Mass Spectrometric Characterization of Native Insulin Hexamers and Non‐Native Heptamers: Formation and Stability Across Insulin Analogs

ABSTRACT

Insulin aggregation and oligomerization present significant challenges in both therapeutic formulation and fundamental studies of amyloid formation, particularly due to the transient and heterogeneous nature of early‐stage oligomers. Here, we employ ion mobility–mass spectrometry (IM–MS) to characterize the oligomeric distributions of human insulin and two clinically relevant analogs, aspart (rapid‐acting) and glargine (long‐acting), in excipient‐containing solutions designed to stabilize native forms of insulin and under aggregation‐inducing conditions. Comparison of insulin analogs revealed distinct aggregation propensities that correlate with each analog's therapeutic design. In general, stable zinc‐coordinated hexamers formed most readily in the presence of excipients, and various oligomers formed under aggregation‐inducing conditions, with a noticeable prevalence of heptamer formation. We further investigated the structure and stability of native hexamers and non‐native heptamers using collision‐induced dissociation and collision‐induced unfolding experiments. Although the expected single‐monomer ejection was the main dissociation pathway for both species, zinc‐coordinated hexamers also dissociated into two zinc‐adducted trimers, whereas heptamers dissociated into dimer/pentamer and trimer/tetramer pairs. Gas‐phase unfolding indicated conservation of subunit tertiary structure in hexamers and no distinct folded structures in heptamer subunits. Overall, this work demonstrates the utility of IM–MS as a rapid, high‐resolution platform for probing insulin aggregation pathways and evaluating current and future insulin analog formulations.

1. Introduction

Insulin is an essential peptide hormone that regulates blood glucose homeostasis and enables cellular glucose utilization [1, 2, 3]. In its native monomeric form, insulin is a 5.8‐kDa predominantly alpha‐helical protein composed of an A chain of 21 amino acids and a B chain of 30 amino acids, linked by two interchain disulfide bonds and one intrachain disulfide bond within the A chain [1, 2, 3, 4, 5]. Despite this defined structure, insulin is highly sensitive to environmental stressors and can undergo denaturation and self‐association under conditions such as elevated temperature, acidic pH, hydrophobic interfaces, and increased ionic strength [5, 6, 7]. These perturbations promote partial unfolding, nucleation, and subsequent elongation into amyloid fibrils [2, 4]. At the same time, insulin also undergoes reversible self‐association into multiple oligomeric states, including monomers, dimers, and hexamers [1, 3, 8]. Under nondenaturing conditions, zinc plays a central stabilizing role by promoting formation of the native hexamer, which consists of three insulin dimers coordinated by two Zn²⁺ ions through residues B10 histidine and B13 glutamic acid [3, 9]. In vivo, this zinc‐coordinated hexamer serves as the principal storage form of insulin and dissociates through dimeric intermediates to release the active monomer [2, 10].

Although insulin aggregation is not typically problematic under physiological storage conditions, it remains a major challenge in pharmaceutical formulations during manufacturing, purification, storage, and transport [2, 5, 11]. This challenge is further complicated by the diversity of clinically relevant insulin products, including human insulin and engineered analogs such as aspart and glargine, which differ in sequence, oligomerization behavior, and pharmacokinetic profile [12, 13]. Formulation additives such as Zn²⁺ and phenolic excipients are widely used to stabilize native insulin assemblies and suppress fibrillation; however, these same additives can alter absorption and biofunctionality through formulation‐dependent interactions [1, 2, 3, 8]. As a result, insulin stability cannot be understood solely in terms of the protein sequence itself but must instead be evaluated within the broader context of analog identity, oligomeric state, and excipient composition. This complexity creates a critical need for rapid analytical approaches capable of characterizing and comparing multiple insulin formulations efficiently, particularly across diverse analog‐excipient combinations relevant to formulation design and stability optimization [11, 14, 15].

