Calpain small subunit homodimerization is robust and calcium‐independent

Abstract

Calpains dimerize via penta‐EF‐hand (PEF) domains, but their mechanical stability and kinetics remain incompletely characterized. Here, we used single‐molecule force spectroscopy on an EGFP‐tagged calpain small subunit PEF (CAPNS1‐PEF) homodimer to quantify dissociation mechanics. Across retraction speeds of 0.2–10 μm/s, homodimer rupture required large forces (> 300 pN). Rupture forces and force‐induced unfolding did not change when Ca²⁺ was removed, suggesting Ca²⁺‐independent mechanics, consistent with crystal structures showing minimal Ca²⁺‐induced conformational changes in the PEF domain. Together, these results indicate that EF‐hands in the CAPNS1 homodimer function primarily as structural elements within a mechanically robust, Ca²⁺‐insensitive dimerization scaffold.

Calpains dimerize via penta‐EF‐hand (PEF) domains. Using single‐molecule force spectroscopy, we measured the strength and kinetics of PEF–PEF homodimer binding. The interaction is robust, shows a transient conformational step before dissociation, and remains largely insensitive to Ca²⁺. Together, these parameters establish a quantitative reference for PEF assemblies and can be adapted to other PEF‐domain pairings.

Abbreviations

AFM, atomic force microscopy

CAPNS1, calpain small subunit

EGFP, enhanced green fluorescent protein

PEF, penta‐EF‐hand domain

SMFS, single‐molecule force spectroscopy

Penta‐EF‐hand (PEF) domains were discovered in calpains, which are intracellular calcium‐activated cysteine proteases [1, 2, 3, 4]. Ca²⁺‐binding EF‐hand motifs, of which there are four in calmodulin, are typically found in pairs [5]. For this reason, the fifth EF‐hand in calpain sequences was overlooked until structural biology revealed its presence. Fittingly, the fifth EF‐hand of the calpain small subunit (CAPNS1) was indeed paired up by being used as a dimerization motif to form a homodimer of the small subunit [6]. The first structures of this PEF domain were solved by two groups. One showed the crystal structure of rat calpain small subunit PEF‐domain homodimer in the presence (1DV1) and absence (1AJ5) of Ca²⁺ [7]. The presence of this divalent metal ion caused a relatively small conformational change, which was the opening of a shallow hydrophobic cleft. The other structure was of pig CAPNS1 PEF‐domain dimer done in the presence of Ca²⁺ (1ALV) [8]. A second structure from the same laboratory was solved with an aromatic compound in the hydrophobic cleft (1ALW). In these structures, EF‐hands 1 and 2, and 3 and 4, were each paired together within the PEF domain. Subsequently, several other PEF‐domain proteins were discovered, all of which dimerize through the pairing of their fifth EF‐hand motif [9]. These include grancalcin, which is present in high levels in neutrophils, [10] ALG‐2, a protein linked to apoptosis, [11] peflin, [12], and sorcin [13].

An interesting feature of PEF domains in calpains is that some of them can form both homo‐ and heterodimers. For example, CAPNS1, which dimerizes with itself, forms heterodimers with the PEF domain at the C terminus of the large subunits of calpain‐1 and calpain‐2 [4, 14, 15]. Calpain‐3 from muscle, which is homologous to the calpain‐1 and calpain‐2 large subunits and has the same four domains ending with the C‐terminal PEF domain, [16] is strictly homodimeric through pairing up of the latter domain [17, 18]. Its PEF domain does not heterodimerize with the calpain small subunit when co‐expressed in E. coli [19].

Despite these structural insights, the energetics and kinetics that dictate PEF–PEF selectivity, that is, why some pairs homodimerize whereas others heterodimerize, remain unclear. Addressing this gap requires measurements that quantify the stability and lifetime of these dimer interfaces. Atomic force microscopy single‐molecule force spectroscopy (AFM–SMFS) enables such measurements by stretching an individual protein–protein complex to dissociation and determining the unbinding force and, across loading rates, the energy–landscape parameters that govern dimer interactions [20, 21, 22].

