Direct correlation of DNA binding and single protein domain motion via dual illumination fluorescence microscopy

Abstract

We report a dual illumination, single-molecule imaging strategy to dissect directly and in real-time the correlation between nanometer-scale domain motion of a DNA repair protein and its interaction with individual DNA substrates. The strategy was applied to XPD, an FeS cluster-containing DNA repair helicase. Conformational dynamics was assessed via FeS-mediated quenching of a fluorophore site-specifically incorporated into XPD. Simultaneously, binding of DNA molecules labeled with a spectrally distinct fluorophore was detected by colocalization of the DNA- and protein-derived signals. We show that XPD undergoes thermally driven conformational transitions that manifest in spatial separation of its two auxiliary domains. DNA binding does not strictly enforce a specific conformation. Interaction with a cognate DNA damage, however, stabilizes the compact conformation of XPD by increasing the weighted average lifetime of this state by 140% relative to an undamaged DNA. Our imaging strategy will be a valuable tool to study other FeS-containing nucleic acid processing enzymes.

Keywords: DNA damage recognition; DNA repair; XPD helicase; multicolor single-molecule detection; protein domain motion; total-internal reflection fluorescence microscopy.

Figure 1

Dual illumination TIRF microscopy setup for simultaneous visualization of domain motion of XPD protein and its interaction with DNA. (a) In vivo, XPD is expected to bind a bubble DNA structure. XPD is an important player in the nucleotide excision repair (NER) pathway where it functions as a component of the TFIIH (transcription factor II–H) complex (schematically depicted as a gray rectangle) together with the second NER helicase XPB. Its native substrate is a DNA bubble. In the XPD–DNA complex, the translocating DNA strand passes through the central pore formed by the ARCH, FeS, and HD1 domains into the secondary DNA binding site at the interface of the FeS domain and HD1. (b) Illustration of the dual illumination single-molecule imaging scheme. Cy3 and Cy5 dyes were simultaneously excited by green (532 nm) and red (640 nm) lasers, respectively, in total internal reflection (TIR) mode. Cy3-AA1-XPD molecules were immobilized on surface of the TIRFM imaging chamber through biotin (b)–Neutravidin (NA) linkage. Transition between “OPEN” (highly fluorescent) and “CLOSED” (quenched) conformations of the ARCH domain of XPD was monitored in the Cy3 emission channel (577 ± 10 nm). The Cy3 fluorescence is quenched and recovers as ARCH domain moves toward and away from the iron–sulfur (FeS) cluster, respectively. Simultaneously, the DNA binding to (“ON”) and dissociation from (“OFF”) the surface-tethered XPD can be monitored in the Cy5 emission channel (690 ± 25 nm). DNA bubble construct used in this study is schematically depicted with the position of the dye (Cy5) and the damage site (CPD) indicated. Freely diffusing Cy5-labeled DNA cannot be excited outside of the evanescent field (“OFF” state) and only becomes visible when it persists near the surface due to its binding to the surface-tethered XPD (“ON” state). EF, Cy3/Cy5 dual-bandpass emission filter; DM, dichroic mirror; M, mirror.

Figure 2

Site-specific labeling of XPD within its ARCH domain. (a) Schematic illustration of the experimental scheme for site-specific labeling of XPD within its ARCH domain. The construct for E. coli expression of the Ferroplasma acidarmanus XPD (FacXPD) contains poly histidine (6His) tag and biotin-acceptor peptide (BAP) at the N-terminus, and the aldehyde tag motif, LCTPSR, inserted in the ARCH domain after N258. The lysine residue within BAP which is biotinylated by E. coli BirA biotin ligase, and the cysteine residue within the aldehyde tag motif which is specifically converted to formyl-glycine, fGly, by formyl-glycine generating enzyme, FGE, are shown in red. After purification, Cy3-hydrazide was specifically and covalently conjugated to the aldehyde group of fGly. (b) Confirmation of purity, biotinylation, and Cy3 incorporation in XPD. SDS-PAGE of the purified aldehyde-tagged XPD (AA1-XPD) from left to right: CBB, AA1-XPD visualized by Coomassie brilliant blue staining, α-bio, antibiotin Western blot of AA1-XPD, Cy3, direct fluorescence imaging of gel of the aldehyde-tagged XPD after labeling with Cy3-hydrazide. (c) Cy3-labeled XPD retains helicase activity. Helicase activity was measured in a standard duplex separation assay where a synthetic Cy5-labeled DNA substrate (10 nM) is incubated with indicated concentrations of XPD and ATP for 15 min and the product of the reaction are separated from the substrate due to difference in their mobility on the polyacrylamide gel. The activity was estimated by measuring the percent decrease in duplex DNA mean band intensity.

