Pulling and gripping: Proteasomal translocation pathway elements differentially contribute to substrate unfolding

Abstract

Eukaryotic proteomes are regulated by the ubiquitin‐proteasome system. Unwanted proteins are tagged with ubiquitin, which is recognized by the 26S proteasome for degradation. To enter the proteasome's 20S core particle and be hydrolyzed into peptides, the substrate must be unfolded and threaded through the 19S regulatory particle. Tyrosine “aromatic paddles” of the regulatory particle's Rpt motor subunits unfold and translocate the substrate by pulling it toward the core particle. We previously showed that for substrates degraded from their N‐termini, polyglycine tracts inserted before a stable domain impair the proteasome's ability to grip or unfold the substrate at multiple points along the translocation pathway. Herein we compare unfolding from the N‐ and C‐termini of model substrates. We find that a combination of local substrate structural elements, grip sequence, and possibly fundamental asymmetry of the proteasomal unfolding machinery that favors N‐terminal unfolding affect the proteasome's ability to unfold substrates. Although the basic unfolding and translocation mechanism is conserved regardless of substrate orientation, we find differences in the ability of polyglycine tracts to disrupt unfolding from the N versus the C terminus. Finally, comparison of the rates of unfolding versus substrate release shows that different regions along the translocation pathway employ different mechanisms to facilitate unfolding and translocation. Interactions with aromatic paddles primarily increase the rate of substrate unfolding, while interactions predicted to take place with the core particle α‐ring N‐termini primarily decrease the rate of premature release of partially degraded substrates.

1. INTRODUCTION

Cells actively regulate their protein content to ensure the correct amount of each protein is present based on cellular needs. Potentially toxic misfolded or damaged proteins and proteins that are no longer required need to be degraded and recycled (Sauer & Baker, 2011). In eukaryotes, the ubiquitin‐proteasome system executes this function by attaching a tag, ubiquitin (Ub), to target proteins, signaling them for proteasome‐mediated degradation. This attachment usually involves conjugation of Ub to a surface‐exposed lysine on the target protein using a three‐step cascade mechanism made up of a ubiquitin‐activating enzyme (E1), a ubiquitin‐conjugating enzyme (E2) and a ubiquitin ligase (E3) (Clague et al., 2015). Since Ub itself contains seven lysines, additional Ubs can be added to a previously conjugated Ub, generating a polyubiquitin (polyUb) chain with a topology that is determined by the specific lysine or lysines used in the chain (i.e., K6, K11, K27, K29, K33, K48, or K63) (Akutsu et al., 2016).

Once a protein is ubiquitinated, polyUb is recognized by one or more of the Ub receptors on the 26S proteasome, Rpn1, Rpn10, and Rpn13. The ATPase motor proteins (Rpt1‐Rpt6) within the 19S Regulatory Particle (RP) engage a requisite unstructured region on the substrate. The substrate is unfolded and translocated through the RP channel as polyUb is removed by the deubiquitinase Rpn11. The channel leads to the 20S Core Particle (CP) proteolytic active sites where the substrate is hydrolyzed into peptide fragments. The unstructured initiation region can come from either end of the protein or from internal loops, thus allowing for N‐to‐C or C‐to‐N degradation (Arkinson et al., 2025).

The Rpt motor proteins contain pore loops, termed “aromatic paddles” because of a critical tyrosine residue, arranged in a spiral staircase lining the RP pore. These paddles engage the unstructured region at the top of the RP pore and move the substrate down toward the CP, each paddle disengaging when it reaches the bottom of the staircase and then returning to engage the substrate upstream (De La Peña et al., 2018). Numerous studies have shown that certain amino acid sequences preceding a stably folded domain on a substrate are able to impair the proteasome's ability to unfold this domain (Daskalogianni et al., 2008; Hoyt et al., ²⁰⁰⁶; Lin & Ghosh, ¹⁹⁹⁶; Ragwan et al., ²⁰²⁴; Tian et al., ²⁰⁰⁵). This impairment could lead to a desired outcome as in the case of the activation of the transcription factor NF‐κB. A C‐terminal inhibitor region of p105 is partially degraded when the proteasome encounters a glycine‐rich region positioned near a stably folded domain, which causes the release of a functional p50 subunit (Lin & Ghosh, 1996; Palombella et al., ¹⁹⁹⁴). Partial degradation could also lead to detrimental outcomes. When a substrate is partially degraded, its ubiquitin tag is often removed, thus eliminating the signal for the proteasome to re‐engage and degrade the substrate. Accumulation of partially degraded fragments can lead to aggregation and cellular toxicity, as seen in several neurodegenerative diseases (Le Guerroué & Youle, 2021).