Addressing insulin aggregation requires analytical methods capable of resolving heterogeneous and rapidly evolving oligomeric ensembles. In particular, the dynamic nature of multicomponent insulin samples complicates the identification, isolation, and characterization of individual oligomeric species, as well as efforts to define the kinetics and equilibria that govern their formation [11, 14, 15, 16]. Prior studies suggest that soluble oligomers arise during the lag phase of fibrillation and subsequently progress to insoluble fibrils, making these early intermediates especially important mechanistic targets [17, 18, 19]. However, this transient regime is often difficult to access using conventional structural methods such as electron microscopy, X‐ray crystallography, and nuclear magnetic resonance spectroscopy, which are either poorly suited to short‐lived intermediates or limited in throughput [20, 21]. Mass spectrometry (MS) offers a powerful alternative because it combines speed, sensitivity, and molecular specificity, enabling direct interrogation of semistable and transient oligomeric states [15, 22]. In particular, electrospray ionization (ESI) preserves noncovalent assemblies sufficiently to detect soluble insulin oligomers, whereas ion mobility (IM) provides an additional dimension of separation for species with overlapping mass‐to‐charge ratios and enables interrogation of their conformational landscapes [23, 24].

Whereas previous work investigating soluble insulin oligomers via IM–MS mainly identified the distribution of various oligomeric states of human insulin and some analogs [25, 26, 27], in this work, we compare human insulin, the rapid‐acting analog aspart, and the long‐acting analog glargine under aggregation‐promoting and zinc‐stabilizing conditions. In addition to characterizing oligomeric distributions, we investigate the structure and stability of prevalent species, including zinc‐associated hexamers and the less commonly reported heptamer. Our results support more recent findings that suggest that monomer addition and loss make substantial contributions to insulin oligomerization dynamics, in contrast to other models that assume other insulin assembly and disassembly pathways [28]. Finally, by employing collision‐induced unfolding (CIU), we demonstrate the ability of IM–MS–based methods to characterize partial unfolding in early oligomeric states thought to precede fibrillation [29, 30]. Collectively, these results demonstrate the utility of IM–MS for rapid characterization of insulin oligomerization and provide a comparative framework for assessing the effects of analog identity and excipient environment across diverse formulations.

2. Experimental

2.1. Chemicals and Reagents

Human insulin (5.8 kDa) (Adooq Bioscience, CA, USA), insulin aspart (5.8 kDa), and insulin glargine (6.1 kDa) (Sigma‐Aldrich, MO, USA) were obtained in pure, powdered form without any formulation excipients and stored at −20°C until use. Ammonium acetate, zinc acetate, ammonium chloride, glacial acetic acid, and phenol were purchased from Sigma‐Aldrich (St. Louis, MO, USA). HPLC‐grade water was purchased from Honeywell/Burdick & Jackson (Charlotte, NC, USA).

2.2. Sample Preparation

To prepare excipient‐containing solutions, 2.2 mg of insulin was initially dissolved in 1 mL of 0.1% acetic acid. A solution of 300‐mM ammonium acetate, 5‐mM phenol, 10‐μM ammonium chloride, and 20‐μM zinc acetate was prepared and adjusted to either pH 7.4 or pH 3.8. The dissolved human and aspart were each diluted to 5 μM with the pH 7.4 solution; the dissolved glargine was diluted to 5 μM with the pH 3.8 solution.

To prepare aggregation‐inducing solutions, 1.4 mg of insulin was initially dissolved in 1 mL of 1% acetic acid. A solution of 300‐mM ammonium acetate and 10‐μM ammonium chloride was prepared and adjusted to either pH 5.0 or pH 6.8. The dissolved human insulin and aspart were each diluted to 5 μM using the pH 5.0 solution; the dissolved glargine was diluted to 5 μM with the pH 6.8 solution. The choice of pH was based on matching the isoelectric point (pI) of each insulin analog to encourage oligomerization.

Solutions were initially prepared at 5 μM to minimize ESI droplet‐induced aggregation, which can obfuscate species accurately captured from solution. To evaluate concentration dependence, additional solutions were prepared at 12 times excess concentrations of insulin, ammonium chloride, and zinc acetate (if applicable). The 60‐μM samples were also used in targeted dissociation and unfolding experiments to enhance the signal of targeted species identified in the original 5‐μM insulin samples.

2.3. MS

Experiments were conducted using a Waters SELECT SERIES Cyclic IMS instrument (Waters Corporation, Wilmslow, UK) that combines cyclic IM with time‐of‐flight mass analysis. The insulin samples were ionized via static spray nano‐ESI. Emitters were fabricated from borosilicate capillaries (1‐mm outer diameter, 0.78‐mm inner diameter) using a micropipette puller (P‐97, Sutter Instruments). A voltage of 0.8 kV was applied to the sample solution using a platinum wire, and an inlet source temperature of 100°C facilitated droplet evaporation. Improved transmission of higher m/z species through the source region, while minimizing in‐source activation and dissociation, was accomplished by applying a 50‐V difference to the stepwave source device (termed “cone voltage” by the control software). Additional transmission voltages were adjusted to minimize ion activation while maintaining transmission. The pre‐IMS trap injection voltage and post‐IMS transfer injection voltage were 4 V, and N₂ collision gas was flowed into both regions at a rate of 3.5 mL/min. The cyclic TW velocity was 375 m/s with a TW static height of 18 V.

