E. coli UvrA protein
Wagner, K.
Citation
Wagner, K. (2011, February 17). Nucleotide excision repair at the single-molecule level : analysis of the E. coli UvrA protein. Retrieved from https://hdl.handle.net/1887/16502
Version: Not Applicable (or Unknown)
License: Leiden University Non-exclusive license Downloaded from: https://hdl.handle.net/1887/16502
Note: To cite this publication please use the final published version (if applicable).
SUMMARY AND GENERAL DISCUSSION
In this thesis, the characteristics of the Escherichia coli UvrA protein were analyzed with microscopy techniques that allow detection of protein complexes at the single-molecule level.
Together with UvrB and UvrC, UvrA catalyzes the excision of damaged DNA from the bacterial genome. This DNA repair mechanism is called Nucleotide Excision Repair (NER) and is also conserved in higher organisms. In the current model for bacterial NER, DNA repair is initiated by a complex of two UvrA and two UvrB subunits, the A2B2-complex. In this complex, the two UvrA subunits probe the genome for potential lesions. After finding such a site, UvrA tries to load UvrB onto the DNA; UvrB verifies the presence of DNA damage. After successfully loading UvrB, the UvrA dimer dissociates from the DNA, leaving the UvrB2-DNA pre-incision complex. This complex is the binding target for UvrC, which incises the damaged DNA strand on both sides of the lesion.
UvrA has two ATPase domains, which both belong to the ABC (ATP Binding Cassette) superfamily of ATPases. This type of ATPase is also found in the repair proteins MutS and Rad50. UvrA hydrolyzes ATP already in the absence of DNA. This stabilizes the UvrA dimer, since in the presence of ATP more UvrA dimers were detected than in the presence of ADP or ATPγS (a non-hydrolysable ATP analog). This indicates that in the most stable form of the UvrA dimer not all bound ATP is hydrolyzed. This form of the UvrA dimer likely contains a mixture of ATP and ADP, although it is unclear how ADP and ATP are divided over the four ATPase domains (chapter 2).
The ATPase activity of UvrA is coupled to DNA binding. When bound to undamaged DNA, ATP hydrolysis is inhibited (chapter 4). The lower ATPase activity of UvrA on undamaged DNA indicates that UvrA consumes no ATP during the scanning for lesions.
Also, the absence of a cofactor does not affect the specificity of UvrA for Menthol- or Cholesterol-DNA, which are both large and bulky DNA lesions. This confirms that binding or hydrolysis of ATP does not contribute to the scanning of DNA for the presence of damage (chapter 2). With fluorescence microscopy, the DNA scanning mechanism of UvrA was imaged, using UvrA labeled with a fluorescent dye.
Theoretically, the search for DNA lesions could occur via a combination of three different search modes: Binding and dissociation at non-specific sites (3D-diffusion), sliding along undamaged DNA (1D-diffusion) and the simultaneous binding of two non-specific sites (intersegmental transfers). We showed that the search for DNA lesions is a very fast process, as in our experiments the vast majority of all observed UvrA-complexes had already bound a preferential binding site before imaging could be started. In our experimental setups no sliding of UvrA along DNA could be detected, but we did find evidence for 3D-diffusion: A small number of events were recorded in which UvrA, after having dissociated from DNA, quickly re-associated on a different DNA site, thus performing 3D-diffusion (chapter 3).
With Atomic Force Microscopy (AFM), the stoichiometry of the UvrA-complex on DNA was determined, finding that only the UvrA dimer stably binds DNA. Furthermore, on the AFM images a significant part of the UvrA-DNA complexes were detected at the ends of the linear DNA substrate. This indicates that, as the DNA helix is partially unwound at a DNA end, UvrA recognizes the destabilization of basepair interactions in double-stranded DNA that is caused by the presence of a DNA lesion or an end. When the UvrA dimer binds a DNA end only one subunit contacts DNA. This is indicated by the detection of UvrA-complexes that appeared to ‘bridge’ two DNA ends (chapter 2).
Notably, the specificity of UvrA for DNA ends did not depend on the presence or absence of a cofactor. Since ‘finding’ DNA ends requires multiple rounds of association and dissociation from undamaged DNA, this means that ATP binding or hydrolysis does not play a role in the search for DNA lesions. Binding/hydrolysis of ATP however does contribute to the binding the damage itself, as in the presence of ATP and in the absence of a cofactor, UvrA had a higher specificity for DNA damage than when ADP or ATPγS was present (chapter 2).
In contrast to binding DNA ends, no ‘bridges’ could be detected between UvrA- complexes bound to an internal lesion. This suggests that when the UvrA dimer binds a damaged site, both subunits contact the DNA flanking the lesion. Apparently when the UvrA dimer is occupied with only ADP or ATPγS, the UvrA dimer has a different structure than the ATP/ADP-mixed form and has a less optimal configuration for binding DNA with both subunits (chapter 2).
When UvrA binds damaged DNA, ATP hydrolysis is activated. This is particularly important for binding non-bulky lesions, as in the absence of a cofactor UvrA recognizes the non-bulky CPD damage with lower specificity than in the presence of ATP.
Since CPD-DNA is the result of UV-irradiation of DNA, the recognition of this ‘non-bulky’
lesion is likely more relevant to DNA repair in vivo than the recognition of synthetic ‘bulky’
lesions such as Cholesterol-DNA that do not occur in vivo.