We recently showed that the insertion of polyglycine (polyG) tracts preceding a stabilized dihydrofolate reductase (DHFR) domain impaired unfolding (Ragwan et al., 2024). We conducted a systematic approach of inserting 10 amino acid‐long glycine tracts at defined distances from the folded domain to generate an unfolding ability profile and confirmed that the biggest impacts on unfolding map onto the regions where the aromatic paddles interact with the substrate (Figure 1). Intriguingly, polyG insertions disrupted unfolding when positioned up to 67 amino acids away from DHFR, implicating much more distal proteasomal translocation elements as also important for unfolding. This prior work was conducted using substrates that were degraded in the N‐to‐C direction. Considering that both the incoming substrate peptide and the hexameric Rpt subunits that contain the aromatic paddles are asymmetrical, we first asked whether there would be differences in unfolding mechanisms based on the direction of substrate translocation.

FIGURE 1.

Model of a proteasome‐bound substrate undergoing N‐to‐C terminal degradation. The model illustrates the translocation pathway, highlighting the features on the proteasome that may contribute to unfolding and translocating the substrate. The colors on the substrate (spheres) correspond to the reduced unfolding ability due to the insertion of a 10 amino acid polyG at that region, with darker representing greater reduction. Blue and orange semi‐transparent surfaces are the regulatory and core particles, respectively, (PDB 6EF3), while yellow, purple and green surfaces are three Rpt subunits, seen in a cutaway view. The red surface is the hard‐to‐unfold DHFR domain (PDB 1DRE). Data from Ragwan et al. (2024).

Here, we find that, contrary to expectations based purely on substrate structure, the proteasome has more difficulty unfolding DHFR from the C‐terminus, consistent with increased proteasomal pulling strength from the N‐terminus. We also show that while there are similarities between the unfolding profile from the N‐ and C‐termini, there are key differences, further suggesting that the proteasome degrades substrates differently based on direction. Finally, we show that different proteasomal translocation pathway elements make distinct mechanistic contributions to processive unfolding and degradation.

2. RESULTS

2.1. Comparing degradation initiated from N‐ and C‐ termini

To compare degradation from the N‐ and C‐termini, we used two model unfolding ability substrates. R‐N2D‐ACTR‐Cys‐DHFR (N‐ACTR_cntrl), which begins with an N‐terminal unstructured region derived from the Neh2 domain of Nrf2 (R‐N2D) that contains the only lysines within the substrate, followed by a Cys‐containing linker for fluorescent labeling, a natively unfolded ACTR domain (Dogan et al., 2012), and finally, a hard‐to‐unfold DHFR domain (stabilized by the addition of the small molecule NADPH). This substrate is degraded from the N‐terminus and has been previously characterized (Ragwan et al., 2024). R‐Cys‐DHFR‐ACTR‐N2D (C‐ACTR_cntrl) is degraded from the C‐terminus, with all elements in the reverse order except for the Cys used for dye labeling and the Arg required for ubiquitination by the N‐end rule ligase Ubr1, which are retained at the N‐terminus (Figure 2a, Supplementary Table S1).

FIGURE 2.

