Dynamic DNA binding, junction recognition and G4 melting activity underlie the telomeric and genome-wide roles of human CST

Abstract

Human CST (CTC1-STN1-TEN1) is a ssDNA-binding complex that helps resolve replication problems both at telomeres and genome-wide. CST resembles Replication Protein A (RPA) in that the two complexes harbor comparable arrays of OB-folds and have structurally similar small subunits. However, the overall architecture and functions of CST and RPA are distinct. Currently, the mechanism underlying CST action at diverse replication issues remains unclear. To clarify CST mechanism, we examined the capacity of CST to bind and resolve DNA structures found at sites of CST activity. We show that CST binds preferentially to ss-dsDNA junctions, an activity that can explain the incremental nature of telomeric C-strand synthesis following telomerase action. We also show that CST unfolds G-quadruplex structures, thus providing a mechanism for CST to facilitate replication through telomeres and other GC-rich regions. Finally, smFRET analysis indicates that CST binding to ssDNA is dynamic with CST complexes undergoing concentration-dependent self-displacement. These findings support an RPA-based model where dissociation and re-association of individual OB-folds allow CST to mediate loading and unloading of partner proteins to facilitate various aspects of telomere replication and genome-wide resolution of replication stress.

Figure 1.

CST binds to ss-dsDNA junctions without melting the DNA duplex. (A) Silver stained polyacrylamide gel (15% on the bottom, 12% on top) showing co-purified CST subunits. (B–D) EMSAs showing CST binding to various substrates. Reactions contained the indicated amounts of CST and 0.1 nM DNA. (B) CST binding to junction substrates with an 18 nt 3′ overhang. Substrates were generated using fold-back oligonucleotides (fb) with: (i) Telomeric sequence at junction, (ii) mixed telomeric and non-telomeric sequence or (iii) non-telomeric sequence. Black lines represent non-telomeric DNA, grey lines indicate telomeric sequence. (iv) Non-telomeric fold-back substrate lacking an overhang. (C) EMSAs showing length of ssDNA substrate needed for CST binding. Non-telomeric substrates were 18, 22, 26 or 32 nt, telomeric substrate was 18 nt. The 18 nt non-telomeric/telomeric substrates had the same sequence as 3′ overhangs of the corresponding fold-back substrates. (D) EMSAs showing CST binding to junction substrates with 10 nt 3′ overhang and 10 or 18 nt 5′ overhangs. (E) Quantification of CST affinity for junction substrates versus ssDNA. Data for binding isotherms to determine apparent dissociation constants (K_d app) were obtained by EMSA and fit to a one site specific binding model. Mean ± SEM, n = 3 independent experiments each with a different protein preparation. ND, K_d not determined as binding undetectable in EMSA.

Figure 2.

CST binds to junction substrates without melting the DNA duplex. (A) Cartoon of strand-melting assay with junction substrate formed by annealing 32 and 22 nt non-telomeric oligonucleotides. Addition of CST to junction substrate could lead to (i) stable binding to ds/ssDNA junction in which case the strands of the DNA substrate would remain annealed or (ii) transient binding to the 10 nt overhang with subsequent unwinding of the adjacent duplex to give separation of the component 32 and 22 nt oligonucleotides. (B) EMSAs showing CST binding to the junction substrate formed from the 32 and 22 nt oligonucleotides, the 32 nt oligonucleotide alone or a fold-back oligonucleotide with the equivalent 22 nt duplex and 10 nt overhang. (C) Native acrylamide gel showing lack of strand-melting after CST binding to junction substrate. Expected positions of partial duplex and ssDNA are shown to the left. Boiled: samples were boiled to melt DNA just prior to gel loading. * indicates ³²P-label.

Figure 3.