The role of each individual ATPase domain in UvrA in damage recognition was studied using two mutants, in which one of both ATPase domains was deactivated (UvrA K37A and UvrA K646A). Both mutants were completely disturbed in ATPase activity, demonstrating that the two domains are tightly coupled. Further analysis indicated that both K37A and K646A cannot bind ATP in their mutated ATPase domain. In the presence of ATP both mutant proteins were not affected in binding the bulky Cholesterol lesion. However, UvrA K37A (and, likely, K646A) was disturbed in binding the non-bulky CPD-damage. Likely, ATP hydrolysis in damage-bound UvrA results in opening the DNA around the lesion. This particularly stimulates binding of non-bulky lesions, as the DNA around this type of damage is generally less unwound than the DNA around a more distorting lesion (chapter 4).
During the preparation of this thesis, the crystal structure of the UvrA dimer was solved.
In this crystal structure, three functional domains of UvrA were identified: The UvrB-binding domain, the insertion domain (ID) and the zinc-finger motif. Similar to the structures of other ABC ATPases, these functional domains are inserted within the ATPase domains of UvrA.
ATPase domain I (residue K37 is part of this domain) contains the UvrB-binding domain and the ID, while the zinc-finger motif is inserted in ATPase domain II (which contains residue K646). Furthermore, two patches of positively-charged residues (one in each monomer) were identified in the crystal structure of UvrA. These patches are surface-exposed and are able to bind DNA. These DNA-binding patches facilitate the initial recognition of damaged DNA, before ATP hydrolysis is activated (chapter 5).
The UvrB-binding domain is a large domain within UvrA and it is, most likely, the only domain that directly associates with UvrB. The zinc-finger motif of UvrA does not resemble the structure and function of a generic zinc-finger, which in other proteins is shown to bind DNA. The zinc-finger motif of UvrA does not have DNA binding activity. Instead, we have shown that this domain is part of the dimer-interface of UvrA and controls the coupling of DNA binding with ATP hydrolysis. When the zinc-finger motif is removed, the ATPase of UvrA no longer responds to the binding of (damaged) DNA (chapter 5).
In E. coli UvrA, the ID is a relatively large domain, but the size and the amino acid sequence of this domain are poorly conserved between different bacterial species. In this thesis, we have demonstrated that the ID of UvrA however plays an important role in binding
In this domain two conserved arginine residues are present and these residues contact DNA.
Before ATP hydrolysis the two IDs of the UvrA dimer likely form a clamp around the DNA, the presence of this clamp positions the DNA such that the DNA binding patches of both monomers can contact DNA. After ATP hydrolysis the ID occupy a new position; this likely results in the local unwinding of DNA that stabilizes the UvrA-complex on non-bulky lesions (chapter 5).
Based on the work presented in this thesis, we have proposed a new model for damage recognition by UvrA. Initially, one of the DNA binding patches binds DNA. The binding of DNA to one DNA binding patch recruits the IDs to the DNA, which clamp around the DNA.
In the UvrA-DNA complex, the two DNA binding patches both try to contact DNA. If both patches can stably bind DNA, this indicates the presence of a lesion. The binding of both monomers results in a structural rearrangement, which separates the two zinc-finger motifs in the UvrA dimer. Separation of the zinc-finger motifs activates ATP hydrolysis. As a result of ATP hydrolysis the ID is repositioned. Because of the movement of the ID, the arginine residues in this domain, which contact DNA, move apart, generating a pulling force on the DNA that separates the two strands, thereby stabilizing the complex on non-bulky lesions.
Likely, the movement of the ID is the result of hydrolysis in ATPase domain I, since the ID (which facilitates the ATPase-dependent binding of non-bulky lesions) is inserted into this ATPase domain (chapter 5).
The two ATPase domains also contribute differently to the interaction with UvrB. Both ATPase mutants are greatly reduced in binding UvrB in solution, although K37A was less affected in binding UvrB than K646A. Even though both mutants have a greatly reduced affinity for UvrB, they still showed considerable incision activity; indicating that both mutants are still able to load UvrB to damaged DNA. Remarkably, the incision rate of both mutant proteins was related to the size of the used substrate. When a long (678 bp) fragment was used, UvrA K646A had a higher activity than K37A; on a short (50 bp) substrate however K37A was more proficient in incision. As both mutants had similar damage specificities, this indicates that size of the DNA differently affects the ability of both mutants to load UvrB to the damage (chapter 4). When UvrB binds DNA approximately 70 basepairs wrap around the protein, which helps UvrB to bind DNA. Therefore, the reduced incision activity of K37A on a longer substrate suggests that, when this mutant binds UvrB, UvrB is not in a position compatible with DNA wrapping. For K646A however, the opposite occurs.
The higher incision of this mutants on longer DNA suggests that, although this mutant is severely affected in binding UvrB, K646A binds UvrB such that DNA can wrap around the protein. The different roles of both ATPase domains in binding UvrB might reflect changes in the interaction between UvrA and UvrB during the loading of UvrB to a damaged site and the subsequent release of UvrA (chapter 4).
Interestingly, in vivo the activity of UvrABC is stimulated by proteins with a repair- helping function, such as photolyase or alkyltransferase-like (ATL) proteins. We have shown that photolyase specifically stimulates UvrA to bind CPD lesions. Photolyase (and presumably ATL proteins) likely accomplishes this through creating a larger DNA distortion around damaged DNA. Removal of the ID, which essentially is a large DNA binding domain, increases the ability of UvrA to bind photolyase-bound DNA lesions. This suggests that the ID actually obstructs the binding of UvrA to damaged sites that are bound by a ‘repair helping’ protein. Some bacterial species carry a UvrA homolog (UvrA class III) that does not contain an ID. Likely, these class III UvrA homologs do not have a primary function in NER, but might function to initiate repair at sites that are bound by repair helping proteins (chapter 5).