Unfolding abilities of K48‐ and K63‐linked substrates degraded from the N‐terminus (a) Schematic depiction of degradation of unfolding ability substrates from either the N‐ or C‐terminus. The rate constant k _deg ^{full‐length} includes substrate binding, engagement, and degradation of the ACTR domain. The DHFR domain can then either be unfolded, translocated, and proteolyzed with a rate constant of k _deg ^frag, or irreversibly released with a rate constant k _rel ^frag. The unfolding ability U is defined as the ratio between degradation and release of the DHFR‐containing fragment, and can be measured by quantifying the extent of complete degradation versus fragment release. (b) Representative SDS‐PAGE gels (left) and quantification (right) of degradation of 20 nM N‐ACTR_cntrl with K48‐ (top), or K63‐ (bottom) linked ubiquitination by 25 nM 26S proteasome. 500 μM NADPH was added to stabilize DHFR. Non‐ubiquitinated substrates are shown in the first lane of the gel. Ubiquitinated substrate is shown in black, and fragment formation is shown in red. Individual data points are shown as dots, averages are solid symbols, error bars represent the SEM of 4 (K48) or 14 (K63) experiments, and fits are global fits to single exponentials. (c) U of K48‐linked (brown) and K63‐linked (purple) ubiquitinated substrates, calculated from data in b. Error bars represent the SEM propagated from curve fitting. (d) U values of K48‐linked (data from Ragwan et al. (2024)) and K63‐linked substrates plotted against distance of the 10gly sequence from the DHFR were normalized to their respective controls, K48‐N‐ACTR_cntrl and K63‐N‐ACTR_cntrl. Error bars represent the SEM of 4–14 experiments.

When the proteasome encounters a substrate, after binding, the proteasome engages at the unstructured terminus of the substrate and begins to degrade the ACTR domain, represented by the rate constant k _deg ^{full‐length} in Figure 2a (Hurley & Kraut, 2021). When the proteasome encounters the hard‐to‐unfold DHFR domain, two possible outcomes occur: (1) DHFR is unfolded, translocated into the CP and degraded with a rate constant k _deg ^frag (where unfolding is rate limiting). (2) The proteasome stalls on the stable DHFR domain, which is irreversibly released with a rate constant k _rel ^frag. The ratio of these outcomes, k _deg ^frag/k _rel ^frag, defines the unfolding ability, U (Equation (1) in Methods), which can be determined from the quantification of the amplitudes of full‐length substrate degraded and DHFR‐containing fragment formed over the course of reaction (Equation (2) in Methods).

C‐ACTR_cntrl was inefficiently ubiquitinated by Ubr1 (Supplementary Figure S1). We therefore ubiquitinated both C‐ACTR_cntrl and N‐ACTR_cntrl with Rsp5 (Supplementary Figures S2 and S3), which generates K63 linkages (Ziv et al., 2011), allowing us to compare both the effect of chain linkage on unfolding ability for degradation from the N‐terminus (K48 as in our previous work (Ragwan et al., 2024) versus K63 herein) and the effect of directionality of degradation.

In principle, unfolding ability should be independent of proteasome concentration, as it is based on proteasome‐bound substrate partitioning between degradation and release. However, we initially observed that higher proteasome concentrations led to increased levels of fragment formation and thus lower unfolding abilities. Additional experiments revealed that ATP present in the purified proteasome (which increased the total ATP concentration in the assay) was leading to these results (Supplementary Figure S4). Indeed, it has been shown that high concentrations of ATP can suppress the activity of the proteasome (Huang et al., 2010). Further control experiments indicated that we could rescue the unfolding defect by increasing [Mg²⁺] (Supplementary Figure S4). We therefore modified our proteasome purification buffers and degradation assay conditions to ensure that the total amount of ATP and Mg²⁺ stayed consistent (see Methods). Under these conditions, unfolding ability was essentially independent of proteasome concentration (Supplementary Figure S4).

