Crystal Structure of Human XLF:
A Twist in Nonhomologous DNA End-Joining
Sara N. Andres,1Mauro Modesti,2Chun J. Tsai,3,4Gilbert Chu,3,4and Murray S. Junop1,*
1Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, ON L8N 3Z5, Canada
2Genome Instability and Carcinogenesis, CNRS FRE 2931, 31 Chemin Joseph Aiguier, 13402 Marseille Cedex 20, France
3Department of Medicine
4Department of Biochemistry
Stanford University School of Medicine, Stanford, CA 94305, USA
DNA double-strand breaks represent one of the
most severe forms of DNA damage in mamma-
lian cells. One pathway for repairing these
breaks occurs via nonhomologous end-joining
(NHEJ) and depends on XRCC4, LigaseIV, and
Cernunnos, also called XLF. Although XLF stim-
ulates XRCC4/LigaseIV to ligate mismatched
and noncohesive DNA ends, the mechanistic
basis for this function remains unclear. Here
we report the structure of a partially functional
224 residue N-terminal fragment of human XLF.
Despite only weak sequence similarity, XLF1–170
shares structural homology with XRCC41–159.
However, unlike the highly extended 130 A˚heli-
cal domain observed in XRCC4, XLF adopts
a more compact, folded helical C-terminal re-
gion involving two turns and a twist, wrapping
back to the structurally conserved N terminus.
Mutational analysis of XLF and XRCC4 reveals
a potential interaction interface, suggesting
of mismatched ends.
Generation of DNA double-strand breaks (DSBs) poses
a serious threat to chromosomal integrity. If left unre-
paired, such breaks can generate destabilizing chromo-
somal rearrangements that may lead to tumorigenesis
(Gao et al., 2000; Lieber, 1998). DSBs occur in response
to exogenous genotoxic agents and as intermediates in
the genomic rearrangements associated with V(D)J
recombination. Eukaryotic cells maintain two systems
for repairing DSBs: homologous recombination, which
facilitates accurate restoration of DNA (Jeggo, 1998),
and nonhomologous end-joining (NHEJ), which facilitates
error-prone repair without a homologous DNA template.
NHEJ requires a core set of five proteins: Ku, DNA-
PKcs, XRCC4, LigaseIV, and Cernunnos, an XRCC4-like
factor hereafter called XLF. Ku initially binds broken DNA
ends, preventing nucleolytic degradation (Liang and
Jasin, 1996), and then recruits DNA-PKcs, which stabi-
lizes synapsis of the DNA ends (DeFazio et al., 2002).
DNA end processing enzymes, including Artemis, en-
hance NHEJ by making DNA ends compatible for ligation
2005). However, in the presence of Ku, XLF stimulates
XRCC4/LigaseIV to ligate mismatched and noncohesive
DNA ends (Tsai et al., 2007). XLF can bind DNA and inter-
act with the XRCC4/LigaseIV complex; however, the
nature of these interactions and the mechanism by which
XLF functions remain poorly understood (Hentges et al.,
2006; Callebaut et al., 2006; Ahnesorg et al., 2006; Lu
et al., 2007).
XLF was identified as the gene mutated in patients
exhibiting immune deficiencies and microcephaly, symp-
toms consistent with impaired NHEJ (Ahnesorg et al.,
2006; Buck et al., 2006). Secondary structure predictions
suggested that XLF and XRCC4 are structurally related,
despite limited sequence similarity (Ahnesorg et al.,
2006). In addition, XLF and XRCC4 exhibit other similari-
ties. Both proteins exist as dimers and have DNA binding
activity dependent on the presence of long (>80 bp) DNA
fragments (Hentges et al., 2006; Lu et al., 2007). Although
XLF and XRCC4 are able to interact, the nature and func-
tional importance of this quaternary structure remain
unknown. Current data suggest that XLF interacts with
XRCC4/LigaseIV primarily through contact with XRCC4,
but the role of DNA and LigaseIV in this interaction has
not been rigorously evaluated (Deshpande and Wilson,
2007; Lu et al., 2007).
Here we report the crystal structure of a partially func-
tional N-terminal fragment of human XLF1–224at 2.5 A˚
resolution. This fragment fails to efficiently bind DNA or
stimulate ligation of noncohesive DNA ends but retains
the ability to directly interact with XRCC4. The N-terminal
region of XLF (amino acids 1–170) adopts a structure
almost identical to XRCC4, as predicted from the amino
acid sequences (Ahnesorg et al., 2006; Callebaut et al.,
2006). However, amino acids 170–224 diverge greatly
from the expected elongated helix seen in XRCC4. Two
helices within this region of XLF fold back onto and twist
Molecular Cell 28, 1093–1101, December 28, 2007 ª2007 Elsevier Inc. 1093
around the main helical stalk of XLF, occluding the analo-
gous LigaseIV binding interface observed in XRCC4. The
structure of XLF therefore explains why LigaseIV is unable
to bind XLF in the same manner observed for XRCC4.
Based on the structures of XLF and XRCC4, we then
conducted mutational analysis aimed at further defining
surfaces involved in the protein-protein interactions.
of XLF interact with XRCC4 and DNA, respectively. Taken
of a functional DNA repair complex.
RESULTS AND DISCUSSION
Crystal Structure of XLF1–224Shows Formation
of a Stable Dimer
Human XLF contains 299 residues (Ahnesorg et al., 2006;
Buck et al., 2006). Comparison with other eukaryotic ho-
mologs reveals a high degree of sequence similarity within
the N-terminal ?220 residues. Interestingly, this does
not extend to the C-terminal portion of XLF. As expected,
similarity within the N-terminal region is most pronounced
among hydrophobic amino acids (Figure 1A). Several
clusters of more highly conserved residues, including
both hydrophobic and charged/polar amino acids, occur
in the N-terminal region (57–65, 108–123, and 160–186)
as well as at a single basic region in the extreme C termi-
nus (residues 288–295) (Figures 1A and 1C). Based on
these observations, we created a truncation to remove
the less conserved C-terminal 75 residues of human
XLF, with the hope of reducing structural flexibility and
thereby facilitating crystallization.
Although native XLF1–224formed crystals, the quality
and resolution of diffraction data were poor; thus, all
data used for both structure solution and refinement
came from selenomethionine-substituted protein. Experi-
mental phases were determined by using single-wave-
length anomalous diffraction (SAD). Crystals formed in
space group P212121, with 2 monomers of XLF in each
asymmetric unit. Regions of disorder within the crystal
prevented modeling of residues 85–92 in chain A and res-
idues 80–92 in chain B. These residues are illustrated
as dashed loops in the final structure (Figure 1C). The
overall structure was refined to R and Rfree values of
25.3% and 28.7%, respectively (PDB 2R9A). Table 1 con-
tains a complete list of data collection and model refine-
XLF1–224is comprised of two principal domains. The
globular N-terminal head domain consists of a seven-
stranded antiparallel b sandwich with a single helix at the
N terminus and a helix-turn-helix motif inserted between
strands 4 and 5 (Figure 1B). The loop region connecting
a3 and b5 is not well ordered and therefore not present
in the final model. Three a helices combine to form an
elongated C-terminal domain, the first of which (a4)
extends ?60 A˚away from the N-terminal head. This rather
long helix terminates with a conserved glycine at position
171. The following extended coiled region (residues 171–
185) reverses the overall trajectory of the polypeptide
chain, bringing the second (a5) and third (a6) helices of
the C-terminal domain back into contact with a4. The
last helix (a6) of the C-terminal domain wedges itself
between a4 and b1, forcing the N-terminal head domain
to adopt an elevated position (Figure 1C). In addition,
of XLF reside near the N-terminal head domain.
XLF formed a dimer within the crystallographic asym-
metric unit. This quaternary structure has also been ob-
served for XLF homologs in S. pombe and S. cerevisiae
(Cavero et al., 2007; Sulek et al., 2007), suggesting that
human XLF forms a homodimer in the absence of DNA
and other NHEJ proteins. XLF1–224monomers are related
by a 2-fold noncrystallographic axis and interact primarily
through conserved hydrophobic regions within a4, listed
in detail in Figure S1 available online. The dimer is also
held in place by compensatory interactions between a4
of one monomer and a5 of the opposing monomer.
Further stabilization occurs from the clamp formed by a5
and a6 as they twist upward and wrap around a4 from
the opposite subunit. The dimer interface buries a surface
area of ?6500 A˚2, the greatest contribution coming from
interactions thatresult fromthefolding back andclamping
of a5 and a6 onto a4. The combined effects of burying
extensive hydrophobic surface and structural support
gained by wrapping a5 and a6 onto a4 suggest that the
XLF dimer is highly stable. However, the possibility re-
mains that opposing tails (a5 and a6) from two XLF dimers
could interchange to form a tetramer similar to XRCC4
(Junop et al., 2000). Preliminary data from gel filtration
analysis suggest that XLF can form concentration-depen-
dent higher-order complexes (Figure S2).
Interestingly, three patches of highly conserved resi-
dues in XLF1–224do not contribute significantly to the
dimer interface (Figure 1). Many of the conserved residues
in these regions are exposed and likely involved in other
interactions. Patch III (residues 160–186) is N-terminal to
the region in XLF (residues 186–209) corresponding to
the region in XRCC4 that binds to LigaseIV. However,
Patch III forms a hinge that folds back to occlude the
potential LigaseIV binding surface. Although Patch I and
II are noncontiguous in primary structure, these regions
form a continuous surface in the tertiary structure of XLF
(Figure 1C). Many of the conserved residues in these
two patches consist of exposed polar/charged residues
that are not expected to contribute to protein folding. A
stretch of three exposed hydrophobic residues (amino
acids 115, 116 and 118; see Figure S1) is also found in
this region, specifically within the short loop joining b6
and b7. These features raise the possibility that the Patch
I and II surfaces mediate interactions with XRCC4 or other
The XLF C-Terminal Domain Is Required for DNA
Binding and Mismatched End Ligation
To determine the role of the conserved N-terminal region
of XLF in NHEJ, we tested XLF1–224for its ability to
1094 Molecular Cell 28, 1093–1101, December 28, 2007 ª2007 Elsevier Inc.
Crystal Structure of Human XLF
Junop, M.S., Modesti, M., Guarne, A., Ghirlando, R., Gellert, M., and
Yang, W. (2000). Crystal structure of the Xrcc4 DNA repair protein
and implications for end joining. EMBO J. 22, 5962–5970.
Liang, F., and Jasin, M. (1996). Ku80-deficient cells exhibit excess
degradation of extrachromosomal DNA. J. Biol. Chem. 24, 14405–
Lieber, M.R. (1998). Warner-Lambert/Parke-Davis Award Lecture.
Pathological and physiological double-strand breaks: roles in cancer,
aging, and the immune system. Am. J. Pathol. 5, 1323–1332.
Lu, H., Pannicke, U., Schwarz, K., and Lieber, M.R. (2007). Length-
dependent binding of human XLF to DNA and stimulation of
XRCC4.DNA ligase IV activity. J. Biol. Chem. 15, 11155–11162.
Ma, Y., Pannicke, U., Schwarz, K., and Lieber, M.R. (2002). Hairpin
teinkinasecomplex in nonhomologous end joining and V(D)J recombi-
nation. Cell 6, 781–794.
Maiti, R., Van Domselaar, G.H., Zhang, H., and Wishart, D.S. (2004).
SuperPose: a simple server for sophisticated structural superposition.
Nucleic Acids Res. 32, 590–594.
Modesti, M., Hesse, J.E., and Gellert, M. (1999). The binding of Xrcc4
protein is associated with V(D)J recombination but not with stimulation
of DNA ligase IV activity. EMBO J. 18, 2008–2018.
Modesti, M., Junop, M.S., Ghirlando, R., van de Rakt, M., Gellert, M.,
Yang, W., and Kanaar, R. (2003). Tetramerization and DNA ligase IV in-
teraction of the DNA double-strand break repair protein XRCC4 are
mutually exclusive. J. Mol. Biol. 2, 215–228.
Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of
macromolecular structures by the maximum-likelihood method. Acta
Crystallogr. D Biol. Crystallogr 53, 240–255.
Nick McElhinny, S.A., Snowden, C.M., McCarville, J., and Ramsden,
D.A. (2000). Ku recruits the XRCC4-ligase IV complex to DNA ends.
Mol. Cell. Biol. 20, 2996–3003.
Nick McElhinny, S.A., Havener, J.M., Garcia-Diaz, M., Juarez, R., Be-
benek, K., Kee, B.L., Blanco, L., Kunkel, T.A., and Ramsden, D.A.
(2005). A gradient of template dependence defines distinct biological
roles for family X polymerases in nonhomologous end joining. Mol.
Cell 3, 357–366.
Pflugrath, J.W. (1999). The finer things in X-ray diffraction data collec-
tion. Acta Crystallogr. D Biol. Crystallogr. 55, 1718–1725.
Sibanda, B.L., Critchlow, S.E., Begun, J., Pei, X.Y., Jackson, S.P.,
Blundell, T.L., and Pellegrini, L. (2001). Crystal structure of an Xrcc4-
DNA ligase IV complex. Nat. Struct. Biol. 12, 1015–1019.
Sulek, M., Yarrington, R., McGibbon, G., Boeke, J.D., and Junop, M.S.
(2007). A critical role for the C-terminus of Nej1 protein in Lif1p associ-
ation, DNA binding and non-homologous end-joining. DNA Repair
(Amst.) 6, 1805–1818.
Tsai, C.J., Kim, S.A., and Chu, G. (2007). XLF promotes the ligation of
mismatched and noncohesive DNA ends. Proc. Natl. Acad. Sci. USA
The coordinates of the molecular structure of human XLF encompass-
Bank under accession code 2R9A.
Molecular Cell 28, 1093–1101, December 28, 2007 ª2007 Elsevier Inc. 1101
Crystal Structure of Human XLF