For dissociation studies, beam‐type collision‐induced dissociation (CID) was applied to isolated species in the pre‐IMS stacked‐ring electrode trap or the post‐IMS segmented quadrupole transfer cell (both pressurized with a nitrogen flow of 3.5 mL/min) with increased kinetic energy. CIU was employed to measure conformational changes in species as a function of collision energy. Ion injection and transmission settings were identical to CID experiments, except that a lower 10 V “cone voltage” was used to minimize in‐source activation. Collected spectra were averaged for 5 min with a m/z range of 500–8000. The cIMS has a mobility spectral resolution of 200 data points or bins. Data presented here use a mobility axis of uncalibrated drift bins, where each bin is equivalent to 0.5 ms. Since all experiments used the same cIMS settings, the measured CIU plots are directly comparable. All data were analyzed and visualized using MassLynx v4.2, Driftscope v3.0 (Waters Corporation, USA), and CIUSuite 2. Data presented include typical mass spectra (m/z vs. intensity) and heatmaps (m/z vs. arrival time vs. intensity).

3. Results

3.1. Oligomer Distributions of Insulin Analogs

Because insulin aspart is a fast‐acting analog, the oligomeric distribution as captured by nano‐ESI would be expected to be dominated by monomers, whereas insulin glargine, being a long‐acting analog, would likely produce higher order oligomers. To thoroughly characterize oligomer distributions, we examined human insulin, aspart, and glargine mass spectra produced from two contrasting solution conditions. The first condition contained Zn²⁺ and phenol excipients to promote the formation of stabilized zinc‐coordinated hexamers, whereas the second was formulated near the isoelectric point (pI) of each analog to favor self‐association and aggregation.

Because of nano‐ESI's ability to generate oligomer ions across multiple charge states, species of different oligomeric size can appear at identical m/z values. For example, a 4+ insulin dimer overlaps in m/z with a 6+ insulin trimer. In some cases, isotopic resolution was sufficient to assign oligomer size and charge state unambiguously. However, unresolved isotopic envelopes frequently arose from overlapping, highly charged oligomers. In these cases, IM separation was used to resolve species of identical m/z. Throughout this work, oligomers are denoted as n ^z+, where n indicates the number of insulin monomer units and z the charge state. For zinc‐containing species, the number of coordinated Zn²⁺ ions is specified explicitly; for example, the 12+ zinc‐coordinated hexamer is denoted 6¹²⁺ + 2Zn, corresponding more explicitly to [6 insulin + 2Zn + 8 H]¹²⁺.

3.2. Excipient‐Containing Solutions Favor Native Hexamer Formation

Since zinc‐coordinated hexamers have a relatively low propensity for aggregation, stabilizing these assemblies in solution suppresses the formation of intermediate and higher order oligomers. This behavior was evident for human insulin at 5 μM, as shown by Figure 1a, and at 60 μM (Figure S1a), where intermediate and larger oligomers were absent. The insulin analogs exhibited distinct responses under the same stabilizing conditions. By contrast, Figure 1b shows inefficient Zn²⁺ coordination of aspart with no observable hexamers at 5 μM and minimal hexamer formation at 60 μM (Figure S1b). This behavior is consistent with aspart's lower tendency to self‐associate. Interestingly, low‐intensity heptamers were also detected in 60‐μM aspart. Glargine displayed a contrasting pattern. At 5 μM, glargine produced abundant zinc‐coordinated hexamers together with relatively intense dimer peaks (Figure 1c). At 60 μM, however, oligomers up to heptamers were observed, whereas zinc‐coordinated hexamers were no longer detected (Figure S1c). These data suggest that although Zn²⁺ can effectively coordinate and stabilize glargine, increasing concentration shifts the oligomeric equilibrium toward non‐native self‐association, thereby competing with hexamer stabilization.

FIGURE 1.

