shimazaki, N., Goodman, M.F., and Lieber, M.R. (2007a). XRCC4:DNA ligase IV
can ligate incompatible DNA ends and can ligate across gaps. The EMBO journal 26, 1010-1023.
Gu, J., Lu, H., Tsai, A.G., Schwarz, K., and Lieber, M.R. (2007b). Single-stranded DNA ligation and XLF-
stimulated incompatible DNA end ligation by the XRCC4-DNA ligase IV complex: influence of terminal
DNA sequence. Nucleic acids research 35, 5755-5762.

86

Gu, J., Lu, H., Tsai, A.G., Schwarz, K., and Lieber, M.R. (2007c). Single-stranded DNA ligation and XLF-
stimulated incompatible DNA end ligation by the XRCC4-DNA ligase IV complex: influence of terminal
DNA sequence. Nucleic Acids Res 35, 5755-5762.
Harrington, J., Hsieh, C.L., Gerton, J., Bosma, G., and Lieber, M.R. (1992). Analysis of the defect in DNA
end joining in the murine scid mutation. Molecular and cellular biology 12, 4758-4768.
Helmink, B.A., and Sleckman, B.P. (2012). The response to and repair of RAG-mediated DNA double-
strand breaks. Annual review of immunology 30, 175-202.
Howard, F.B., Chen, C., Ross, P.D., and Miles, H.T. (1991). Hairpin formation in the self-complementary
dodecamer d-GGTACGCGTACC and derivatives containind GA and IA mispairs. Biochemistry 30, 779-782.
Huang, Y., Giblin, W., Kubec, M., Westfield, G., St Charles, J., Chadde, L., Kraftson, S., and Sekiguchi, J.
(2009). Impact of a hypomorphic Artemis disease allele on lymphocyte development, DNA end
processing, and genome stability. The Journal of experimental medicine 206, 893-908.
Iles, N., Rulten, S., El-Khamisy, S.F., and Caldecott, K.W. (2007). APLF (C2orf13) is a novel human protein
involved in the cellular response to chromosomal DNA strand breaks. Molecular and cellular biology 27,
3793-3803.
Kanno, S., Kuzuoka, H., Sasao, S., Hong, Z., Lan, L., Nakajima, S., and Yasui, A. (2007). A novel human AP
endonuclease with conserved zinc-finger-like motifs involved in DNA strand break responses. The EMBO
journal 26, 2094-2103.
Kim, C., Snyder, R.O., and Wold, M.S. (1992). Binding properties of replication protein A from human and
yeast cells. Molecular and cellular biology 12, 3050-3059.
Kurosawa, A., Koyama, H., Takayama, S., Miki, K., Ayusawa, D., Fujii, M., Iiizumi, S., and Adachi, N.
(2008). The requirement of Artemis in double-strand break repair depends on the type of DNA damage.
DNA and cell biology 27, 55-61.
Larsen, K.S., and Auld, D.S. (1989). Carboxypeptidase A: mechanism of zinc inhibition. Biochemistry 28,
9620-9625.
Li, G., Alt, F.W., Cheng, H.L., Brush, J.W., Goff, P.H., Murphy, M.M., Franco, S., Zhang, Y., and Zha, S.
(2008). Lymphocyte-specific compensation for XLF/cernunnos end-joining functions in V(D)J
recombination. Molecular cell 31, 631-640.
Li, X., Hejna, J., and Moses, R.E. (2005). The yeast Snm1 protein is a DNA 5'-exonuclease. DNA Repair
(Amst) 4, 163-170.
Lieber, M.R. (2008). The mechanism of human nonhomologous DNA end joining. The Journal of
biological chemistry 283, 1-5.
Lieber, M.R. (2010). The Mechanism of Double-Strand DNA Break Repair by the Nonhomologous DNA
End-Joining Pathway. Annu Rev Biochem 79, 181-211.

