Academic Background

Dr. Knight received his Ph.D. from the Department of Biochemistry at U.C. Riverside in 1981. He worked as an American Cancer Society post-doctoral fellow with Kevin McEntee at the U.C.L.A. School of Medicine from 1981-1985, and was awarded a Charles A. King Trust research fellowship for post-doctoral work with Bob Sauer in the Department of Biology at M.I.T. from 1985-1989.

Associate Dean of the GSBS Basic and Biomedical Sciences Division & UMMS Assistant Vice-Provost of Admissions

Homologous recombinational DNA repair in human cells

We are interested in understanding the catalysis and regulation of Rad51-mediated DNA repair in human cells.  Our research involves biochemical and cell biological studies of Rad51, as well as the contribution of various Rad51-interacting proteins to the overall mechanism of Rad51 function, e.g. Rad51C, Xrcc3, Brca2, etc.  We are also engaged in biochemical and biophysical studies of the molecular design of proteins that participate in the catalysis of homologous genetic recombination and recombinational DNA repair.  Our research takes advantage of a number of experimental approaches in genetics,RNAi, cell biology, microscopy, molecular biology, protein chemistry/enzymology and biophysics aimed at providing a molecular and cellular description of the catalytic organization of the Rad51 protein and Rad51 protein complexes.  

Human Rad51 and related proteins

The majority of our work focused on studies of the coordinated action of the human Rad51 and the Rad51 paralog proteins (Xrcc3 and Rad51C).  In our initial work we took advantage of several improvements we made in immunofluorescence methods to show that Rad51, Rad51C, and Xrcc3 have a significant cytoplasmic presence in human cells, and that as a function of DNA damage Rad51 and Rad51C are re-distributed toward the nuclear periphery and into the nucleus (Figs. 1 & 2, and see Forget, Bennett & Knight, 2004; Bennett & Knight, 2005).  Further studies revealed that Rad51C contributes to a DNA damage-induced and BRCA2-independent increase in nuclear levels of Rad51 (Gildemeister, Sage & Knight, 2009).  An unexpected outcome of these studies was our discovery that human Rad51, Rad51C and Xrcc3 function in mitochondria to maintain mtDNA copy number in response to DNA damage.  We found that DNA damage results in increased mitochondrial levels of each protein as well as an increased physical association of Rad51 with mtDNA (Sage, Gildemeister & Knight, 2010).  Additionally, we showed that Rad51 promotes mtDNA synthesis under conditions of replication stress in either the presence or absence of DNA damage (Sage & Knight, 2013).

For our mutagenesis studies of human Rad51 we are creating single and multiple mutations in targeted domains, e.g. the ATP and DNA binding sites, the oligomeric interface, protein-protein interaction domains, etc. to identify important determinants of different aspects of protein structure and function.  Specific mutant proteins are chosen to analyze using cell-based recombination and DNA repair assays, and for purification and detailed in vitro analyses of their biochemical and structural properties.  Our recent study of mutations in the Rad51 ATP binding site revealed an unexpected result that, in fact, supports a long-standing model for how recombinase enzymes such as Rad51 and bacterial RecA use the energy derived from ATP binding and turnover/release to promote recombination.  We found that a mutant Rad51 carrying a Lys-to-Ala substitution in the Walker A motif of the ATP binding site was partially functional in a cell-based DNA repair assay (see Forget et al., 2007).  In this work we showed that the purified protein binds ATP but with an affinity approximately100-fold lower than wild type Rad51 (Fig. 3).  This was important because it is commonly assumed that this mutation prohibits ATP binding in any ATP-binding protein into which it is introduced.  Therefore, our data supports a model in which ATP binding promotes nucleoprotein filament assembly and function in the recombination process, while ATP hydrolysis and release of ADP (or simple release of non-hydrolyzed ATP in the case of the Lys-to-Ala mutant protein) promotes filament disassembly.