Here, we apply AFM–SMFS to quantify the force required to dissociate the calpain small subunit PEF homodimer. Rupture forces fall in the few hundred of piconewtons—exceeding values reported for biotin‐streptavidin interactions, among the strongest noncovalent linkages in biology (affinity of 10¹³–10¹⁵ m ⁻¹), under comparable loading rates [23, 24, 25, 26]. In a subset of force curves, we also observe an intermediate we attribute to partial unfolding of the PEF prior to dimer separation. The associated contour‐length increase of this partially unfolded PEF domain is consistent with unfolding of up to a single EF‐hand motif before dimer dissociation. In the presence of Ca²⁺, no significant difference in the rupture force is detected, indicating that subtle conformational changes seen in the crystal structure do not measurably alter the dimerization interface between the PEF domains. The AFM–SMFS framework provides a quantitative baseline for the small subunit homodimer and can be adapted to evaluate factors that modulate PEF‐mediated assemblies in calcium signaling and protease regulation.

Materials and methods

Cloning and expression of CAPNS1 PEF domain

The protein construct is based on the crystal structure (PDB: 1DVI), corresponding to UniProtKB accession Q64537 (residues 95–270). A DNA sequence encoding an N‐terminally Enhanced GFP (EGFP)‐tagged CAPNS1 PEF‐domain construct, was synthesized by GenScript with codon optimization for expression in E. coli. This construct was ordered into a pET‐24a(+) vector to provide a C‐terminal hexahistidine (His6) tag. In the EGFP‐CAPNS1 construct, the EGFP sequence included an A206K mutation to prevent EGFP dimerization [27]. Additionally, alanine and cysteine were inserted at Positions 2 and 3, respectively, after the start codon to enable anchoring of the fusion protein via thiol‐gold interactions to both the gold cantilever tip and the gold basal surface [28, 29].

The EGFP‐CAPNS1 construct was expressed and purified as previously described [19]. Briefly, BL21 (DE3) E. coli carrying expression vectors with EGFP‐CAPNS1 were grown at 37 °C in a shaking incubator until they reached an OD_600nm of 0.6, at which point expression was induced by the addition of 1 mg isopropyl thio‐β‐d‐galactoside (IPTG) per liter of culture. Cultures were then grown overnight at room temperature with shaking. The next day, cells were harvested by centrifugation, resuspended with lysis buffer [25 mm Tris/HCl pH 7.6, 5 mm EDTA, 5% (v/v) glycerol, 10 mm 2‐mercaptoethanol, and 0.1 mm phenylmethanesulfonyl fluoride], and lysed by sonication. The cellular debris was pelleted by centrifugation, and the supernatant was incubated with Ni²⁺‐nitrilotriacetate‐agarose (Ni‐NTA) resin for 20 min at 4 °C with constant stirring. The Ni‐NTA resin was then transferred to a column and washed with nickel wash buffer [50 mm Tris/HCl pH 7.6, 150 mm NaCl, 2% (v/v) glycerol, 10 mm imidazole]. Bound proteins were eluted off the column with nickel wash buffer containing 250 mm imidazole. The collected fractions were analyzed by SDS/PAGE, concentrated to 2.5 mg/mL, aliquoted, and snap frozen in liquid nitrogen. The Ni‐affinity purified EGFP‐CAPNS1 PEF construct was dialyzed against 50 mm sodium phosphate pH 7.2 with 50 mm NaCl and 10 mm EDTA.

AFM‐based single‐molecule force spectroscopy (SMFS)

Force spectroscopy was performed on a JPK Nanowizard® 4 (JPK Instruments, Berlin, Germany) mounted on an inverted optical microscope (Zeiss Axio Observer 7). A 50‐μL sample of EGFP‐CAPNS1 homodimer (1 mg/mL) was immobilized on a piranha‐cleaned, gold‐coated Si wafer overnight at 4 °C. The resulting substrate was extensively washed in 20 mm HEPES (pH 7.5) supplemented with either 5 mm Ca²⁺ for force measurement in the presence of Ca²⁺ or 1 mm EDTA for Ca²⁺‐free measurements. Gold‐coated NPG‐D cantilevers (Bruker, k ~ 60 pN·nm⁻¹) were brought into contact with the immobilized protein for 3 s to establish an Au–S bond with accessible cysteines. Cantilever spring constants were determined by the thermal noise method and ranged from 0.01 to 0.04 N/m [30]. Force–extension curves were acquired at six pulling speeds (0.2, 0.5, 1, 2, 5, and 10 μm/s) in the presence of calcium, while Ca²⁺‐free measurements were collected at 1 μm/s. All experiments were performed at room temperature. To account for possible heterogeneities of the surface, a total of 5000–10,000 approach–retraction cycles were recorded across a 32 × 32 grid within a 5 μm × 5 μm area.