87

Lu, H., Schwarz, K., and Lieber, M.R. (2007). Extent to which hairpin opening by the Artemis:DNA-PKcs
complex can contribute to junctional diversity in V(D)J recombination. Nucleic Acids Res.
Lu, H., Shimazaki, N., Raval, P., Gu, J., Watanabe, G., Schwarz, K., Swanson, P., and Lieber, M.R. (2008). A
biochemically defined system for coding joint formation in human V(D)J recombination. Molecular Cell
31, 485-497.
Ma, Y., Pannicke, U., Lu, H., Niewolik, D., Schwarz, K., and Lieber, M.R. (2005a). The DNA-dependent
protein kinase catalytic subunit phosphorylation sites in human Artemis. The Journal of biological
chemistry 280, 33839-33846.
Ma, Y., Pannicke, U., Lu, H., Niewolik, D., Schwarz, K., and Lieber, M.R. (2005b). The DNA-PKcs
phosphorylation sites of human artemis. J. Biol. Chem. 280, 33839-33846.
Ma, Y., Pannicke, U., Schwarz, K., and Lieber, M.R. (2002a). Hairpin opening and overhang processing by
an Artemis/DNA-dependent protein kinase complex in nonhomologous end joining and V(D)J
recombination. Cell 108, 781-794.
Ma, Y., Pannicke, U., Schwarz, K., and Lieber, M.R. (2002b). Hairpin opening and overhang processing by
an Artemis:DNA-PKcs complex in V(D)J recombination and in nonhomologous end joining. Cell 108, 781-
794.
Ma, Y., Schwarz, K., and Lieber, M.R. (2005c). The Artemis:DNA-PKcs Endonuclease Can Cleave Gaps,
Flaps, and Loops. DNA Repair 4, 845-851.
Macrae, C.J., McCulloch, R.D., Ylanko, J., Durocher, D., and Koch, C.A. (2008). APLF (C2orf13) facilitates
nonhomologous end-joining and undergoes ATM-dependent hyperphosphorylation following ionizing
radiation. DNA repair 7, 292-302.
Mandel, C.R., Kaneko, S., Zhang, H., Gebauer, D., Vethantham, V., Manley, J.L., and Tong, L. (2006).
Polyadenylation factor CPSF-73 is the pre-mRNA 3'-end-processing endonuclease. Nature 444, 953-956.
Marti, T.M., and Fleck, O. (2004). DNA repair nucleases. Cellular and molecular life sciences : CMLS 61,
336-354.
Meek, K., Dang, V., and Lees-Miller, S.P. (2008). DNA-PK: the means to justify the ends? Advances in
immunology 99, 33-58.
Meek, K., Gupta, S., Ramsden, D.A., and Lees-Miller, S.P. (2004). The DNA-dependent protein kinase:
the director at the end. Immunol. Rev. 200, 132-141.
Mehrotra, P.V., Ahel, D., Ryan, D.P., Weston, R., Wiechens, N., Kraehenbuehl, R., Owen-Hughes, T., and
Ahel, I. (2011). DNA repair factor APLF is a histone chaperone. Molecular cell 41, 46-55.
Mimitou, E.P., and Symington, L.S. (2009). DNA end resection: many nucleases make light work. DNA
repair 8, 983-995.