4Pi microscopy

We published the first successful use of high-resolution 4Pi microscopy to visualize endogenous nuclear proteins (see Bewersdorf, Bennett & Knight, 2006).  4Pi microscope technology was created in the laboratory of Stefan W. Hell, and offers a significant improvement in z-axis resolution (the optical axis).  The illuminating light from two perfectly aligned opposing objective lenses interferes constructively within the sample, thereby improving resolution (Fig. 4).  A similar constructive interference of the collected fluorescent light adds to the increase in resolution, providing an approximate 6-fold improvement in z-axis resolution over standard confocal microscopes.  http://www.mpibpc.gwdg.de/groups/hell/4Pi.htm

Our first efforts were directed at imaging histones H2AX and g-H2AX, a phosphorylated version of H2AX that appears immediately following DNA damage and serves as a mark for the site of a DNA break.  This work supports a model in which chromatin structures carrying clusters of H2AX serve as platforms for the immediate response to DNA damage, and also serve to limit the spread of H2AX phosphorylation (Figs. 5 & 6; movies 1 & 2; see Bewersdorf, Bennett & Knight 2006).

Movie One

Three-dimensional video representation of a 4Pi microscopy data set showing nuclear distribution of H2AX and g-H2AX at 15 min after exposure of cells to ionizing radiation. The movie is based on the same data set used in Fig. 1 (15 min) and was created with the rendering software Amira. H2AX is in green and g -H2AX is in red. The grid has a 1-m m raster and axes are as indicated in Fig. 1 in the full manuscript (Bewersdorf et al. 2006).

Movie Two

Three-dimensional video representation of a 4Pi microscopy data set showing nuclear distribution of H2AX and g-H2AX at 180 min after exposure of cells to ionizing radiation. The movie is based on the same data set used in Fig. 1 and was created with the rendering software Amira. H2AX is in green and g -H2AX is in red. The grid has a 1-m m raster and axes are as indicated in Fig. 1 in the full manuscript (Bewersdorf et al. 2006).

Bacterial RecA

Our studies of genome stability in human cells and Rad51-mediated DNA repair derive from work on the related bacterial RecA protein.  Both RecA and human Rad51 bind to single-stranded regions at DNA breaks and form a helical nucleoprotein filament, which then initiates strand exchange and transfer of information between homologous DNA molecules, most frequently a sister chromatid following DNA replication (McGrew & Knight, 2003).  RecA and human Rad51 have similar structural and functional properties, but also show distinct differences in their structure-function relationships (see De Zutter & Knight, 1999).  We showed that Phe217 is a critical subunit interface residue in RecA that regulates the flow of ATP-mediated allosteric information throughout the protein filament (see De Zutter et al., 2001).  However, the human Rad51 protein does not have an aromatic residue conserved at this position, and any ATP-mediated allosteric effect on its overall function appears to be significantly more subtle than that observed for RecA (Fig 3 above; see Forget et al., 2007).

In our final project with bacterial RecA, we addressed a long-standing question in the recombination field – what is the protein unit that initiates filament assembly on ssDNA?  Because purified RecA protein exists as a heterogeneous mix of various oligomeric states, e.g. monomer, dimer, trimer, hexamer, short filaments, etc., it has been impossible to confirm the original idea that a monomer of RecA is the nucleating unit for filament assembly.  However, we have recently published a study in which we created a fused RecA dimer and showed that this protein is perfectly capable of forming filaments on ssDNA and catalyzing recombinational DNA repair (see Forget et al., 2006).  Therefore, in the cell, where it is important for RecA to respond immediately to DNA damage, it is very likely that a mixture of monomers, as well as pre-formed dimers and perhaps trimers of RecA nucleate filament formation.

In addition to providing a detailed molecular understanding of homologous genetic recombination, and the mechanisms by which multifunctional, oligomeric enzymes and protein complexes coordinate a variety of activities to carry out the ordered enzymatic process, our work will offer insights into processes critical for maintenance of genome stability. 

Movie One

Movie Two