2.2. K63 linkages reduce unfolding ability, but not the effect of polyG insertions

Consistent with previous reports in which K63‐linked chains were compared to K48‐containing mixed‐linkage chains (Reichard et al., 2016), U modestly decreased with K63‐linked chains versus K48‐linked chains for our N‐terminally degraded substrate (Figure 2b,c, p = 0.02). However, when we replaced 10 amino acids of ACTR with polyG at different distances from DHFR to generate an unfolding ability profile, the effects of polyG substitution were nearly identical to our previous K48‐linked profile (Figure 2d, Supplementary Table S2, Supplementary Figures S5 and S6), with the largest decrease in U when polyG was inserted 18–27 amino acids from DHFR, and with impairment of unfolding occurring out to 67 amino acids away from DHFR.

2.3. Unfolding abilities are higher for N‐terminal degradation, and reveal direction‐dependent differences in the unfolding ability profile

Having obtained a profile for N‐terminal degradation, we next assessed C‐terminal degradation. C‐ACTR_cntrl (Figure 3a) gave a lower unfolding ability (U) than N‐ACTR_cntrl (2.9 ± 0.3 vs. 9 ± 2, p = 0.008). There are at least three possible explanations for this result. First, the proteasome interacts with a different stretch of amino acids within ACTR while attempting to unfold DHFR from either the C‐ or N‐terminus, and might therefore be able to pull with more or less force. Second, DHFR may have a different intrinsic resistance to unfolding from the C‐ or N‐terminus based on the structural features being pulled on. Finally, there may be underlying differences in the ability of the proteasome to apply force when pulling in either the C‐ or the N‐terminal direction.

FIGURE 3.

Differences in the way polyG inserts impair unfolding ability based on direction of degradation. (a), (b) Representative SDS‐PAGE gels (left) and quantification (right) of degradation of C‐ACTR_cntrl (a) and the polyG substrate that produced the highest fragment formation from degradation assays done with K63‐linked ubiquitinated substrates, C‐ACTR_10gly53 (b). Non‐ubiquitinated substrates are shown in the first lane of the gel. Ubiquitinated substrate is shown in black, and fragment formation is shown in red. Individual data points are shown as dots, averages are solid symbols, and fits are global fits to single exponentials. (c) Unfolding abilities of C‐terminally degraded substrates. * indicates p< 0.05 relative to control. (d) Unfolding abilities from of N‐ and C‐ terminal degradation were normalized to their respective control substrates, N‐ACTR_cntrl and C‐ACTR_cntrl. For (c), (d), dashed lines represent control values. (e) Comparison of unfolding abilities of the control (ACTR_cntrl) and reverse (ACTR_rev) ACTR constructs in both N‐ and C‐terminal degradation. For (a)–(f) error bars represent SEM of between 3 and 14 experiments.

To address the first possibility, we generated two additional constructs (C‐ACTR_rev and N‐ACTR_rev) which reverse the amino acid sequence of ACTR. If the sequence interacting with the proteasome caused the lower unfolding ability from the C‐terminus, we would predict that C‐ACTR_rev would be degraded with a higher unfolding ability than C‐ACTR_cntrl while N‐ACTR_rev would be degraded with a lower unfolding ability than N‐ACTR_cntrl. However, we find that C‐ACTR_rev does not increase the unfolding ability of the proteasome relative to C‐ACTR_cntrl (U = 2.6 ± 0.4 vs. 2.9 ± 0.3), and although there is a decrease between N‐ACTR_cntrl and N‐ACTR_rev (U = 9 ± 2 vs. 5 ± 1) the effect is not statistically significant (p = 0.2) (Figure 3e). Thus, the sequence of the substrate regions in the translocation channel likely do affect the proteasome's unfolding ability, but are unlikely to account for all of the proteasome's greater ability to unfold from the N‐terminus.

