A Sequential Assembly Mechanism for Stable Cdc13 Dimerization on Telomeric DNA

Reviewer #1 (Public review):

Summary:

This manuscript investigates how the Saccharomyces cerevisiae telomere-binding protein Cdc13 assembles on a 12-nucleotide single-stranded telomeric DNA substrate. Using complementary smFRET and CoSMoS measurements, together with dimerization- and DNA-binding-defective mutants, mass photometry, photobleaching analysis, and kinetic modeling, the authors assign state II to a DNA-bound Cdc13 monomer and state III to a stable complex containing two Cdc13 molecules. They propose that the stable DNA-bound dimer forms predominantly through sequential recruitment of two monomers, while direct binding of a preformed dimer represents a less frequent pathway. This work addresses an important mechanistic question in telomere biology because the pathway of Cdc13 assembly may influence telomere recognition, end protection, and recruitment of telomere-maintenance factors.

Strengths:

Overall, the manuscript is well written, and the combination of two complementary single-molecule approaches is a strength. The central model is interesting and potentially important.

Weaknesses:

Several issues require clarification or additional analysis.

Major comments:

(1) The mass photometry results are central to the mechanistic model and should be presented more prominently.

The conclusion that Cdc13 binds DNA predominantly through sequential monomer recruitment depends strongly on the oligomeric state of Cdc13 at the concentrations used in the single-molecule experiments. The mass photometry results currently provide the principal direct evidence that Cdc13 is predominantly monomeric at 5 nM but exists as a mixture of monomers and dimers at 20 nM. These results should therefore be included in a main figure rather than only in the Supplementary Information. The authors should also provide a complete description of the mass photometry in the Methods section.

Related to lines 191-192, the authors should discuss the estimated cellular or nuclear concentration and abundance of Cdc13 and compare these values with the experimental concentrations at which states II and III are populated. Because the relevant quantity may be the effective local concentration at a telomere rather than the average nuclear concentration, this distinction should also be acknowledged. Such a discussion is needed to establish under what physiological conditions sequential monomer loading versus binding of a preassembled dimer would be expected.

(2) 75% labeling efficiency must be clearly defined and incorporated into both the stoichiometric and kinetic analyses.

In line 251, the authors state that the labeling efficiency of DY-649P1-Cdc13 is 75%, but it is not clear how this value was measured. The authors should state whether 75% refers to the efficiency of the sortase reaction, the fraction of labeled molecules in the final purified preparation, or a value inferred from the plateau in Figure 3B. If it was inferred from the binding plateau, the plateau below 100% could also arise from inactive or inaccessible DNA molecules, incomplete colocalization detection, inactive protein, or an effect of the fluorophore on binding. An independent measurement, such as absorbance-based determination of the dye-to-protein ratio, quantitative gel analysis, or intact-mass analysis, would be preferable.

Incomplete labeling has direct consequences for the interpretation of Figure 3. With a labeling probability of 0.75, a true Cdc13 dimer would contain zero, one, or two fluorophores. Thus, even among detectable dimers, 40% would appear as one-step photobleaching events. A one-step event therefore cannot automatically be equated with a monomer without correcting for labeling efficiency. The authors should quantitatively account for incomplete labeling when inferring the relative monomer and dimer populations from the photobleaching data.

The same issue is even more important for the CoSMoS kinetic model. The authors should incorporate labeling efficiency into the observation model, or at minimum perform simulations or a sensitivity analysis demonstrating that the inferred transition rates and state assignments are robust to 75% labeling.

(3) Apparent state I→III transitions do not by themselves demonstrate direct binding of a preformed Cdc13 dimer.

At lines 326-330, the authors interpret state I→III transitions as direct binding of a solution dimer and conclude that sequential binding is approximately eightfold faster than direct dimer binding. This interpretation is not yet sufficiently established. An observed I→III transition demonstrates only that no state II intermediate was resolved; it does not distinguish true binding of a preformed dimer from sequential binding in which the state II lifetime is shorter than the temporal resolution of the experiment.