Results

Unfolding of EGFP reporters precedes PEF dimer rupture

We used AFM–SMFS to measure the force required to dissociate the CAPNS1‐PEF homodimer. A key experimental challenge was to distinguish the actual dimer interface rupture from nonspecific tip–surface adhesion and multiple simultaneous attachments. To clearly identify single‐molecule events, we fused an EGFP reporter to each PEF domain, generating an EGFP‐CAPNS1 PEF–PEF CAPNS1‐EGFP dimer complex. The characteristic mechanical unfolding of full‐length EGFP served as an internal molecular fingerprint to ensure we only included traces with single‐molecule extensions and terminal rupture of the PEF–PEF dimer interface [31, 32]. To enforce a defined pulling geometry, we introduced a cysteine near the EGFP N terminus; its thiol group formed Au–S bonds with the gold substrate (anchoring one EGFP) and with the gold‐coated AFM tip (via the second EGFP) [32, 33]. In each pulling cycle, a 3 s tip‐sample dwell time was applied to promote Au–S bond formation [34].

Figure 1A illustrates the experimental set‐up and the sequence of events during AFM single‐molecule pulling that led to PEF homodimer dissociation. As the AFM tip was retracted from the surface, the first EGFP unfolded and extended (0 and I, respectively), followed by unfolding (II) and extension (III) of the second EGFP, and finally rupture of the PEF–PEF dimer interface (IV). Although the schematic shows the substrate‐anchored EGFP unfolding first, the unfolding order is stochastic and unbiased, that is, either of the two EGFP may unfold first. Because the dimer was directly immobilized on both gold surfaces (i.e., AFM tip and gold substrate), the first EGFP would unfold early in the pulling process, within only a few nanometers of tip–surface contact. As such, this event was often obscured by strong nonspecific adhesion between the tip and the gold substrate (Fig. 1B–0) [29]. Thus, only the subsequent extension of the first unfolded EGFP was visible in the force–extension profile (Fig. 1B–I). Once the first EGFP had fully extended, unfolding of the second EGFP could be observed (Fig. 1B–II).

Fig. 1.

Single‐molecule force spectroscopy of CAPNS1 penta‐EF‐hand (PEF) dimer dissociation. (A) Schematic of the EGFP‐CAPNS1 PEF dimer chimera protein stretched between a gold surface and a gold‐coated atomic force microscopy (AFM) tip, showing the unfolding sequence: (0) initial unfolding of the first EGFP, (I) stretching of the first EGFP, (II) unfolding of the second EGFP followed by its (III) stretching, and (IV) final dissociation of the dimer interface. (B) Typical force–extension traces of EGFP‐CAPNS1 PEF dimer, showing the unfolding stages (0‐IV) described in (A). Dashed lines correspond to worm‐like chain (WLC) fits, highlighting the contour‐length difference (ΔL_EGFP) as corresponding to the full extension of an EGFP domain. (C) Histogram of ΔL_EGFP corresponding to EGFP stretching, centered at 73 ± 6 nm (mean ± std dev of the Gaussian fit), for all selected force curves (n = 258). Inset: Distribution of contour lengths obtained from WLC fits of all selected force curves with peak values at 67 ± 11 nm and 140 ± 16 nm. (D) Distribution of unfolding forces obtained from WLC fits of Fig. 1B–I, which corresponds to the unfolding of an EGFP domain, 97 ± 26 pN (mean ± std dev), pulled at 1 μm/s (n = 32). EGFP, enhanced green fluorescent protein.

The stretching of the first unfolded EGFP was then analyzed using the worm‐like chain (WLC) model (Fig. 1B‐lime green‐dashed line in I). From selected force curves, the measured contour length was 67 ± 11 nm (Fig. 1C, inset‐first peak), shorter than the theoretical contour length of 84–87 nm for a fully extended EGFP construct (241 amino acids, 0.35–0.36 nm per residue). This discrepancy likely arises from nonspecific interactions between the EGFP and the gold surface, effectively reducing the apparent contour length. The unfolding force for EGFP obtained at a pulling velocity of 1 μm/s was 97 ± 26 pN (mean ± sd from Gaussian fit, Fig. 1D), consistent with previously reported values for GFP unfolding [32, 35].