To address the second possibility, we generated DHFR circular permutants which reconfigured DHFR to contain new N‐ and C‐termini that vary the local structure where pulling begins (Supplementary Figure S7). Table 1 details the permutants and the local structures that are initially pulled on for each variant, and a preferred direction of degradation proposed based on its structural features, computational experiments (Tian et al., 2015; Varikoti et al., ²⁰²²) and on similar studies done with both the proteasome and other proteases (Lee et al., 2001), where the general expectation is the easiest unfolding when initiating degradation from an α‐helix and the most difficult unfolding from internal ß‐strands (Lee et al., 2001). We then conducted degradation experiments initiated from both termini to observe which orientation was preferred. In all cases, either the direction was as predicted from structural features, or degradation was easier from the N‐terminus than predicted (Table 1, Supplementary Figures S8 and S9). Thus, while the local structure is likely the dominant factor determining whether a protein is unfolded or not, the proteasome may also have some bias toward unfolding from the N terminus.

TABLE 1.

Degradation of DHFR and DHFR circular permutants and local stability.

DHFR construct	Direction of degradation	Initial structural element pulled	Predicted preferred direction of degradation based on previous studies and/or structure	Observed preferred direction of degradation in this study
WT	N	Internal ß‐strand	C (Tian et al., 2015; Varikoti et al., 2022)	N
	C	Edge ß‐strand
CP38	N	Internal ß‐strand	C (Lee et al., 2001)	C
	C	α‐helix
CP70	N	Edge ß‐strand	?	N
	C	Coil leading to internal ß‐strand
CP145	N	Internal ß‐strand	C	No preference observed
	C	Edge of ß‐strand

Next, we inserted 10‐amino acid long polyG tracts at varying distances from DHFR to generate a C‐terminal unfolding ability profile (Figure 3c, Supplementary Table S3, Supplementary Figures S10 and S11). This profile contained two minima: polyG insertions 13–27 and 53–62 amino acids away from DHFR significantly impaired unfolding, while other regions of the profile were not significantly different from the control (Figure 3c, Supplementary Table S5). In contrast, for N‐terminal degradation, polyG insertions up to 58 amino acids away from DHFR all caused a significant reduction in unfolding ability (Figure 3d, Supplementary Tables S2 and S4). Moreover, N‐terminal degradation appears to be more sensitive to polyG insertions. When compared to their respective controls, the largest effect on unfolding with N‐terminal degradation was an ~8‐fold reduction in U (N‐ACTR_cntrl = 9 ± 2 vs. N‐ACTR_10gly18 = 1.2 ± 0.2) in contrast to an ~3‐fold effect with C‐terminal degradation (C‐ACTR_cntrl = 2.9 ± 0.3 vs. C‐ACTR_10gly18 = 1.0 ± 0.2). Taken together, these findings suggest differences in substrate processing based on direction. From both directions, larger effects are seen in the region predicted to correspond to aromatic paddles (~18 amino acids from DHFR) and ß‐subunits (53–58 amino acids from DHFR) (Ragwan et al., 2024), suggesting polyG impairs some of the same translocation mechanisms regardless of direction.

2.4. Varying proteasome regions within the translocation pathway contribute to unfolding via different mechanisms

We next turned to kinetic modeling to gain mechanistic insight into the effects of directionality of degradation and polyG insertions. Reductions in U can come from either a decreased DHFR unfolding rate or an increased rate of DHFR‐containing fragment release. By varying the concentration of NADPH and thus the extent to which DHFR was stabilized, we were able to find conditions where DHFR‐containing fragment would transiently accumulate before being degraded or released (Figure 4a, Supplementary Figure S12). Globally fitting results with multiple substrates, proteasome concentrations and NADPH concentrations to a model derived from Figure 2a then allowed the direct determination of k _deg ^frag and k _rel ^frag (Kraut et al., 2012). We collected datasets for three constructs from each direction: control substrates (N‐ACTR_cntrl and C‐ACTR_cntrl), substrates with polyG located near the proteasome's aromatic paddles (N‐ACTR_10gly18 and C‐ACTR_10gly18, predicted to interact with the aromatic paddles), and substrates that position polyG further into the translocation pathway that impaired unfolding (N‐ACTR_10gly38 and C‐ACTR_10gly53). We first confirmed that there was good agreement between U calculated using kinetic modeling (Equation (1) in Methods) (Ragwan et al., 2024) and from endpoints of degradation assays (Equation (2) in Methods), validating the rate constants determined by kinetic modeling (Table 2).