88

Moshous, D., Callebaut, I., Chasseval, R.d., Corneo, B., Cavazzana-Calvo, M., Diest, F.L., Tezcan, I., Sanal,
O., Bertrand, Y., Philippe, N., et al. (2001). Artemis, a novel DNA double-strand break repair/V(D)J
recombination protein, is mutated in human severe combined immune deficiency. Cell 105, 177-186.
Niewolik, D., Pannicke, U., Lu, H., Ma, Y., Wang, L.C., Kulesza, P., Zandi, E., Lieber, M.R., and Schwarz, K.
(2006a). DNA-PKcs dependence of Artemis endonucleolytic activity, differences between hairpins and 5'
or 3' overhangs. The Journal of biological chemistry 281, 33900-33909.
Niewolik, D., Pannicke, U., Lu, H., Ma, Y., Wang, L.C., Kulesza, P., Zandi, E., Lieber, M.R., and Schwarz, K.
(2006b). DNA-PKcs dependence of artemis endonucleolytic activity: differences between hairpins and 5'
or 3' overhangs. J. Biol. Chem. 281, 33900-33909.
Noordzij, J.G., Verkaik, N.S., van der Burg, M., van Veelen, L.R., de Bruin-Versteeg, S., Wiegant, W.,
Vossen, J.M., Weemaes, C.M., de Groot, R., Zdzienicka, M.Z., et al. (2003). Radiosensitive SCID patients
with Artemis gene mutations show a complete B-cell differentiation arrest at the pre-B-cell receptor
checkpoint in bone marrow. Blood 101, 1446-1452.
Pannicke, U., Honig, M., Schulze, I., Rohr, J., Heinz, G.A., Braun, S., Janz, I., Rump, E.M., Seidel, M.G.,
Matthes-Martin, S., et al. (2010). The most frequent DCLRE1C (ARTEMIS) mutations are based on
homologous recombination events. Human mutation 31, 197-207.
Pannicke, U., Ma, Y., Lieber, M.R., and Schwarz, K. (2004). Functional and biochemical dissection of the
structure-specific nuclease Artemis. EMBO J. 23, 1987-1997.
Pawelczak, K.S., and Turchi, J.J. (2010). Purification and characterization of exonuclease-free Artemis:
Implications for DNA-PK-dependent processing of DNA termini in NHEJ-catalyzed DSB repair. DNA repair
9, 670-677.
Povirk, L.F., Zhou, T., Zhou, R., Cowan, M.J., and Yannone, S.M. (2007). Processing of 3'-
phosphoglycolate-terminated DNA double strand breaks by Artemis nuclease. The Journal of biological
chemistry 282, 3547-3558.
Raghunathan, G., Jernigan, R.L., Miles, H.T., and Sasisekharan, V. (1991). Conformational feasibility of a
hairpin with two purines in the loop. 5'-d-GGTACIAGTACC-3'. Biochemistry 30, 782-788.
Riballo, E., Kuhne, M., Rief, N., Doherty, A., Smith, G.C., Recio, M.J., Reis, C., Dahm, K., Fricke, A.,
Krempler, A., et al. (2004a). A pathway of double-strand break rejoining dependent upon ATM, Artemis,
and proteins locating to gamma-H2AX foci. Molecular cell 16, 715-724.
Riballo, E., Kuhne, M., Rief, N., Doherty, A., Smith, G.C.M., Recio, M.-J., Reis, C., Dahm, K., Fricke, A.,
Kempler, A., et al. (2004b). A pathway of double-strand break rejoining dependent upon ATM, Artemis,
and proteins locating to gamma-H2AX foci. Mol. Cell 16, 715-724.
Rooney, S., Alt, F.W., Lombard, D., Whitlow, S., Eckersdorff, M., Fleming, J., Fugmann, S., Ferguson, D.O.,
Schatz, D.G., and Sekiguchi, J. (2003a). Defective DNA repair and increased genomic instability in
Artemis-deficient murine cells. J. Exp. Med. 197, 553-565.