After the first EGFP reached full extension, continued retraction caused the second EGFP to unfold (Fig. 1B–II) and extend (Fig. 1B–III), followed by final rupture (Fig. 1B–IV). The WLC fit of this second EGFP extension (green‐dashed line in Fig. 1B–III) gave a contour length of 140 ± 16 nm (Fig. 1C, inset, second peak), ~ 73 nm longer than the first peak. Because rupture of the PEF–PEF dimer would separate the tip from the surface and occur only after the second EGFP was fully extended, these signature traces therefore imply that the final event (Fig. 1B–IV) must correspond to the dissociation of the PEF–PEF dimer interface. By restricting analysis to single‐molecule traces with this characteristic profile and a contour‐length difference of 73 ± 6 nm between the first and second rupture (Fig. 1C), we isolated and quantified the force required to break the PEF–PEF dimer interface while minimizing contributions from nonspecific and multiple protein‐tip contacts.

A partially unfolded intermediate is observed before PEF–PEF dimer rupture

Although force–extension profiles resembling those in Fig. 1B were observed in most selected force curves, intermediate peaks preceding the terminal rupture were occasionally observed (Fig. 2). Prior single‐molecule force spectroscopy experiments of protein unfolding indicate that such features often correspond to partially unfolded intermediates, in which a substructure unravels before the entire domain is completely destabilized [36, 37]. In our experiments—because we are pulling apart a dimer interface—these intermediate peaks likely reflect partial unfolding of PEF‐domain motifs prior to dimer separation.

Fig. 2.

Partially unfolded intermediate observed for a subset of traces before PEF–PEF dimer rupture. Representative AFM–SMFS force–extension curve (left, black), like those shown in Fig. 1B, but showing an intermediate peak before the final rupture event (n = 11 of 39 force curves at 1 μm/s). The lime green‐dashed line is the worm‐like chain (WLC) fit of the extension of the first enhanced green fluorescent protein (EGFP). The green‐dashed line is a WLC fit to the intermediate peak, attributed to extension of the second EGFP immediately before partial unfolding within the PEF domain. The red dashed line is a WLC fit to the subsequent extension of the newly unfolded PEF segment before final PEF–PEF dissociation. The boxed region is magnified (right) to show the contour‐length difference between intermediate peak and final rupture event, that is, ΔL_Unfolding. Inset: Boxplot showing the distribution of the length of unfolded segment (ΔL_Unfolding) from n = 11 traces, with horizontal line indicating the median; box range: 25^th/75^th percentile; and whiskers range: 5^th/95^th percentile. Note that the slanted slope near zero extension likely reflects sample heterogeneity and/or minor drift in optical lever sensitivity, and may also arise from the curve‐alignment step in our analysis. PEF, penta‐EF‐hand; AFM, atomic force microscopy.

To estimate the extent of prerupture unraveling, we quantified the contour‐length difference between the intermediate peak and the final rupture (ΔL_Unfolding in Fig. 2‐right). The distribution of ΔL_Unfolding for 11 apparent unfolding events (out of 39 force curves) at 1 μm/s was relatively broad (Fig. 2‐right inset), consistent with heterogeneous partially unfolded states. However, the majority (~ 70%) of these partial unfolding events showed extensions < 11 nm before dimer rupture. This contour‐length difference corresponds to < 30 amino acids (assuming 0.36 nm per residue), approximately the length of one EF‐hand motif in the PEF domain [38]. This implies that, for some force‐induced breaking of the dimer interface, either up to one EF‐hand motif unravels from a single PEF domain of the dimer, or up to one of the two helices of the EF‐hand motif unravels from each PEF domain prior to dissociation. Our single‐molecule pulling data cannot discriminate between these possibilities. Nonetheless, the data support the existence of a partially unfolded intermediate preceding force‐induced dimer rupture.

