*2.2. E*ff*ect of DA and EGCG on the Conformational Ensemble*

The SMFS measurements were repeated in the presence of either 200 μM DA or 25 μM EGCG (see Figures S3–S6 and Tables S2–S6 for more details). These concentrations were chosen to compare SMFS results with native-MS data [42]. The statistical distributions of AS conformations obtained by SMFS in the presence or absence of ligands are reported in Figure 2A. In solution, at neutral pH and without ligands, AS behaves partially as RC (62% of the molecules) and partially populates collapsed states (~30%, mainly stabilized by WI and ~8%, mainly stabilized by SI), consistent with previously reported SMFS data [14–16]. The addition of either ligand leads to a loss of the RC conformation in favor of the SI conformation, with the most pronounced effect of EGCG (drop of RC from 62% to 36%). The same conditions had been investigated by native MS, showing the presence of intermediate states (I1 and I2), together with random coil (RC) and compact (C) conformations (Figure 2B).

**Figure 2.** Species distributions as obtained by (**A**) SMFS and (**B**) native MS. The intensity-weighted average charge of the peak envelopes is reported in brackets (i.e., RC = 17.3; I1 = 13.7; I2 = 10.5; C = 8.4). Error bars in panel (A) represent the standard deviation calculated for the normal distribution. Error bars in panel (B) represent the standard deviations from three independent experiments. (**C**) RC reduction in response to ligand binding, relative to the free protein, as obtained by SMFS and native MS, considering the 1:1 protein:ligand complexes (1:1) or the cumulative MS data (all). Error bars in panel (C) represent the propagated standard deviation.

A quantitative comparison between the species distributions obtained by SMFS and native-MS data is shown in Figure 2C. An intrinsic difference between SMFS and MS concerns the discrimination between free and ligand-bound protein molecules, which is possible only by the latter technique. Thus, native-MS data in Figure 2C have been processed by two alternative ways. In one case, only signals of the 1:1 protein:ligand complexes have been considered. This procedure yields more reliable information on the conformational changes induced by ligand binding but is, at the same time, not exactly comparable to the blind molecular selection performed by SMFS. Thus, "cumulative" MS data are also shown (labeled as ESI-MS(all) in Figure 2C), derived by Gaussian fitting of the artificial CSD obtained by the summation of the species-specific CSDs corresponding to the different binding stoichiometries, including the free protein. In either way, the aggregated data for the unstructured (RC) component, represented as relative change from the reference condition of the protein in the absence of ligands, indicate a remarkable loss of the most disordered conformation induced by ligand binding, as assessed by both techniques.

### *2.3. Comparison to CD and FTIR*

For comparison with complementary spectroscopic methods, sensitive to protein secondary structure, far-UV circular dichroism (CD) and Fourier-transform infrared spectroscopy (FTIR) analyses were performed. Representative results are reported in Figure 3. It can be noted that AS spectra in the presence or absence of the ligands, acquired by either technique under the same conditions employed for SMFS experiments, are almost superimposable. Thus, bulk methods probing secondary structure do not capture the conformational changes induced by ligand binding in monomeric AS in solution.

**Figure 3.** Secondary-structure content as obtained by CD and Fourier-transform infrared spectroscopy (FTIR) techniques. (**A**) Far-UV CD spectra of 20 μM AS in PBS buffer in the absence of ligands (gray), in the presence of 200 μM DA (orange) or 80 μM epigallocatechin-3-gallate (EGCG) (purple). (**B**) Second derivatives in the Amide I region of the FTIR absorption spectra of 340 μM AS in deuterated PBS buffer in the absence of ligands (gray) and in the presence of 1 mM DA (orange) or 800 μM EGCG (purple).

#### **3. Discussion**

The results reported here provide direct evidence of the different conformers populated by AS in solution and the structural effects elicited by ligand binding, resulting in a rearrangement of

the conformational ensemble [50]. The structural heterogeneity of free AS in solution captured by SMFS is consistent with previous reports by the same approach [14,15], as well as with results from native MS [23,29,33,42], computational simulations [51,52], and chemical crosslinking [53], indicating the presence of at least three different conformational states characterized by different degrees of intramolecular interactions. Furthermore, SMFS is applied here for the first time to probe the effects of the fibrillation inhibitors DA and EGCG on AS conformational properties in solution.

Since the reliability of CSD analysis in the investigation of IDP conformational ensembles by native MS has been questioned [44], the SMFS results obtained in this work are compared to native-MS data. In analogy with SMFS, the CSD analysis of nano-ESI-MS spectra identifies, in addition to the RC component, the presence of three non-RC components, namely the intermediate species I1, I2, and the compact conformation C. Furthermore, both techniques indicate a loss of the most disordered component in response to ligand binding, resulting in the accumulation of the more structured species. Thus, not only the presence of multimodal profiles is confirmed by both techniques, but also a reorganization of the conformational ensemble in the same direction is consistently indicated in the presence of ligands.