Mass spectra of excipient‐containing samples with 5‐μM insulin analogs. Each peak is identified by its oligomeric and charge state. (a) Human insulin at pH 7.4 with peaks representing monomers, and a magnified region of the spectrum showing low‐intensity zinc‐coordinated hexamers. (b) Aspart at pH 7.4 with peaks denoting mainly monomers. (c) Glargine at pH 3.8 with peaks denoting highly intense zinc‐coordinated hexamers along with monomers and dimers. *Peak represents a slight human insulin impurity.

3.3. Oligomer‐Inducing Samples

Samples prepared near the pI of each insulin analog were used to favor self‐association and aggregation. Under these oligomer‐inducing conditions, distinct analog‐dependent differences were observed. Figure 2a illustrates the oligomerization of human insulin, with mainly smaller oligomers, up to tetramers, formed at 5 μM and larger oligomers, up to octamers, clearly observed at 60 μM (Figure S2a). By contrast, aspart was dominated by monomer signal, with only minimal dimer and trimer abundance at 5 μM, again consistent with its reduced self‐association tendency (Figure 2b). At 60 μM, however, aspart demonstrated a similar degree of self‐association as human insulin (Figure S2b). Glargine, by comparison, produced a broad oligomeric distribution extending to heptamers, regardless of concentration, as shown in Figures 2c and S2c. Whereas human insulin and aspart primarily showed increased abundance of the same general oligomer classes observed at 5 μM, glargine displayed a relative enrichment in heptamers, accompanied by a decrease in smaller oligomers. This redistribution suggests that glargine may be progressing toward the formation of less soluble aggregates under these conditions, thereby reducing the population of smaller soluble species available for detection by ESI‐MS.

FIGURE 2.

Mass spectra of oligomer‐inducing samples with 5‐μM insulin analogs. Each peak is identified by its oligomeric and charge state. (a) Human insulin at pH 5.0 with peaks representing multiple small oligomers. (b) Aspart at pH 5.0 with peaks denoting minor oligomerization. (c) Glargine at pH 6.8 with peaks representing various oligomers up to heptamers.

3.4. Structural Comparisons of Native Hexamers and Non‐Native Heptamers

A consistent result across all three insulin variants was the preferential formation of zinc‐coordinated hexamers in excipient‐containing solutions and non‐native heptamers in oligomer‐inducing solutions. Although native insulin hexamers are well established, the recurrent appearance of heptamers as a comparatively stable higher order oligomer is less commonly discussed. To compare these assemblies, we used CID to probe their dissociation patterns and CIU to assess conformational stability.

3.5. Dissociation of Native Zinc‐Coordinated Hexamers vs. Non‐Native Heptamers

Across all charge states, zinc‐coordinated hexamers dissociated predominantly by monomer ejection, with the larger complementary fragment generally retaining the zinc ions. This dissociation preference is illustrated by Figure 3, which compares the dissociation of human insulin and glargine 12+ hexamers (6¹²⁺ + 2Zn) at 5 μM, and Figure 4, which compares the dissociation of human insulin and aspart 12+ hexamers at 60 μM. Note that aspart failed to form hexamers at 5 μM (Figure 1b) and glargine failed to form hexamers at 60 μM (Figure S1c). Although all analogs showed mainly monomer ejection, glargine produced monomers with a larger average charge state. This could be a result of glargine's design, which includes two extra basic arginine residues. Product ions such as zinc‐bound pentamers were consistently observed, indicating that metal coordination remains associated with the larger oligomeric fragment during dissociation. Another recurring feature was the formation of trimeric species containing one zinc ion, which were resolved by combining mass spectral and IM information (Figure S3). Rather than supporting the zinc‐coordinated trimer‐of‐dimers structure, this result seems to indicate a dimer‐of‐trimers structure, where each trimer is coordinated by one Zn²⁺ ion.

FIGURE 3.

CID mass spectra of native zinc‐coordinated hexamers (6¹²⁺ + 2Zn) resulting from excipient‐containing solutions with 5‐μM insulin analogs. The applied CID voltage was 30 V. (a) Human insulin at pH 7.4. (b) Glargine at pH 3.8.

FIGURE 4.

CID mass spectra of native zinc‐coordinated hexamers (6¹²⁺ + 2Zn) resulting from excipient‐containing solutions with 60‐μM insulin analogs. The applied CID voltage was 30 V. (a) Human insulin at pH 7.4. (b) Aspart at pH 7.4.