Dimer rupture requires large forces and is Ca²⁺‐independent

Having isolated rupture events corresponding to PEF homodimer dissociation (i.e., rupture events in selected traces matching the profiles in Figs 1B and 2A), we quantified the forces required to separate the homodimer. At a pulling velocity of 1 μm/s, the distribution of rupture forces measured in the presence of 5 mm Ca²⁺ (n = 39) range from 120 to 600 pN with a mean ± sd rupture force of 380 ± 150 pN (Fig. 3A). Because prior work indicated that Ca²⁺ does not alter the binding of the PEF homodimer, [39] we repeated the measurements in the absence of Ca²⁺. The resulting forces (370 ± 150 pN, n = 32) did not differ significantly from the Ca²⁺‐bound condition (two‐sided $t$‐test, $P-\mathit{\text{value}}$ > 0.05, Fig. 3A). Furthermore, the absence of Ca²⁺ also did not affect the partial unfolding observed during pulling of Ca‐bound PEF. The frequency of unfolding events was comparable under both conditions: 28% of selected force curves displayed an unfolding event in the presence and absence of calcium (w/ Ca²⁺: 11 of 39 and w/o Ca²⁺: 9 of 32); and the extent to which the PEF domain(s) unfolded prior to rupture also remains statistically similar between these two conditions (Fig. 3B; $t$‐test, $P$ > 0.05). The forces required to partially unfold PEF in the absence of Ca²⁺ (220 ± 80 pN) were comparable ($t$‐test, $P$ > 0.05) to those measured in the presence of Ca²⁺ (240 ± 100 pN, Fig. 3C). Taken together, these suggest that the minimal conformational differences between Ca²⁺‐bound and Ca²⁺‐free PEF observed in structural studies do not translate into a measurable change in the mechanical stability of the PEF homodimer or in the sequence of force‐induced events. These observations support a model in which the energy barrier governing dimer failure, and the associated mechanical sequence (occasional partial unfolding followed by rupture), is not appreciably modulated by Ca²⁺, implying that the Ca²⁺‐binding sites are not substantially load‐bearing along the dominant unbinding pathway under these conditions.

Fig. 3.

Effect of Ca²⁺ on PEF–PEF mechanics. (A) The rupture forces of PEF–PEF dimer dissociation, (B) the contour‐length difference of the partially unfolded intermediate (ΔL_Unfolding), and (C) the forces needed to partially unfold the PEF domain, with their corresponding dimer rupture force, are similar in the presence (w/ Ca²⁺) and absence (w/o Ca²⁺) of calcium. All data were acquired at a pulling velocity of 1 μm/s. Solid dots indicate the mean; horizontal line indicating the median; box range: 25^th/75^th percentile; and whiskers range: 5^th/95^th percentile. Statistical significance was determined using a two‐sample t‐test (Origin Pro). PEF, penta‐EF‐hand.

Rupture forces increase with loading rate

To gain more insights into the energetics of CAPNS1 PEF homodimerization, we performed dynamic force spectroscopy on the EGFP‐CAPNS1 PEF homodimer in the presence of Ca²⁺ over loading rates spanning 10³–10⁵ pN/s by varying the retraction speed (0.2–10 μm s⁻¹). Similar force–extension profiles were observed across all pulling speeds (Fig. S1), resulting in a 73 ± 6 nm contour‐length increase (ΔL_EGFP) between unfolding of the second EGFP and the final rupture event (Fig. S2). The speed invariance of ΔL_EGFP further supports a common structural pathway (complete EGFP unfolding) preceding the terminal rupture. Partial unfolding of PEF was also observed across pulling speeds (Fig. S3), and the distribution of the unfolded length prior to rupture (Fig. S4) shows that most events have ΔL_Unfolding < 11 nm, consistent with up to 1 EF motif unraveling before rupture. With increasing pulling speed, however, the distribution of final rupture force shifted to higher values (Fig. 4A). The mean rupture force increased linearly with the logarithm of the applied loading rate (Fig. 4B), consistent with dissociation governed by a single dominant energy barrier over the loading rates tested, as predicted by Evans and Ritchie to describe the dynamics of adhesion bonds [23, 40]. The Bell‐Evans model, [23] allowed us to estimate the potential barrier width, $x_{\mathrm{\beta }}$ ~ 0.2 nm, beyond which the dimer was completely dissociated along the applied force direction, and a zero‐force off‐rate, $k_{\mathrm{off}}$ ~ 3 × 10⁻⁵ s⁻¹ for PEF homodimer dissociation. An $x_{\mathrm{\beta }}$ of 0.2 nm indicates a very short distance to the transition state, consistent with a compact interface where only a few short‐range contacts, likely within the EF5‐EF5 dimer interface, bear the load at rupture. From the zero‐force off‐rate, we estimate the activation barrier for homodimer rupture to be 40 $k_{\mathrm{B}}\mathrm{T}$ and the lifetime of CAPNS1 dimer to be τ₀ ~ 1/$k_{\mathrm{off}}$ ~ 9 h. These values indicate that, in the absence of load, the homodimer can persist for long periods, such that only very rare thermal fluctuations would surmount the barrier.