This concern is particularly important because the CoSMoS experiments were conducted at only 0.3-1.25 nM Cdc13, whereas mass photometry indicates that Cdc13 is predominantly monomeric even at 5 nM. In addition, the smFRET signals were averaged over a sliding window of ten 50-ms frames, which could obscure short-lived intermediate states.

The authors should show representative raw traces containing apparent I→III transitions in a supplementary figure and quantify the shortest state II dwell time that could be detected under the acquisition, smoothing, and HMM procedures used.

(4) The interpretation of the WT-Cdc13/Cdc13^R635C mixture requires further clarification.

For Figures 2G-H and lines 209-218, the reduced state III population in the mixture of 2.5 nM WT Cdc13 and 2.5 nM Cdc1^R635C is interpreted as evidence for a solution monomer-dimer equilibrium and formation of a nonfunctional WT-mutant heterodimer. However, at least two nonexclusive explanations should be considered:

a) Formation of WT-mutant heterodimers in solution could reduce the concentration of free WT monomers and WT homodimers available to form state III.
b) A WT-mutant heterodimer, or recruitment of Cdc13^R635C to a DNA-bound WT molecule through protein-protein interactions, could produce a DNA-bound complex that cannot adopt the state III conformation because only one subunit has an intact DNA-binding interface.

The authors should discuss these possibilities explicitly and clarify expected FRET states.

For direct visual comparison, Figure 2G should include the FRET histograms for 2.5 nM WT Cdc13 alone and 5 nM WT Cdc13 alone, in addition to the WT-mutant mixture. The concentrations of both the initially loaded WT Cdc13 and the WT or mutant protein added during the chase experiment in Figure 2I should also be stated in the main text and figure legend.

(5) The physical basis of the different FRET values for states II and III should be explained earlier and more carefully.

The assignment of state II and state III to one and two bound Cdc13 molecules is supported by the combined smFRET and CoSMoS results. However, the manuscript should explain earlier why the addition of a second Cdc13 molecule is expected to produce a further decrease in FRET. Because the fluorophores are attached to the DNA, the different FRET values imply a change in the distance, orientation, or local photophysical environment of the DNA-linked dyes when the second Cdc13 binds. CoSMoS establishes a change in protein stoichiometry, but it does not by itself establish that the DNA has undergone a particular conformational change.

The discussion at lines 404-423 suggests that the second Cdc13 induces a rearrangement of the first Cdc13-DNA complex. This is a reasonable hypothesis, but it should be presented as an inference rather than as a demonstrated DNA conformational transition. References 44-46 describe different RPA binding modes and rearrangements of protein-DNA contacts; they do not directly demonstrate the specific DNA conformational change proposed here. The authors should either provide more direct support or revise the discussion accordingly. The distance estimates should also be described cautiously because they assume that dye orientation and photophysical properties are unchanged between states.

The rationale for the internally positioned Cy3 constructs in lines 147-149 should also be explained more clearly. Why does it demonstrate that Cdc13 cannot bind duplex DNA?

Reviewer #2 (Public review):

Summary:

The manuscript presents an interesting and potentially important single-molecule study of Cdc13 assembly on telomeric ssDNA. The experimental observations are intriguing, particularly the identification of distinct FRET states associated with different Cdc13 occupancies. However, I have substantial concerns about whether the current data support the central mechanistic conclusion as strongly as the authors claim. In particular, the manuscript does not yet clearly distinguish between the observation that two Cdc13 molecules are associated with DNA and the stronger mechanistic claim that Cdc13 loads sequentially as monomers and subsequently dimerizes on DNA.

Major concerns

(1) The central mechanistic conclusion is stronger than the evidence

The manuscript's principal model is that Cdc13 proceeds through the pathway monomer → DNA-bound monomer → recruitment of a second monomer → stable DNA-bound dimer. However, the experiments establish this sequence only indirectly.

The authors show that FRET state II is associated with one Cdc13 molecule. FRET state III is associated with two Cdc13 molecules. Cdc13-DM produces state II but not state III. WT Cdc13 can undergo I→II→III transitions. Direct I→III transitions also occur.