89

Rooney, S., Alt, F.W., Lombard, D., Whitlow, S., Eckersdorff, M., Fleming, J., Fugmann, S., Ferguson, D.O.,
Schatz, D.G., and Sekiguchi, J. (2003b). Defective DNA repair and increased genomic instability in
Artemis-deficient murine cells. The Journal of experimental medicine 197, 553-565.
Rooney, S., Chaudhuri, J., and Alt, F.W. (2004). The role of the non-homologous end-joining pathway in
lymphocyte development. Immunological reviews 200, 115-131.
Ryan, K., Calvo, O., and Manley, J.L. (2004). Evidence that polyadenylation factor CPSF-73 is the mRNA 3'
processing endonuclease. Rna 10, 565-573.
Shirodkar, P., Fenton, A.L., Meng, L., and Koch, C.A. (2013). Identification and Functional
Characterization of a Ku-binding Motif in Aprataxin Polynucleotide Kinase/Phosphatase-like Factor
(APLF). The Journal of biological chemistry 288, 19604-19613.
van Overbeek, M., and de Lange, T. (2006). Apollo, an Artemis-related nuclease, interacts with TRF2 and
protects human telomeres in S phase. Curr Biol 16, 1295-1302.
West, R.B., Yaneva, M., and Lieber, M.R. (1998). Productive and nonproductive complexes of Ku and
DNA-dependent protein kinase at DNA termini. Molecular and cellular biology 18, 5908-5920.
Yannone, S.M., Khan, I.S., Zhou, R.Z., Zhou, T., Valerie, K., and Povirk, L.F. (2008). Coordinate 5' and 3'
endonucleolytic trimming of terminally blocked blunt DNA double-strand break ends by Artemis
nuclease and DNA-dependent protein kinase. Nucleic Acids Res 36, 3354-3365.
Zhang, X., Succi, J., Feng, Z., Prithivirajsingh, S., Story, M., and Legerski, R.J. (2004a). Artemis is a
phosphorylation target of ATM and ATR and is involved in the G2/M DNA damage checkpoint response.
Mol. Cell. Biol. 24, 9207-9220.
Zhang, X., Succi, J., Feng, Z., Prithivirajsingh, S., Story, M.D., and Legerski, R.J. (2004b). Artemis is a
phosphorylation target of ATM and ATR and is involved in the G2/M DNA damage checkpoint response.
Molecular and cellular biology 24, 9207-9220.

Abstract (if available)

Abstract DNA double-stranded breaks (DSB) can occur through programmed mechanisms such as V(D)J recombination and class switch recombination or pathological mechanisms such as ionizing radiation. To maintain genomic stability and integrity it is essential for DSB to be repaired properly and efficiently. Repair of DSB is predominantly carried out by the non-homologous end joining (NHEJ) pathway. Broken DNA ends are recognized and bound by the Ku70/80 heterodimer, followed by DNA-PKcs binding, processing by nucleases and polymerases, before being joined by the XRCC4-Ligase IV complex. Unrepaired or improperly repaired DSB results in apoptosis, slowed cell growth, or translocation. Here we characterize the nucleases that are involved in a subset of NHEJ as well as explore the implications of inhibiting these nucleases for therapeutic purposes. ❧ PNK Aprataxin like Factor (PALF also referred to as APLF) was first identified by searching through protein libraries for proteins containing the FHA domain. PALF is able to bind to NHEJ factors, Ku80 and XRCC4, as well as to ssDNA repair proteins, PARP1 and PARP3. It has been proposed that PALF can function as a nuclease during DNA repair. We explore this possibility in a biochemical system using purified PALF protein. The data presented here demonstrates that PALF has 3’ exonuclease activity independent of binding interaction with other NHEJ components. PALF also has endonuclease activity at 5’ and 3’ overhangs. Additionally, in a reconstituted biochemical joining system, PALF is able to participate in ligation by functioning to process incompatible DNA ends. In a cellular system, small interfering RNA targeted against PALF slows repair kinetics after treatment with ionizing radiation. ❧ Artemis is the only known mammalian nuclease which can open hairpin DNAs. This essential feature of Artemis makes it indispensable for processing coding ends during V(D)J recombination. Human and mice lacking Artemis protein suffer from immunodeficiency due to lack of T and B cells. Data presented here explores the nucleolytic capabilities of Artemis on a single-stranded DNA. We show that Artemis has low levels of nuclease activity on single-stranded DNA. Artemis preferentially processes single-stranded pyrimidines and not purines. The single-stranded endonuclease activity of Artemis is further stimulated by the addition of DNA-PKcs. ❧ Previous data suggests that Artemis has intrinsic 5’ exonuclease activity which elutes at the same position as the intrinsic endonuclease activity across several different purification columns. We wanted to further evaluate whether the 5’ exonuclase is intrinsic to Artemis by creating a point mutant of Artemis. The point mutant of Artemis, ARM14, has a single amino acid substitution at a conserved histidine residue (H115). Purified ARM14 mutant, unlike wild type Artemis protein, does not have endonuclease activity or 5’ exonuclease activity. CD spectrum demonstrates that purified ARM14 retains similar structural profile to wild-type Artemis in solution. Manipulation of the divalent cation concentration affects both the 5’ exonuclease and the endonuclease with similar behavior. Furthermore, we show that several different selective chemicals are able to inhibit both the endonuclease as well as the exonuclease activity of Artemis. We conclude that wild-type Artemis has intrinsic 5’ exonuclease activity which utilizes the same catalytic site as its endonuclease activity. ❧ To explore the possibility of inhibiting Artemis as a potential treatment for acute lymphoblastic leukemia (ALL), we utilized a high-throughput screen to search for small molecule inhibitors of Artemis. In a biochemical reaction, purified Artemis is added to a fluorescent substrate and the reaction is miniaturized to a 4 to 6ul volume format to be compatible with conventional HTS screening. In preliminary assay optimization, we were able to achieve a robust HTS, based on standard HTS criteria (Z scores and signal to noise ratios in a 1536 well assay plate format). We carried out a small scale HTS using 31,000 compounds from an NIH library. Of the 31,000 compounds screened, we observed a hit rate of 0.59% when using a threshold of inhibition greater than 30%. Three of the molecules with the most potent IC50 and greatest efficacy were further tested in a secondary gel based assay and counter-screened against mung bean nuclease. From this initial screen we have identified 2 compounds that are able to inhibit Artemis in our testing scheme.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Studies on the role of Artemis in non-homologous DNA end-joining to understand the mechanism and discover therapies