Fig. 4.

Dynamic force spectroscopy of PEF–PEF dimer rupture. (A) PEF homodimer rupture force distributions obtained from individual single‐molecule events selected using the EGFP‐reporter signature and measured at different retraction velocities, 0.20–10 μm/s. Red dots represent the mean dimer rupture force; horizontal line indicating the median; box range: 25^th/75^th percentile; and whiskers range: 5^th/95^th percentile. (B) Dynamic strength spectrum for the rupture of CAPNS1‐PEF homodimer with the Bell‐Evans model fit (red line). Mean rupture forces and standard deviations (error bars) were obtained from n = 39 to 70 dimer rupture events. EGFP, enhanced green fluorescent protein; PEF, penta‐EF‐hand.

Discussion

Our single‐molecule pulling strategy relies on the assumption that the PEF–PEF dimer interface is mechanically more stable than the folded EGFP reporters. If the interface were weaker, the terminal rupture would occur within a few nanometers of tip–surface contact and be masked by nonspecific adhesion (Fig. 1B–0), and the characteristic force–extension curve (Fig. 1B) expected from our engineered protein construct (Fig. 1A) would not be observed. However, the single‐molecule force–extension curves we obtained demonstrate that EGFP unfolding occurs before dimer rupture, providing a robust molecular fingerprint that we used to select force curves for further data analysis (i.e., the characteristic extension of the first EGFP, followed by the unfolding and extension of the second EGFP, before a terminal rupture corresponding to PEF–PEF dissociation; Figs. 1B–I–IV). The inclusion of EGFP handles therefore provided an internal quality‐control criterion to identify correctly tethered single dimers and allowed us to exclude nonspecific or multi‐molecular PEF–PEF rupture events. Furthermore, we note that the terminal rupture events could, in principle, arise from failure of the Au–S linkage used to attach EGFP‐PEF dimer on Au‐coated substrate and AFM tip; however, the forces observed here are well below those typically required to break Au–S bonds, making this alternative explanation unlikely [34].

Within this framework, PEF homodimer unbinding consistently required large forces—on average, > 300 pN across all loading rates tested. Comparable magnitudes have been reported for proteins involved in synaptic vesicle fusion (200–300 pN), [41] antibody–antigen pairs (244 ± 22 pN, $K_{\mathrm{d}}$ = 10 nM) [42] and biotin‐streptavidin (207 ± 58 pN, $K_{\mathrm{d}}$ ~ 10 fM), [25] under similar pulling velocities. These comparisons support a very tight CAPNS1 dimer interface, consistent with our dynamic force spectroscopy experiments where the estimated off‐rate ($k_{\mathrm{off}}$) for the dissociation for CAPNS1 homodimer at zero‐force was ~ 3 × 10⁻⁵ ${\mathrm{s}}^{-1}$.

Our force spectroscopy measurements also identified limited structural rearrangements within the PEF domains prior to rupture. Across different pulling speeds, ~ 66–70% of force curves exhibiting partial unfolding intermediates extended by ≤ 11 nm, consistent with unraveling of < 30 residues, on the order of a single EF‐hand motif, before dimer separation. These intermediates occurred at high forces (> 150–200 pN, Fig. S5) and were more frequent in traces with larger terminal rupture forces, suggesting that they accompany loading of an already robust interface rather than initiate failure. Thus, partial local unraveling appears to be off‐pathway or rapidly reversible relative to the rate‐limiting barrier for dimer dissociation. Consistent with this, the partial unfolding within the PEF domain does not appear to occur at the dimerization interface (i.e., EF5), which would otherwise trigger rupture. Instead, this partial unfolding likely occurs within one of the remaining four EF‐hand motifs. Determining which EF‐hand unfolds prior to the final force‐induced dimer rupture, however, remains to be elucidated. The magnitudes of these unfolding forces agree with prior measurements on calcium‐loaded calmodulin (200–600 pN), [43] supporting the view that EF‐hand helices in CAPNS1 are mechanically resilient.