These observations are consistent with sequential binding, but they do not uniquely demonstrate that the second Cdc13 molecule first binds as a monomer and subsequently undergoes dimerization on DNA. In particular, the direct I→III events indicate that a preassembled dimer or another cooperative pathway can also contribute.

I therefore recommend substantially tempering statements such as "Cdc13 initially loads onto telomeres as a monomer." A more defensible formulation would be: "The data support a kinetically favored sequential pathway involving an initial Cdc13 binding event followed by recruitment of a second Cdc13 molecule." The authors could still present sequential loading as the preferred model, but the language should clearly distinguish a kinetically supported pathway from a uniquely established molecular mechanism.

(2) The assignment of FRET states II and III to exactly one and two Cdc13 molecules requires stronger validation

The interpretation of states II and III as one- and two-Cdc13 states is central to the entire mechanistic model. Because labeling efficiency, incomplete labeling, photophysics, and heterogeneous molecular populations can all affect the observed distributions, the authors should provide a quantitative probabilistic model. Specifically, the authors should calculate the expected distribution of one- and two-labeled-Cdc13 species given the experimentally determined labeling efficiency and compare these expectations with the observed FRET-state populations. This analysis would provide an important independent validation of the proposed stoichiometric assignments.

(3) The TG12 substrate raises an important stoichiometric and conformational question

The manuscript argues that a single Cdc13 molecule binds approximately 11 nt, while two Cdc13 molecules can bind a 12-nt ssDNA substrate. This raises an immediate mechanistic question: if one Cdc13 occupies approximately 11 nt, how can two Cdc13 molecules simultaneously associate with only 12 nt of ssDNA? This issue should be addressed experimentally rather than only structurally or schematically. A particularly informative experiment would be to systematically vary ssDNA length. The probability and kinetics of state III formation could then be quantified as a function of substrate length. Such an experiment would determine whether formation of the two-Cdc13 state genuinely requires additional DNA and whether the two proteins occupy overlapping or distinct regions of the substrate.

(4) The "salt-resistant" interpretation is overstated

The manuscript repeatedly describes state III as "salt-resistant" and uses this observation to support the existence of a highly stable physiological complex. However, the reported experiment examines only a relatively modest range of NaCl concentrations (50, 75, and 100 mM).

I recommend either expanding the salt-dependence analysis substantially or using more quantitative language. For example, the authors could report the fraction and lifetime of state III as a function of salt concentration and define explicitly what they mean by "salt-resistant." The current data support persistence under the tested conditions, but they do not by themselves establish exceptional physiological stability.

(5) The mass-photometry experiment does not establish the physiological solution equilibrium

The mass-photometry data are useful for demonstrating that Cdc13 can exist in monomeric and dimeric forms, but the current experiment does not establish the equilibrium between these species under physiologically relevant conditions.

A concentration series would be valuable. The observed monomer/dimer populations should be fit to an explicit equilibrium model to obtain an apparent dimerization constant and assess how strongly the equilibrium depends on Cdc13 concentration. This would also help connect the solution behavior to the single-molecule observations and determine whether the observed DNA-bound dimer could plausibly arise from a pre-existing solution dimer.

(6) The WT + R635C mixing experiment is overinterpreted

The interpretation of the WT + R635C experiment is currently complicated and somewhat speculative. The experiment is potentially informative, but the conclusions appear stronger than what can be directly inferred from the data. The authors should clearly distinguish between the observations directly supported by the mixing experiment and the mechanistic interpretation proposed from them. In particular, the experiment does not necessarily establish the precise sequence of DNA binding and dimerization events.

(7) The requirement for DNA-binding activity in both Cdc13 molecules is not fully established

The manuscript concludes that both Cdc13 molecules must possess DNA-binding activity. However, the R635C experiment does not clearly distinguish between:
two independently DNA-bound Cdc13 molecules; and one Cdc13 molecule directly bound to DNA plus a second molecule whose DNA-binding surface is required for allosteric stabilization of the dimer.

This distinction is mechanistically important. Additional experiments using DNA-binding-defective mutants in defined heterodimeric configurations would help determine whether both molecules directly contact DNA or whether DNA binding by one molecule promotes recruitment/stabilization of the second through protein-protein interactions.