Figure 3

Representative Cy3 fluorescence intensity trajectories derived from individual surface-tethered XPD protein molecules. Highly fluorescent set of substates comes from more open ARCH domain conformations (“OPEN”), and weakly fluorescent (quenched) set of substates comes from closed conformations (“CLOSED”). The half a mplitude threshold for the normalized fluorescence signal is marked by the black dashed line. A less populated intermediate state approximately at the level of the threshold is shown as “I”. This state is excluded from statistical analysis. The single irreversible photobleaching step at the end of each trajectory is indicated by the black arrow.

Figure 4

Multiexponential nature of dwell time distributions of ARCH domain conformations. (a) Normalized cumulative distributions for the ARCH domain dwell times in open (green) and closed (red) conformations. The dashed lines are best fits. Distribution for the open state is fit to double-exponential, whereas that of the closed state is fit to a triple exponential. The open–closed equilibrium favors the closed conformation. (b) Time constants of the individual exponential components of the dwell time distributions of each protein state derived from the experiments carried out at different laser powers. For each laser power, dwell time data points of each protein state were compiled into one cumulative distribution from several movies recorded on different days. Cumulative dwell time distribution of each protein state for each laser power value was constructed and fitted in the same way. Error bars (standard errors from fitting) were too small to be clearly visible.

Figure 5

Simultaneous direct real time observation of the DNA binding and dissociation and the conformational transitions of a domain of a repair protein. (a) Representative fluorescence intensity time trajectories of a dual illumination and dual detection experiment, with the green trace (Cy3) showing an individual XPD protein ARCH domain opening and closing and the red trace (Cy5) showing the association and dissociation of individual damaged DNA molecules. The Cy5-labeled DNA concentration was 150 pM. Distinct protein states (open and unbound, closed and unbound, closed and bound, and open and bound) are schematically shown above the corresponding fluorescence states. (b) Normalized cumulative distributions for the ARCH domain dwell times in “open” conformation in absence of DNA from the imaging chamber (black), only when simultaneously detected with binding events of undamaged DNA (green) and only when simultaneously detected with binding events of damaged DNA (red). The dashed lines represent the best fits to double-exponential. (c) Normalized cumulative distributions for the ARCH domain dwell times in “closed” conformation in absence of DNA from the imaging chamber (black), only when simultaneously detected with binding events of undamaged DNA (green) and only when simultaneously detected with binding events of damaged DNA (red). The dashed lines are best fits to triple-exponentials. The difference between the red and the green curves indicates that DNA damage stabilizes the closed conformation of ARCH domain.

Figure 6

Single-molecule analysis of binding of undamaged and damaged Cy5-labeled bubble DNA to Cy3-labeled XPD. (a) Representative fluorescence intensity time trajectories of a dual illumination experiment from a single XPD molecule in the presence of bubble DNA. The green trace (Cy3) showing an ARCH domain persisting in the open state for 10s of seconds and the red trace (Cy5) showing the association and dissociation of individual bubble DNA molecules. This form of simple association/dissociation binding represents the most dominant type of events. Inverse of binding on-time is the apparent dissociation rate, and the inverse of binding off-time is the apparent association rate for each specific DNA concentration. (b) Binding on-time histogram for undamaged (left) and damaged (right) bubble DNA (at 150 pM concentration) fitted to a single exponential decay. Errors shown between brackets are the standard errors from fitting the on-time distributions. Durations of individual events were collected from approximately 50 XPD molecules and compiled in the distribution for each type of DNA. Apparent dissociation rates (inverse of binding on-time) for both types of bubble DNA are almost the same. (c) Effect of DNA concentration on the apparent association rate (inverse of binding off-time). Binding off-time at each DNA concentration was calculated in the same way as described above for binding on-time. Damaged bubble DNA shows a slightly higher apparent association rate than undamaged bubble at all DNA concentrations tested.

Figure 7

Minimal model for the role of ARCH domain dynamics in a kinetically enhanced damage detection process and in the recruitment of downstream factors of the NER pathway. Damaged and undamaged DNA can bind to both open and closed states. The association rate constants obtained without specifying the conformational state during which any binding event occurs, however, is higher for damaged DNA resulting in a slightly higher affinity (Figure 6). The ARCH domain undergoes conformational transitions both in the DNA bound and free states of XPD. Time values shown above the arrows associated with opening and closing reactions are “weighted means” of the time constants of the exponentials used to fit the lifetime distributions of the conformational states (Table 1). We propose a kinetically enhanced damage detection process composed of two steps: the first level of discrimination takes place at the moment of binding undamaged or damaged substrate, whereas the second occurs at the lifetime scale of ARCH domain conformational states. DNA binding is not strictly coupled to ARCH domain motion, but DNA damage slightly shifts the conformational equilibrium toward closed state, providing a means for a “kinetic amplification” of XPD damage detection (or discriminative power). The mechanism of signaling the presence of damage relies mainly on the increased lifetime of the closed state of the ARCH domain. Thus, a slight shift in the conformational equilibrium provides a platform for assembly of the downstream factors in NER pathway.