Moreover, our results show that Ca²⁺ has no detectable effect on any of the mechanical readouts from pulling individual dimers apart. Neither the distribution of rupture forces nor the frequency and extent of partial unfolding differed significantly between Ca²⁺‐bound and Ca²⁺‐free conditions, in marked contrast to calmodulin with four EF motifs, in which Ca²⁺ binding dramatically stabilizes and elevates unfolding forces of holo‐CaM versus that of Ca²⁺‐free apo‐CaM [43]. The observation that PEF–PEF binding remains robust and largely Ca²⁺‐insensitive suggests that this interaction may primarily serve a structural role—maintaining stable subunit association under basal cytosolic Ca²⁺ (~ nM) concentrations and during transient Ca²⁺ elevations (~ μM) that trigger calpain activation. These findings are consistent with crystallographic reports of only minor conformational differences between Ca²⁺‐bound and Ca²⁺‐free PEF domains, [7, 14, 44] as well as with observations that the dissociation constant is unchanged upon calcium removal [39]. EF5 has been reported to lack Ca²⁺ coordination in CAPNS1 PEF structures and, as the unpaired fifth EF‐hand, mediates homodimerization via the EF‐hand ‘embrace’. These structural features are consistent with our interpretation that the dimer interface may preferentially fail near EF5 during force‐induced dissociation. Together, these findings support the view that the PEF‐domain function primarily as structural elements within a mechanically robust dimerization scaffold rather than as classical Ca²⁺‐dependent switches for structural transformation, like calmodulin. It is important to note that structural data have shown that Ca²⁺ binding to CAPNS1 PEF domain, specifically by EF‐hand 2, helps stabilize the active conformation of the enzyme [14, 45]. The PEF domains in calpains might also serve to provide a local high concentration of Ca²⁺ needed to activate the protease core. The two Ca²⁺‐binding sites in the core could promote a structural change that forms the catalytic cleft. When transient Ca²⁺ signals activate the enzyme, the EF‐hands may help ensure that the Ca²⁺ signal persists long enough to activate the protease [1].

The avidity with which the CAPNS1 PEF domains bind each other also raises interesting issues about how heterodimerization of the large and small calpain subunits wins out over small subunit homodimerization to form heterodimeric calpain‐1 and calpain‐2. It has been reported via quantitative split‐NanoLuc PEF biosensors that CAPNS1 forms detectable homodimers (K_D ~ 0.6 μm) while CAPNS1–CAPN1/2 heterodimers show higher affinity, consistent with preferential heterodimerization under physiological conditions [39]. Molecular chaperones could also influence preferential heterodimerization in vivo by affecting subunit folding or assembly kinetics. With this AFM‐based single‐molecule force measurement established, it should be possible to investigate other PEF‐domain pairings such as those in calpain‐3, whose large subunit homodimerizes exclusively despite the presence of excess CAPNS1, and also to identify the contributions of individual residues to dimerization. Single‐molecule force measurements could therefore provide reference quantities that can be compared across different PEF–PEF pairings to evaluate how sequence‐level differences translate into dimerization preference and mechanical robustness—key factors that could modulate PEF‐mediated assemblies.

Author contributions

NMOA, TDM, and MB performed the experiments and analyzed the data. PD and RMAS supervised the study. All authors contributed to the project conceptualization and experiment design, contributed to writing of the manuscript and have approved the final version.

Supporting information

Fig. S1. Overlay of AFM–SMFS force–extension curves of CAPNS1 PEF homodimer at different pulling speeds.

Fig. S2. EGFP contour‐length increment did not change across pulling speeds.

Fig. S3. Partial unfolding intermediates preceding PEF–PEF rupture across pulling speeds.

Fig. S4. Distribution of prerupture unfolding lengths.

Fig. S5. Forces for partial unfolding intermediates versus terminal PEF–PEF rupture across pulling speeds.

Data accessibility

The data that support the findings of this study are available in the supplementary material of this article.