(8) Stronger controls are needed for FRET-state assignment

The FRET states are treated as discrete molecular states, but alternative explanations for heterogeneous FRET populations should be considered more explicitly.

Important controls would include: concentration-dependent FRET measurements in the absence of DNA binding; fluorescence controls to determine whether the observed states could arise from dye-protein interactions; Cdc13 mutants with altered DNA-binding specificity; alternative dye positions; demonstration that the major FRET states are reproduced with independent labeling configurations. The duplex-positioned Cy3 controls, which show little FRET change, are useful. However, they do not completely exclude the possibility that protein-induced changes in DNA conformation contribute to the observed FRET states. Independent labeling geometries would substantially strengthen the assignment.

(9) The physiological relevance of the 12-nt substrate requires better justification

The authors use TG12 as their primary substrate and state that telomeres contain approximately 12-14 nt of ssDNA during most of the cell cycle. This rationale requires greater biological context. Telomere length and the extent of the exposed G-rich strand are dynamic and heterogeneous, and Cdc13 has established functions throughout telomere replication. The authors should explain more carefully why TG12 is biologically representative and how the proposed mechanism is expected to behave on substantially longer telomeric substrates. The TG25 experiment is useful, but at present it functions primarily as a stoichiometric observation. A systematic substrate-length analysis, as suggested above, would turn this observation into a mechanistic test.

(10) The relationship to existing structural studies requires deeper discussion

The manuscript presents sequential loading as a novel mechanism, but existing structural and biochemical studies of Cdc13 dimerization and Cdc13-DNA architecture are essential for interpreting these observations. The authors should explicitly reconcile their proposed model with the existing structural literature. In particular, they should address:

Does the known Cdc13 dimerization interface permit simultaneous DNA binding by both subunits?

Is the dimerization interface compatible with the proposed DNA-bound state II?

Could DNA binding alter the dimerization interface?

Are the two Cdc13 molecules predicted to bind overlapping or distinct portions of the telomeric sequence?

Can the structural models accommodate the apparent stoichiometry on a TG12 substrate?

Without this reconciliation, the proposed sequential-loading mechanism remains somewhat disconnected from the established structural framework.

Other important issues:

(11) The Kd comparisons are confusing

The manuscript should include a table summarizing the different Kd values and explicitly explaining why they differ. In particular, describing 3.7 nM as the Kd for the first monomeric binding step while reporting an apparent Kd of 1.2 nM requires careful kinetic and statistical justification. The authors should distinguish clearly among microscopic Kd values, apparent Kd values, and parameters inferred from kinetic models.

(12) The direct-dimer pathway deserves greater attention

The occurrence of direct I→III transitions is mechanistically important and should not be treated primarily as an exception to the sequential pathway. A more balanced conclusion would be:

"Both pathways contribute to formation of the final Cdc13-DNA complex, with the sequential pathway being kinetically favored under the experimental conditions." This interpretation appears better aligned with the data and would still constitute a strong mechanistic conclusion.

(13) The "kinetic proofreading" interpretation is currently speculative

The Discussion proposes that the first Cdc13 monomer provides a kinetic proofreading step. This is an interesting hypothesis, but it is not directly demonstrated by the current experiments.

I recommend changing this to language such as "a kinetic proofreading-like mechanism may be possible" unless the authors can provide direct evidence that the first binding event selectively promotes productive complex formation or rejects nonproductive substrates.

(14) Biological-function claims should be clearly separated from the in vitro findings

The manuscript frequently connects the stable state III complex with telomere protection, telomere length regulation, CST formation, Est1 recruitment, Pol α recruitment, and prevention of DNA degradation. None of these functions are directly tested in the present study.

The experiments establish a biochemical/single-molecule mechanism in vitro. They do not establish that state III is the functional protective species in vivo.

This distinction should therefore be maintained throughout the Abstract, Discussion, Key Points, and concluding statements. The authors can appropriately discuss these possibilities as
implications or hypotheses, but should avoid presenting them as demonstrated functions of state III.