Similar to native hexamer dissociation, non‐native heptamers almost exclusively dissociated via monomer ejection. Again, glargine produces monomers at a higher average charge state, likely due to the longer amino acid sequence with additional basic residues. Minor dimer/pentamer and trimer/tetramer product pairs were observed at low abundance, as shown in Figure 5. Although low in intensity, the formation of dimer/pentamer and trimer/tetramer product pairs potentially indicates some higher order structure in the heptamer.

FIGURE 5.

CID mass spectra of non‐native heptamers (7¹³⁺) resulting from oligomer‐inducing solutions with 60‐μM insulin analogs. The applied CID voltage was 30 V. (a) Human insulin at pH 5.0. (b) Aspart at pH 5.0. (c) Glargine at pH 6.8.

Although dissociation patterns among the insulin analogs were identical, the energies required for complete dissociation appeared to mirror their individual propensities for oligomer formation, as shown in Table 1. The 12+ glargine hexamer required ~10 V (~120 eV) more to fully dissociate than human insulin and aspart. Similarly, the glargine 13+ heptamer required ~15 V (~195 eV) more than the human insulin and aspart 13+ heptamers for complete dissociation. Thus, the higher prevalence of either native hexamer or non‐native heptamer formation by glargine appears to come with an increased gas‐phase stability of the complexes. Although aspart does not form oligomers as readily as human insulin, the dissociation experiments suggest that their stabilities once formed are similar. Additionally, comparing the collision energies calculated from collision voltage and charge state suggests that each analog's heptamer is more stable than its hexamer.

TABLE 1.

Approximate CID energies required for complete dissociation of insulin analog native 12+ hexamers and non‐native 13+ heptamers.

	Hexamer (612+ + 2Zn)			Heptamer (713+)
	Human	Aspart	Glargine	Human	Aspart	Glargine
CID (eV)	480	480	600	520	520	715

3.6. Gas‐Phase Unfolding of Hexamers and Heptamers

In the case of protein complexes, CID typically results in monomer subunit unfolding preceding dissociation from the rest of the complex. The CIU experiment of human insulin 12+ hexamer demonstrates this behavior, as shown in Figure 6a. As stated previously, the 12+ hexamer at m/z 2915 can dissociate into two 6+ trimers, each with a Zn²⁺ ion, at the same m/z. Because CIU monitors the change in IMS arrival time for a defined m/z window, all isobaric species are included in the CIU fingerprint. For m/z 2915, this produces a clear picture of hexamer unfolding at ~21 V, followed by a clear dissociation event at ~40 V. The identity of CIU features can be verified by inspecting the isotopic spacing of each IMS peak (Figure S3).

FIGURE 6.

(a) CIU of native zinc‐coordinated hexamers (6¹²⁺ + 2Zn) indicates partial unfolding of the oligomer at 21 V. (b) CIU of 5⁹⁺ + 2Zn, which is a major dissociation product of the native zinc‐coordinated hexamers (6¹²⁺ + 2Zn), also indicates partial unfolding of the dissociation product.

Additionally, dissociation products of zinc‐coordinated hexamers showed evidence of unfolding. An example of this is the 9+ pentamer product (5⁹⁺ + 2Zn) shown in Figure 6b. By contrast, isolated oligomer species ranging from monomers to pentamers appeared to show no evidence of unfolding. Besides providing additional evidence that ESI under native‐like conditions can preserve protein tertiary structure in the gas phase, this observation indicates that the larger subcomplex product maintains a defined structure following monomer ejection.

Although the heptamer demonstrated solution stability over other oligomers and potentially some quaternary structural characteristics, the CIU experiment indicates no clear tertiary structure in heptamer subunits. Extracted arrival time distributions from the CIU fingerprint illustrate this behavior most clearly (Figure S4). Initially, there is a defined peak for the heptamer at low activation energies; however, increasing energy does not result in a clear shift to a new structure like the zinc‐coordinated hexamer. Instead, the peak appears to broaden until signal is lost, indicating that the complex is adopting a continuous range of unstructured conformations rather than distinct states.

4. Discussion

Comparing the measured oligomer distributions agreed well with expected behavior based on analog pharmaceutical design. In general, the fast‐acting aspart showed resistance to oligomerization, and the long‐acting glargine showed a prevalence for aggregation when compared with human insulin. In addition, aspart's resistance to aggregation and glargine's propensity to aggregate were observed in the excipient‐containing samples based on how well Zn²⁺ appeared to facilitate the formation of zinc‐coordinated hexamers. Although excipients stabilized the long‐acting glargine at low concentrations, they appeared ineffective at higher concentrations. Identifying correct solution conditions appears particularly important for glargine formulations to retain their intended pharmaceutical activity.

