Non-homologous End Joining

Yeast, as all eukaryotes, have acquired many different mechanisms to repair DNA double strand breaks (DSBs) in chromosomes. Several types of break-induced homologous recombination have been described in the Introduction such as gene conversion (SDSA), break-induced replication (BIR) and single-strand annealing (SSA), all of which depend on the RAD52 gene and are independent of the DNA end-joining proteins yKu70 and yKu80.

An alternative set of non-homologous, or illegitimate recombination processes can rejoin DNA ends in the absence of significant homology. There are at least four such NHEJ processes, all of which are independent of RAD52 and RAD51, but dependent on the specialized DNA ligase IV (Dnl4) and its associated Xrcc4 homolog, Lif1.   Dnl4 and Lif1 interact with a third protein, Nej1.  In budding yeast, diploid cells expressing both MATa and MATa repress expression of NEJ1; hence end-joining is turned off in diploids and in haploids expressing the a1-a2 repressor.   Consequently non-homologous end-joining (NHEJ) is repressed in haploids with mutations in the SIR genes, when the normally silent HMLa and HMRa loci are expressed.

Of the 4 NHEJ processes in budding yeast three require the Ku proteins. One pathway is the precise joining (re-ligation) of short overhanging, complementary ends, such as those produced by site-specific endonucleases. This is a relatively efficient process.  When a DSB is created by the HO endonuclease at MATa, about 10% of the cells re-ligate the ends while the remaining 90% undergo homologous recombination with HML or HMR.  When the HML and HMR donors are deleted re-ligation increases to about 15%.  The limitation is that the DSB ends are resected by 5’ to 3’ nucleases; apparently as ssDNA regions increase efficient NHEJ decreases.  In G1-arrested cells, where resection is turned off in the absence of CDK1 activity, NHEJ efficiency rises to 65%.   Re-ligation also requires the Mre11-Rad50-Xrs2 (MRX) complex. 

When the ends are not fully complementary or when the continued presence of an endonuclease precludes precise re-ligation (because the ligated sequence is re-cut), there are two additional, but much less efficient, NHEJ processes.  One process involves misalignment of overhanging ends, apparently by pairing of as few as one base pair, followed by filing-in by a DNA polymerase, and results in the ligation of ends with the insertion of a few bp (see illustration below).  This process is strongly cell cycle regulated and requires MRX.   The filling-in of misaligned ends leads to the creation of small templated insertions of 2 or 3 bp (see illustration).  This process requires the specialized DNA polymerase, Pol4.  Alternatively, annealing of microhomologies of one to a few bp and removal of single-stranded tails, leads to the formation of deletions ranging from a few bp to several kb.  This process requires Ku but not MRX.  Both of these NHEJ mechanisms are inefficient, allowing only about 1 in 1000 cells to survive.

A fourth process, that is independent of Ku but dependent on MRX proteins, is termed Microhomology-Mediated End-Joining (MMEJ) and has been characterized primarily in the lab of Sang Eun Lee.  MMEJ is seen when the 3’ overhanging ends have no complementary base-pairs and in the absence of Ku proteins.  Whereas the absence of Ku reduces “normal” NHEJ by 100-fold, MMEJ efficiencies are as high as those seen with NHEJ using partially complementary ends (about 1 in 1000).  MMEJ requires MRX as well as Rad1-Rad10, the nuclease that can cleave off 3’ ended ssDNA tails that would be generated by base-pairing of the resected ends.  MMEJ generally have junctions with 5 or more bp.  

NHEJ and MMEJ in fission yeast and mammalian cells. Both Ku-dependent NHEJ and Ku-independent MMEJ have been reported in mammalian cells.  Ku-dependent events also require DNA-PKcs (not present in budding yeast).  Most NHEJ events require DNA ligase 4, Xrcc4 and the (highly diverged but recognizable) Nej1 homolog known as XLF or Cernunnos.  Fission yeast also have XLF but curiously may not have Xrcc4.  Because the Mre11-Rad50-Nbs1 (MRN) complex is essential in mammals, the role of MRN has been hard to assess with certainty.  In fission yeast, MRN is not required for most NHEJ events, but those observed in G1-arrested cells do require MRN.  There are still NHEJ events in the absence of mammalian DNA ligase 4.  These may use DNA ligase 3.  These events seem to occur by MMEJ.

Systems to study NHEJ. The HO endonuclease cleaves the MAT locus to leave a 4 bp, 3’ overhanging sequence, AACA. In the presence of the unexpressed donor sequences HML and HMR, the HO-cleaved MAT locus recombines and switches mating-type. However, when both HML and HMR are deleted, the cell can only repair the DSB by nonhomologous recombination or by re-ligation. When HO is continuously expressed from a galactose-inducible promoter, simple re-ligation is prevented, since the restored recognition site will be cut again; consequently, survivors, arising at a frequency of about 2 in 1000, have mutations that alter the HO cleavage site. The majority have 2- and 3-bp insertions of CA and ACA arising by misalignment and filling-in of the overhanging ends that create MATa1 mutations. The rest have deletions ranging from 3 bp (-ACA) to several kb, usually containing between 1 and 5 bp of microhomology at the junction. Deletions of less than about 300 bp also create mata1 mutations, while larger mutations eliminate both MATa1 and MATa2 and produce an a-mating phenotype.

Model of Basic NHEJ Pathway. The mechanistic steps and intermediates of NHEJ remain unclear, even though it has been extensively studied. The following is a working model incorporating the basic features of NHEJ as proposed by Michael Lieber, providing a good overview of the mechanism of NHEJ.