Nonetheless, the structural intermediates detected by SMFS and native MS cannot be related in a straightforward way. These discrepancies can be due to the fact that the physical properties detected by the two techniques are different. While SMFS discriminates protein structures according to their mechanical stability under an external tension (quantified by the unfolding force), native MS is affected by structural compactness (quantified by the acquired net charge). Different compaction levels can correspond to similar unfolding force and vice-versa. Accordingly, the WI state, as detected by SMFS, is characterized by a number of variable peaks ranging from 1 to 3 different species, which could be compatible with different AS compaction states. Furthermore, the SMFS instrumental noise, related to the minimum measurable force (around 20pN) limits the minimal detectable unfolding force, below which the less stable AS compact states are counted as RC molecules.

It should also be noted that the conformations with lower unfolding force, as detected by SMFS, could include some components with higher charge-state detected by ESI-MS. This hypothesis can be verified by comparing the two techniques in terms of the response of the RC component to the binding of the ligand. Indeed, upon binding of either ligand there is a compatible trend of loss in such a component, as observed by both techniques, in favor of more compact structures (native MS) or stronger interactions (SMFS). Therefore, interpreting the low-charge components of CSDs as collapsed and partially structured conformational states leads to compatible pictures delivered by SMFS and native MS. Both techniques reveal the presence of partially structured conformers, thus suggesting that the bimodal or multimodal CSDs detected by native MS do not simply reflect artefacts of the ESI mechanism. It is worth pointing out that the conditions employed in this work do not lead neither to AS oxidation (Figure S7) nor to AS oligomerization, which requires incubation at 37 ◦C, shaking and higher protein concentrations [54], as also indicated by the lack of higher-order aggregates in native-MS spectra [42].

It cannot be ruled out that different ionization and/or transmission efficiency of compact and extended protein ions in native MS might lead to distortions of the apparent molecular ensemble, adding to the difficulties of direct comparison with SMFS data. Indeed, it has been suggested that folded and unfolded molecules could undergo different ESI mechanisms, resulting in different signal yields [55]. However, this effect seems to be protein-specific, since quantitative agreement with solution methods has been observed describing, for instance, the pH-dependent unfolding transition of cytochrome *c* [56]. The underlying mechanism has been identified in the different hydropathy of the exposed regions of normally folded proteins in different conformational states, which could affect their surface activity inside ESI droplets [55]. Such an effect is expected to be much more modest for IDPs, which lack a structured hydrophobic core and whose collapsed conformations are mostly promoted by electrostatic interactions [57]. More systematic, quantitative comparison between native MS and solution methods will be required to further elucidate this point. This first comparative study between

a single-molecule technique and CSD analysis by native MS supports the feasibility of combined approaches to describe IDP molecular ensembles.

Based on this study, it seems safe to conclude that SMFS and structural interpretation of CSDs consistently indicate the simultaneous presence of collapsed and partially structured conformers of AS monomer in solution and, most importantly, reveal induced-folding transitions elicited by ligand binding. Furthermore, this study shows that single-molecule protein unfolding can capture changes in AS conformational landscape, induced by variable solution conditions, with remarkable sensitivity and reproducibility. These results indicate that the conformational ensemble depicted by two orthogonal biophysical principles is heterogeneous and reshaped in the same direction by ligand binding.

Another implication of this study is that the AS conformational transitions detected by SMFS under these conditions should not be interpreted in terms of secondary structure formation [14,15]. Indeed, the measured WI and SI components cannot be simply seen as the distinct contributions of van-der-Waals interactions and ordered secondary structure, respectively. In fact, an increase of almost 30% in the SI component, as observed here, would be detected by CD and FTIR spectroscopies, if ascribable to secondary structure. It is conceivable that the AS conformational components detected by SMFS under these conditions differ by contact order, type, and number of interactions, within a picture of similar secondary-structure content. Hence, a new structural interpretation of SMFS data is proposed, in particular for the SI population, differing from the one reported in the literature [14,15], where the SI component was directly associated with the presence of secondary structure.

This comparison points out that the ion-sorting mechanism inherent to MS analyses makes the MS methods more comparable to single-molecules approaches, rather than to bulk spectroscopic techniques, and underscores the importance of multi-technological approach to ensemble characterization. Nevertheless, the WI population detected by SMFS and the intermediate species (I1 and I2) detected by native MS do not necessarily coincide. Actually, two intermediates are detected by MS and only one by SMFS and the WI species found by SMFS does not respond to ligands, while the MS-detected intermediates do. These results indicate that both techniques capture the decrease in structural disorder induced by the ligands, but they describe the partially structured species of the conformational ensemble in different ways. In particular, it seems that the collapsed and partially structured species detected by MS contribute cumulatively to the SI component by SMFS, while the WI component by SMFS does not find correspondence in the MS spectra. These interactions could be too weak to survive the ionization/desolvation step.