Single molecule FRET showing CST binds to ss-dsDNA junctions. (A) Cartoon showing design of the non-telomeric junction substrate and anticipated FRET signals in the presence or absence of CST. In the absence of CST the flexibility of the 18 nt ssDNA will bring the Cy3 donor and Cy5 acceptor into close proximity. If CST binds without melting the anchoring DNA duplex, the high FRET signal will be lost but emission from the Cy3 donor (green) will be retained. If CST melts the DNA duplex, both the high FRET and the Cy3 donor signal will be lost. (B) FRET histograms generated from FRET measurements of >4000 individual molecules. Top: DNA alone, bottom: DNA + 2 nM CST (C) Representative smFRET real-time trace showing change in individual Cy3 and Cy5 signals (top) and FRET (bottom) with time.

Figure 4.

CST binds G4 DNA. (A) Nomenclature, sequence and expected structures of the G4-forming oligonucleotides used in EMSAs. Substrates had 3′ ssDNA (₃G4), 3′ or 5′ ssDNA (G5 and ₃G4₆) or no 3′ or 5′ ssDNA (G4). (B) Circular dichroism measurements on G4-forming oligonucleotides to confirm folding into anti-parallel G-quartet. Measurements were in 150 mM NaCl. (C) EMSAs showing CST can bind to telomeric DNA oligonucleotides that form G4 structures. Binding reactions contained the indicated concentrations of CST and 0.1 nM ³²P-labeled DNA in 150 mM NaCl, 3 mM MgCl₂.

Figure 5.

CST unfolds G4 DNA. (A) Schematic of the G4 smFRET substrates showing expected FRET signals with/without G4 unfolding. (B–D) Representative FRET histograms showing CST can bind and unfold G4 DNA. Top, DNA alone; bottom, DNA +2 nM CST. CST was added for 10 min, then excess protein was washed out prior to FRET measurement. (B and C) Unfolding of ₃G4 and ₃G4₆ in 150 mM NaCl + 3 mM MgCl₂. (D) Unfolding of ₃G4 in 100 mM KCl + 3 mM MgCl₂. (E) Representative smFRET real-time trace showing a decrease in FRET after CST binding. Top, Cy3 and Cy5 signals showing complementary transition upon CST binding. Bottom, FRET signal. Measurement was performed in the presence of 2 nM CST (no washout).

Figure 6.

Dissociation of CST from DNA is concentration dependent. (A and B) Representative smFRET real-time traces showing CST dissociation from the non-telomeric junction substrate with 18 nt overhang (A) or the ₃G4 G-quadruplex forming substrate (B). 2 nM CST was added to the slide in the imaging buffer and the recording was initiated soon afterwards. T = 0 indicates the start of recording. (C) Fraction of traces showing one or more dissociation events from non-telomeric junction substrate (pink) or ₃G4 substrate (blue) after CST removal (0 nM), in the continued presence of 2 nM CST, or after addition of 5 nM CST. N = 3 independent experiments ± S.E.M., 260–560 real-time traces were analyzed for each protein concentration with the junction substrate and >1000 with ₃G4. (D) Dwell times for CST binding to ₃G4 substrate before protein dissociation. Traces analyzed were as in C, a minimum of 179 dissociation events were analyzed for each CST concentration (B). (E) Model illustrating how unbound CST could cause facilitated displacement of bound CST. The CST complex is shown with a 1:1:1 stoichiometry of CTC1, STN1 and TEN1. Individual OB folds (known for STN1 and TEN1, predicted for CTC1 (18)) are represented as oblongs. Note, the FRET data cannot distinguish between complete CST dissociation from the DNA versus some residual CST binding (e.g. to the ss-dsDNA junction or G4 structure) that exposes sufficient ssDNA to reform the high FRET state.

Figure 7.

Models for CST mechanism of action. (A) Model illustrating how junction recognition and dynamic binding could lead to Pol α loading on the telomeric G-strand overhang to achieve C-strand fill-in. (B) Model for how CST binding could relieve genome-wide replication stress by promoting G4 unfolding. Oblongs represent individual OB folds in CTC1, STN1 and TEN1