Although glargine native hexamers and non‐native heptamers demonstrated the highest resistance to dissociation via CID, there was no clear difference between dissociation thresholds for human insulin and aspart complexes. The observed higher gas‐phase stability of glargine correlates with its long action behavior. The more soluble aspart monomer results in fewer solution‐formed oligomers being transferred to the MS than human insulin; however, once successfully transferred to the gas‐phase, human insulin and aspart oligomers exhibit similar stability in the absence of solvent, suggesting some limitation in characterizing long action vs. rapid action simply by CID. Although these data cannot directly evaluate in vivo pharmacokinetics of oligomer assembly and stability, the combination of observed oligomer distributions and measured dissociation thresholds does provide useful qualitative trends.

Regardless of analog aggregation propensity, non‐native heptamers appeared to be relatively stable structures, indicating some preferential higher order structure that might mirror native zinc‐coordinated hexamers. One potential explanation is that an insulin monomer could function as a pseudo‐coordination center via protonated N‐termini or protonated basic residues. Because monomer unfolding is expected to precede aggregation [2], the heptamer subunits would have no defined tertiary structure, even if the quaternary structure is favored. CIU experiments indicate that hexamers maintain aspects of their defined solution‐phase tertiary structure, whereas heptamers have no defined tertiary structure.

Investigation of hexamer and heptamer quaternary structure via CID mainly produced the expected monomer ejection as the primary dissociation pathway for both species. In addition to monomer ejection, hexamers generated zinc‐adducted trimers, whereas heptamer dissociation resulted in minor dimer/pentamer and trimer/tetramer product pairs. Although monomer ejection would not be expected from the accepted trimer‐of‐dimer structure of insulin hexamer, there is evidence for insulin hexamer disassembly via monomer dissociation in solution [28]. Based on the CIU, we showed evidence of hexamer unfolding preceding dissociation and further unfolding of hexamer dissociation products. Although heptamers appear to be stable oligomers that might mimic native zinc‐coordinated hexamers, the different unfolding behaviors and minor dissociation pathways indicate unique structural differences.

5. Conclusions

Using IM–MS, we characterized insulin analog oligomerization resulting from aggregation and zinc‐coordination. The observed oligomer distributions and gas‐phase stabilities of insulin analogs were consistent with their clinical designations as rapid‐acting or long‐acting species. These experiments provide a basis for initial rapid characterization of additional insulin analogs and the effects of additional excipients that can be validated via more rigorous pharmacokinetic studies. Although non‐native heptamers behaved distinctly from native hexamers during gas‐phase activation, the prevalence of heptamers in oligomer‐inducing solutions indicates a potential for insulin to self‐coordinate during aggregation, reminiscent of zinc‐assisted coordination. Further investigation into the structure and stability of intermediate oligomers, such as the aggregation‐induced heptamer, would elucidate more information about the full dynamics of insulin analog aggregation pathways for better design of insulin formulations, including analogs and excipients. Beyond insulin, this work establishes approaches that can be applied to other proteins whose aggregation is a primary concern in long‐term human health.

Supporting information

Figure S1: Mass spectra of excipient‐containing samples with 60‐μM insulin analogs. Each peak is identified by its oligomeric and charge state. (a) Human insulin at pH 7.4 with peaks representing monomers, dimers, and zinc‐coordinated hexamers. (b) Aspart at pH 7.4 with peaks representing monomers, dimers, and low‐intensity zinc‐coordinated hexamers. (c) Glargine at pH 3.8 with peaks representing various oligomers up to heptamers, and a lack of zinc‐coordinated hexamers.

Figure S2: Mass spectra of oligomer‐inducing samples with 60‐μM insulin analogs. Each peak is identified by its oligomeric and charge state. (a) Human insulin at pH 5.0. (b) Aspart at pH 5.0. (c) Glargine at pH 6.8. All analogs show oligomerization up to heptamers or octamers, with relatively intense heptamer peaks as compared with other oligomers.

Figure S3: Mass spectra and arrival time distributions of m/z 2915 during CID of this precursor ion, illustrating the dissociation of human insulin 6¹²⁺ + 2Zn to 3⁶⁺ + 1Zn.

Figure S4: Arrival time distribution of 13+ human insulin heptamer measured following increasing collisional activation. The experiment shows a broadening distribution before complete signal loss without a clear unfolding event.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.