PDF

Molecular elucidation of nonhomologous DNA end-joining in the context of nucleosome core particles

PDF

Mechanism of human nonhomologous DNA end joining

PDF

Biochemistry and reconstitution of V(D)J recombination in a purified system

PDF

X-ray structural studies on DNA-dependent protein kinase catalytic subunit:DNA co-crystals

PDF

Current models of non-homologous end joining and their implications in gene therapy

PDF

Mechanistic basis for chromosomal translocations at the E2A gene

PDF

The mechanism of R-loop formation in mammalian immunoglobulin class switch recombination

PDF

Study of protein deamidation in innate immune signaling

PDF

Characterizing and manipulating homology-directed gene editing in human cells

PDF

UVRAG maintains genomic integrity

PDF

Biochemical mechanism of TopBP1 recruitment to sites of DNA damage

PDF

Decreased levels of expression of transmembrane protein 56 (TMEM56) in breast cancer tissues

PDF

Studies on the role of TMEM56 in tumorigenesis by using PTEN knockout mouse model

PDF

Rational selection of CRISPR/Cas9 guide RNAs for homology directed genome editing and its utility in the development of gene therapies

PDF

Effects of chromatin regulators during carcinogenesis

PDF

Biochemical characterization and structural analysis of two hexameric helicases for eukaryotic DNA replication

PDF

Characterization of three novel variants of the MAVS adaptor

PDF

Multiple functions of the PR-Set7 histone methyltransferase: from transcription to the cell cycle

PDF

Molecular genetic analysis of human genetic lesions

Asset Metadata

Creator Li, Sicong (author)

Core Title Mechanisms of nucleases in non-homologous DNA end joining

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Genetic, Molecular and Cellular Biology

Publication Date 11/01/2013

Defense Date 10/11/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Artemis,DNA repair,NHEJ,nuclease,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Hsieh, Chih-Lin (committee chair), Lieber, Michael R. (committee chair), Zandi, Ebrahim (committee member)

Creator Email sicongli@usc.edu,sicongli0603@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-342055

Unique identifier UC11296607

Identifier etd-LiSicong-2125.pdf (filename),usctheses-c3-342055 (legacy record id)

Legacy Identifier etd-LiSicong-2125.pdf

Dmrecord 342055

Document Type Dissertation

Format application/pdf (imt)

Rights Li, Sicong

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA