| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Warner-Lambert/Parke-Davis Award Lecture |
From the Departments of Pathology, Biochemistry and Molecular Biology, and Molecular Microbiology and Immunology, Norris Comprehensive Cancer Center, University of Southern California School of Medicine, Los Angeles, California
Abstract
Pathological agents such as ionizing radiation and oxidative free radicals can cause breaks in both strands of the DNA at a given site (double-strand break). This is the most serious type of DNA damage because neither strand is able to provide physical integrity or information content, as would be the case for single-strand DNA damage where one strand of the duplex remains intact. The repair of such breaks usually results in an irreversible alteration of the DNA. Two physiological forms of intentional double-strand (ds) DNA breakage and rejoining occur during lymphoid differentiation. One is V(D)J recombination occurring during early B and T cell development, and the other is class switch recombination, occurring exclusively in mature B cells. The manner in which physiological and most pathological double-strand DNA breaks are rejoined to restore chromosomal integrity are the same. Defects during the phases in which pathological or physiological breaks are generated or in which they are joined can result in chromosomal translocations or loss of genetic information at the site of breakage. Such events are the first step in some cancers and may be a key contributor to changes in DNA with age. Inherited defects in this process can result in severe combined immune deficiency. Hence, pathological and physiological DNA double-strand breaks are related to immune defects and cancer and may be one of the key ways in which DNA is damaged during aging.
. One can broadly classify the types of DNA repair
as those dedicated to repair of damage to a single DNA strand
(single-strand break repair) or those dedicated to repair of damage
where both DNA strands are broken (double-strand break repair) (Figure 1
, Table 2
).
|
|
|
Single-celled eukaryotes, such as yeast, can use the other chromosome when they are in their diploid phase to copy the information by a process called homologous recombination (HR). However, unlike yeast, the genomes of higher eukaryotes have an abundance of repetitive DNA, with similar or identical repeats scattered over many different chromosomes. HR under such circumstances may be difficult because of the abundance of similar sequences throughout the genome.
Instead, cells of multicellular eukaryotes rely on a process where the two broken DNA ends are joined back together, even if information between the two broken ends is lost. This process is called nonhomologous DNA end joining (NHEJ) to distinguish it from HR.
Steps and Proteins Involved in the Joining of Broken DNA Ends
When double-strand breaks occur in DNA, the biochemical configuration of the broken ends can be any of a large number of possibilities. Hence, putting the two ends back together cannot usually be achieved by a simple ligation step.
For the two DNA ends to be processed so that they can be joined, they
must be maintained in physical proximity (Figure 2)
. The
term synapsis can be used to describe this step. Concurrently, proteins
may bind that signal that there is a double-strand break, a step that
can be referred to as end activation. One of the first proteins to bind
is Ku. Ku binds to DNA ends and can diffuse to internal positions from
the end. It is not yet clear whether Ku is responsible for synapsis of
the DNA ends, as some reports have suggested,2,3
or whether
additional proteins are involved.
|
When the two broken DNA ends are brought into proximity, the exact point of joining can occur at any base pair. There is a propensity for the joining to occur at one to four nucleotides that are complementary between the two ends. Therefore, if the DNA end sequence at one end is —-TTGGT* (where the dashes represent the rest of the chromosome and the asterisk represents the break site) and the other end begins as *aggcc—, then the sequence at the point of joining has a higher probability of being —TTGGcc—-. This means that the T at one end and the a at the other end were removed, and the joining occurred at the two base pairs GG or gg, where the two ends shared these two nucleotides of microhomology. The reason for this preference at the junction is because this is where the two DNA ends can base pair. This tendency is called microhomology usage. It is not an essential feature, because two ends that do not share microhomology can still be joined at normal efficiencies. Rather, the microhomology usage is something that occurs at a much higher frequency than random positioning of the join site. This tendency for microhomology usage is seen in general DNA end joining and in specialized physiological forms of double-strand break repair, such as V(D)J recombination.4
When the two DNA ends do undergo alignment at points of microhomology,
then there is often either excess DNA beyond the point of alignment
that must be removed by nucleases or there are gaps that must be filled
by polymerases. We call this the DNA end processing step. The
intriguing feature about this step is that the two DNA ends must be
held in proximity to prevent them from diffusing away from one another,
and yet the means by which those ends are held cannot prevent access by
the nucleases and polymerases that must process the DNA termini. The
nucleases and polymerases involved at this step are not yet defined.
However, we have identified a structure-specific nuclease called FEN-1
(flap endoNuclease-1) that cleaves flap structures at DNA branch
points. Such a nuclease may account for removal of the type of overhang
shown in Figure 2
(X. Wu, T. Wilson, and M. Lieber, submitted).
Regarding polymerases, we have some data suggesting that polymerase ß
accounts for some of the filling of gaps, but polymerases ß, 
, and
may all contribute (T. Wilson and M. Lieber, submitted).
The last step in restoring chromosomal integrity at a DNA break site is ligation of each of the two DNA strands. Recently, we determined that the DNA ligase responsible for repairing double-strand breaks is DNA ligase IV. In mammalian cells, DNA ligase IV exists in a complex with XRCC4. Without XRCC4, mammalian DNA ligase IV is unstable and has much lower activity.
In eukaryotic cells, absence of any of the components along this pathway results in an inability to repair some or most double-strand breaks caused by ionizing radiation and oxidative agents (such as bleomycin) and breaks caused by physiological processes such as V(D)J recombination and class switch recombination.
DNA Breakage and Rejoining Phases of V(D)J Recombination
All vertebrate animals have a specific immune system in addition
to the nonspecific one that all animals possess. The central feature of
the specific immune system is its ability to generate an enormous
repertoire of antigen receptors by a process called V(D)J
recombination. During this process, any one from an array of V
(variable) gene segments can be joined to one of many D (diversity) or
J (joining) gene segments to generate a new exon encoding the
antigen-binding domains of immunoglobulins or T-cell receptors (Figure 3)
. The V(D)J recombination process
happens millions of times each day in the proliferating pool of
lymphoid precursors of vertebrates. The cis elements
mediating the site specificity of the recombination reaction are
recombination signal sequences (RSSs) flanking the V, D, and J gene
segments. They consist of a palindromic heptamer separated from an
A/T-rich nonamer by a spacer of either a 12- or 23-bp
length.5
A single recombination event is directed by a pair
of these joining signals with different spacer lengths, a restriction
referred to as the 12/23-bp rule. The recombination process is
initiated by two lymphoid-specific proteins that are encoded by the
RAG1 and -2 genes.6,7
Remarkable progress has been made
recently in understanding the mechanism of V(D)J recombination.
|
The first cell-free V(D)J cutting activity has been described by
van Gent and co-workers in the Gellert laboratory using nuclear
extracts from lymphoid cells supplemented with a purified, recombinant
core RAG1 protein.8
They later showed that the RAG proteins
alone were sufficient to catalyze DNA cleavage at RSSs (Figure 4)
.9
The reaction occurs by a
two-step mechanism, during which one DNA strand is initially nicked at
the heptamer of the RSS. The 3'OH group thus generated attacks the
phosphodiester bond of the anti-parallel strand in a direct
transesterification reaction10
leaving hairpinned coding
ends and blunt signal ends (with 5'P and 3'OH). This is reminiscent of
the chemistry involved in nicking and strand transfer seen in reactions
of transposable elements in prokaryotes and in retrotransposons
(including retroviruses) in eukaryotes.11
Nicking and
hairpin formation at isolated signals occurs in a stable RAG-1:RAG-2
DNA complex that disassembles after completion of the
reaction.12,13
|
There are still many remaining questions regarding the initiation of V(D)J recombination. First, is the 12/23 rule enforced at the nicking or hairpinning steps? Second, which regions of RAG-1 and RAG-2 carry out which aspects of the reaction, including binding to the signal, nicking, hairpinning, and 12/23 rule enforcement? Two studies employing either an indirect yeast-one hybrid system or surface plasmon resonance to measure DNA protein interaction suggest that a region in RAG-1 with homology to bacterial Hin recombinases is required for binding to the nonamer of the RSS.18,19 Both studies did not detect DNA-binding activity for RAG-2, and the significance of the highly conserved heptamer sequence of the signals for interaction with RAGs remains unclear. Third, how does the nucleosome affect binding, nicking, and hairpinning? This is critical to understanding how chromatin structure affects V(D)J recombination. A crucial aspect of the regulation of V(D)J recombination is that some antigen receptor loci remain inaccessible whereas others are recombined. Although transcription itself may not control the accessibility for V(D)J recombination,20-23 enhancers clearly do.24-26 Enhancers may direct changes in nucleosome binding and positioning. They may also affect CpG methylation27 and histone acetylation, which have major effects on determining accessibility.28-30 Initial progress has been made at studying the action of exogenous RAG proteins on genomic DNA sites in isolated nuclei.31,32 Cell-free V(D)J recombination using recombinant RAG proteins and nuclei prepared from cells representing different developmental stages indeed recapitulates the tight developmental regulation of antigen receptor rearrangements observed in vivo. Fourth, mutations in the RAG-1 and -2 genes have been shown to account for ~14% of human SCID.33 It will be interesting to see whether components in the joining phase of V(D)J recombination can account for SCID as well.
Rejoining Phase of V(D)J Recombination
After generation of the DNA ds break at RSS, hairpinned coding
ends must be opened to permit their processing and joining (Figure 5)
. Analysis of V(D)J recombination
intermediates in scid mice has demonstrated that these mice are unable
to resolve hairpinned coding end structures at their antigen receptor
loci.34
Scid cells display no DNA-dependent kinase
activity, due to a carboxyl-terminal truncation of the 470-kd DNA-PK
protein encoded by the scid gene.35,36
One possibility is
that DNA-PK is needed to phosphorylate the hairpin endonuclease to
activate it for hairpin opening. Possible candidates for the hairpin
endonuclease are the RAG proteins themselves, because they are the only
known components required for V(D)J recombination that display nuclease
activity. A requirement for RAGs for later stages of V(D)J
recombination has been documented and will be discussed
below.37,38
|
Ku may serve additional functions in V(D)J recombination beyond stimulation of DNA-PK activity. This is suggested by the different phenotypes of mutants in DNA-PK, ie, scid mice, as opposed to Ku-deficient animals. Absence of DNA-PK activity results in failure to form coding joints, but signal joints form at near normal efficiencies.50 In contrast, mutations in Ku affect both signal and coding joints.40 One possible explanation of this is that Ku may dislodge the RAG postsynaptic complex from the signal ends to allow them to be ligated.40 There is precedent for postsynaptic cleavage complexes in prokaryotic recombination systems requiring chaperone proteins to assist in debinding from the signal sequences.51
Once the hairpins are opened, all four DNA ends are available for
modification52
(Figure 6)
.
Terminal deoxynucleotidyl transferase (TdT) is an optional
lymphoid-specific component that can add nucleotides to coding ends to
generate most of the junctional diversity that is a critical extension
to the combinatorial diversity of antigen receptor
assembly.53
A point of mechanistic interest is that TdT can
add N-nucleotides at signals ends as well, even though those
signal ends are not subject to base loss.52
This suggests
that the RAG postsynaptic complex protects them from nuclease action
but not from TdT template-independent nucleotide addition. It has been
shown that a post-cleavage complex protects signal ends in
vitro from exonuclease digestion.54
In contrast, the
coding ends are available for both TdT addition as well as nuclease
action. The nucleases responsible for end nibbling have not yet been
identified with certainty but may include FEN-1. The modification of
all four ends is likely to occur in a complex where all four ends are
in proximity to each other. This is suggested by the occurrence of
nonphysiological alternative products called hybrid
joints.55
These involve the mistaken joining of a coding
end to the other signal end, instead of to the other coding end. Such
an occurrence would require that all four ends remain in proximity
until ligation. Using a cell-free system for concerted cleavage at two
signals, it has now been demonstrated, in crude extracts, that the
signals remain synapsed in a post-cleavage complex.54
|
Class Switch Recombination
Just as V(D)J recombination can be divided into the cutting phase
and the joining phases, class switch recombination can also be
dissected into at least these two stages (Figure 7)
. Concerning the joining phase, there
are indications that this process is like that for general NHEJ. There
are indications of the involvement of DNA-PK in class switch
recombination because cells mutant for DNA-PK (murine scid cells) fail
to carry out class switch recombination in a primary cell culture
system.56
More recent data show that class switch fails to
occur in cells missing Ku.73
Along with previous data
demonstrating deleted circular DNA in switching cells, the requirement
for DNA-PK is compelling evidence that the joining phase of class
switch recombination is that of general NHEJ.
|
Chromosomal Translocations Arising during the Course of V(D)J Recombination and Class Switch Recombination
Translocations Associated with V(D)J Recombination
Many chromosomal translocations in lymphoid cells arise as
mistakes during the process of either V(D)J recombination or class
switch recombination.62
One of the most common of these and
most well characterized is one that occurs in essentially all
follicular lymphomas, the t(14;18) translocation involving bcl-2 and
the IgH locus (Figure 8)
. One scenario
for this translocation is as follows (Figure 8A)
. The cutting event at
the D to J rearrangement step of the IgH locus fails to complete.
Cutting at the bcl-2 locus occurs and may be catalyzed by the RAG
proteins inadvertently. Finally, the ends of the break at the bcl-2
locus become joined to the D and J segments instead of the D and J
being joined.
|
. The RAG
proteins may synapse the J segment recombination signal with a cryptic
signal at the bcl-2 locus. Recent data indicate that cryptic sites
consisting simply of CAC (the essential portion of the heptamer) may be
sufficient to permit synapsis with a partner signal at the antigen
receptor loci.63,64
Once the first step is completed, a
signal joint has been formed at the derivative 18 chromosome. Signal
joints can carry out secondary rounds of rearrangement in which one of
the signals (at d in Figure 8B
) can synapse with the DH
segment signal in a second step. The second step would account for the
configuration for the derivative chromosome 18 in cases for which this
half of the translocation event has been sequenced.65
The model in Figure 8B
is more consistent with what is currently known
about V(D)J recombination and about chromosome breakage. This model
would account for the initiation of breakage at the bcl-2 locus. It
would also reduce the complexity of the process into two steps rather
than trying to explain three concurrent breakages (one at D, one at J,
and one at the bcl-2 locus). This model would predict that the
derivative chromosome 18 could exist in two states, one before and one
after the second step. However, the second step may be quite efficient;
hence, it may be difficult to detect the earlier configuration of the
derivative 18.
One major uncertainty concerns the predilection for breakage at the bcl-2 locus. The promiscuity of the RAG proteins for cleaving at accessible CAC sites is noteworthy and creates a paradox as to how our lymphoid cells survive such promiscuity. It appears that most of the genome is protected by some degree of chromatin structure. The interplay between the RAG cleavage and chromatin structure is an interesting area for future work. Breaks that do not involve adventitious CAC sites may be caused by other processes. Fragile sites within DNA and chromatin are another area of intense interest. Some of these fragile sites may relate to how the G-rich class switch sequences are cleaved.
Translocations Involving Class Switch Sequences
The best characterized translocation involving class switch
sequences is that of c-myc. In sporadic Burkitt's lymphoma, c-myc
typically translocates to the Sµ sequences located upstream of the
constant Igµ region. This process appears to arise as a mistake of
class switch recombination. Instead of the Sµ and one of the
downstream switch regions (such as S
, S
, or S) recombining,
the break site at the Sµ region recombines with a break that arises
at the c-myc gene on chromosome 8. A reciprocal translocation results
between chromosomes 8 and 14 with most of the c-myc coding region now
joined to the IgH locus. A diagram of the translocation process is
shown in Figure 9
(top line). (In endemic
Burkitt's lymphoma, the translocation process typically occurs as a
mistake of V(D)J recombination and gives the type of structural
outcomes shown in the second through fourth lines of Figure 9
.) Because the mechanism of switch recombination is
not understood, the mechanism of the c-myc translocation process is
also not defined.
|
In addition to repair of double-strand DNA breaks by nonhomologous DNA end joining, some double-strand DNA breaks are repaired by homologous recombination (HR). The precise balance of homologous and nonhomologous recombination in cells is not yet fully determined.4,66 It appears that during G0, G1, and early S phases of the cell cycle in vertebrate cells (and those of most multicellular eukaryotes), NHEJ is the dominant pathway. During late S and G2 phases of the cell cycle, HR plays a significant role.67 The enzymes involved in HR include RAD51, -52, -54, -55, and -57.68 BRCA1, BRCA2, and ATM may also play roles in this process. HR may be ideally suited for late S and G2 when the chromatids are aligned. Teleologically, it may be that as eukaryotic genomes became rich with repetitive DNA (which is 40% in humans), HR became less appropriate as a method for repairing double-strand breaks; otherwise, the number of potential, but not ideal, donors for breaks within the repetitive 40% of the genome may have made the donor search process extremely long. Moreover, some HR events involving the wrong donor repeat might lead to chromosomal translocations. NHEJ may be the safer alternative in multicellular eukaryotes during phases of the cell cycle when chromatids are not aligned. This is despite the fact that NHEJ is almost certain to result in the loss of some genetic information between the two DNA ends.
Some protein complexes, such as RAD50-Mre11-XRS(p95) complex, are involved in double-strand break repair, but it is not yet clear whether they are involved in NHEJ or HR or both.69
NHEJ and Aging
As mentioned earlier, defects in the cleavage phase of V(D)J recombination can result in human SCID. Defects in the synapsis phase of V(D)J recombination may result in chromosomal translocations that yield follicular lymphomas, and similar events may occur for switch recombination in the case of sporadic Burkitt's lymphoma.
However, a broader issue in human disease is the extent to which double-strand breaks generate irreversible changes in the genome over time. Any irreversible change in any nondegradable macromolecule may be contributory to the aging process. It is currently unclear as to the extent of the contribution of changes in DNA to aging. However, DNA is the one macromolecule in the cell whose information content cannot be regenerated once it is lost.
Single-strand DNA damage is not as relevant because the other strand provides for the information content. However, double-strand breaks typically result in the loss of the information between the two ends during NHEJ. Hence, double-strand breaks are the likely candidate for aging of DNA.
The one tangible observation thus far is that cells lacking the ability to carry out NHEJ appear to senesce in culture more quickly41,70 (P. Hasty, personal communication). Additional studies will be needed to define whether the basis for this is actual DNA damage.
A key pending question related to this issue is how frequently double-strand breaks are created in the genome from free radicals generated by oxidative respiration and from ionizing radiation. If they are rare, they represent no significant contributory cause of aging. At this point, we do not yet know the rate at which double-strand breaks occur. This could be high and yet our NHEJ (and HR) may be sufficiently effective so as to repair the vast majority of these to restore the integrity of the chromosome. Nevertheless, restoration of chromosomal integrity would still be accompanied by loss of genetic information at the nucleotide level, and this loss is irreversible. There may soon be methods to measure the background level of double-strand breaks caused by oxidative free radicals arising from respiration and from ionizing radiation. Only at that point will we be able to evaluate this issue in a meaningful way.
Concluding Remarks
Biochemical definition of the steps and details of physiological and pathologic double-strand DNA breaks is still a young field. Yet there have been many rewards for this field concerning human disease. This may continue as we examine the relationship to broader issues, such as the time-dependent DNA changes that may occur with aging.
|
I thank Drs. H. T. Blumenthal, T. Steck, M. Gellert, and K. Mizuuchi.
Footnotes
Address reprint requests to Dr. Michael R. Lieber, Department of Pathology, Norris Comprehensive Cancer Center, University of Southern California School of Medicine, 1441 Eastlake Avenue, Los Angeles, CA 91007. E-mail: lieber_m{at}froggy.hsc.usc.edu
Supported by NIH grants and by the Rita and Edward Polusky Basic Cancer Research Professorship. The author is also a Leukemia Society of America Scholar.
Accepted for publication July 13, 1998.
References
gene enhancers confer tissue- and stage-specificity to V(D)J recombination events. EMBO J 1993, 12:4335-4346[Medline]
chain enhancer sequence. Cell 1994, 76:913-923[Medline]
heavy chain switching. Immunity 1996, 5:319-330[Medline]
This article has been cited by other articles:
 |
|
 |
|
  H. Li, H. Vogel, V. B. Holcomb, Y. Gu, and P. Hasty Deletion of Ku70, Ku80, or Both Causes Early Aging without Substantially Increased Cancer Mol. Cell. Biol., December 1, 2007; 27(23): 8205 - 8214. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Kitagawa, K. Inoue, S. Sasaki, Y. Hayashi, Y. Matsuo, M. R. Lieber, H. Mizoguchi, J. Yokota, and T. Kohno Prevalent Involvement of Illegitimate V(D)J Recombination in Chromosome 9p21 Deletions in Lymphoid Leukemia J. Biol. Chem., November 22, 2002; 277(48): 46289 - 46297. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Li and L. Comai Displacement of DNA-PKcs from DNA ends by the Werner syndrome protein Nucleic Acids Res., September 1, 2002; 30(17): 3653 - 3661. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Tanaka, S. J. Tapscott, B. J. Trask, and M.-C. Yao Short inverted repeats initiate gene amplification through the formation of a large DNA palindrome in mammalian cells PNAS, June 25, 2002; 99(13): 8772 - 8777. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Ma and M. R. Lieber Binding of Inositol Hexakisphosphate (IP6) to Ku but Not to DNA-PKcs J. Biol. Chem., March 22, 2002; 277(13): 10756 - 10759. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Yu, A. Taghva, and M. R. Lieber The Cleavage Efficiency of the Human Immunoglobulin Heavy Chain VH Elements by the RAG Complex. IMPLICATIONS FOR THE IMMUNE REPERTOIRE J. Biol. Chem., February 8, 2002; 277(7): 5040 - 5046. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Marculescu, T. Le, P. Simon, U. Jaeger, and B. Nadel V(D)J-mediated Translocations in Lymphoid Neoplasms: A Functional Assessment of Genomic Instability by Cryptic Sites J. Exp. Med., January 7, 2002; 195(1): 85 - 98. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. MUTIRANGURA Serum/Plasma Viral DNA: Mechanisms and Diagnostic Applications to Nasopharyngeal and Cervical Carcinoma Ann. N.Y. Acad. Sci., September 1, 2001; 945(1): 59 - 67. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Yu and M. R. Lieber The Nicking Step in V(D)J Recombination Is Independent of Synapsis: Implications for the Immune Repertoire Mol. Cell. Biol., November 1, 2000; 20(21): 7914 - 7921. [Abstract] [Full Text]
|
|
|
|
|
|
D.-S. Lim, H. Vogel, D. M. Willerford, A. T. Sands, K. A. Platt, and P. Hasty Analysis of ku80-Mutant Mice and Cells with Deficient Levels of p53 Mol. Cell. Biol., June 1, 2000; 20(11): 3772 - 3780. [Abstract] [Full Text]
|
|
|
|
|
|
K. Yu and M. R. Lieber Mechanistic Basis for Coding End Sequence Effects in the Initiation of V(D)J Recombination Mol. Cell. Biol., December 1, 1999; 19(12): 8094 - 8102. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Vogel, D.-S. Lim, G. Karsenty, M. Finegold, and P. Hasty Deletion of Ku86 causes early onset of senescence in mice PNAS, September 14, 1999; 96(19): 10770 - 10775. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Li and L. Comai Functional Interaction between Ku and the Werner Syndrome Protein in DNA End Processing J. Biol. Chem., September 8, 2000; 275(37): 28349 - 28352. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. C. Raghavan, I. R. Kirsch, and M. R. Lieber Analysis of the V(D)J Recombination Efficiency at Lymphoid Chromosomal Translocation Breakpoints J. Biol. Chem., July 27, 2001; 276(31): 29126 - 29133. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||
chain
loci) and all three TCR loci
(