FIGURE 4.

Kinetic modeling reveals mechanism of unfolding ability reduction by polyG stretches. (a) Representative degradation experiments with N‐ACTR_10gly18 using 25, 100, 200 and 500 μM NADPH, monitoring the disappearance of substrate (closed circles) and accumulation of fragment (open circles) at various time points. Lines are global fits to the kinetic model in Figure 2a (as described in Methods), error bars represent the SEM of 3–4 experiments. (b) Unfolding ability (U) at 500 μM NADPH calculated from kinetic modeling (closed symbols and from the quantification of end points of single‐exponential fits of full‐length and fragment amplitudes (open circles). N‐terminal and C‐terminal degradation are shown as squares and circles, respectively. Error bars represent SEM propagated from curve fitting errors. (c)‐(d) Relative effects on k _deg ^frag and k _rel ^frag caused by polyG insertion, calculated by normalizing each rate to the rate of the control for (c) N‐terminal and (d) C‐terminal degradation. Error bars represent SEM propagated from curve fitting errors.

TABLE 2.

Rates obtained from kinetic modeling.

	k deg frag (min−1)	k rel frag (min−1)	U model	U endpoints
N‐ACTRcntrl	1.11 ± 0.04	0.19 ± 0.01	5.7 ± 0.4	9 ± 2
N‐ACTR10gly18	0.16 ± 0.01	0.15 ± 0.01	1.1 ± 0.1	1.2 ± 0.2
N‐ACTR10gly38	1.7 ± 0.3	0.49 ± 0.09	3.5 ± 0.9	4.1 ± 0.6
C‐ACTRcntrl	0.71 ± 0.09	0.26 ± 0.03	2.8 ± 0.5	2.9 ± 0.3
C‐ACTR10gly18	0.19 ± 0.03	0.14 ± 0.02	1.3 ± 0.3	1.2 ± 0.3
C‐ACTR10gly53	0.22 ± 0.05	0.27 ± 0.06	0.8 ± 0.2	1.0 ± 0.2

With N‐terminal degradation, polyG insertions predicted to interact with the aromatic paddles (N‐ACTR_10gly18) primarily reduced the observed unfolding rate rather than affecting the release rate (Figure 4c, Supplementary Tables S6–S9), in agreement with our previous work (Kraut et al., 2012). In contrast, for polyG inserts at a position predicted to interact with the CP α ring N‐termini (Ragwan et al., 2024) (N‐ACTR_10gly38) the opposite was true; i.e. U was diminished mostly due to an increase in k _frag ^rel (Figure 4c, Supplementary Tables S6–S9). These data suggest that two different mechanisms, corresponding to two different regions within the translocation pathway, are driving polyG's impairment of U. Similarly, with C‐terminal degradation, although the effect of polyG insertion was smaller, making it harder to draw strong conclusions, insertions at the aromatic paddles significantly decreased unfolding, with little effect on release. There was no observed U effect at positions predicted to interact with the CP α ring N‐termini, and at the region predicted to interact with the CP ß ring, the observed defect is again driven by reduced unfolding.

3. DISCUSSION

In the experiments outlined above, we compared the proteasome's unfolding ability with substrates initiated from the N versus the C terminus. We confirmed that processivity in both directions relies on multiple unfolding elements within the translocation pathway as outlined in Figure 1, likely the aromatic paddles of the RP and α and β subunits of the CP. Despite this conservation of overall mechanism, there are noteworthy differences. When using our model substrates, the proteasome's unfolding ability was higher with N‐to‐C versus C‐to‐N‐terminal degradation. Substrate sequence differences that arise when altering the direction of degradation may account for some, but likely not all, of this difference. Why might this asymmetry exist?

As previously noted, differences in directionality could be attributed to the architecture of the substrate and/or inherent properties of the proteasome. Looking at the substrate, the DHFR domain may unfold differently when pulled on from the N‐ versus C‐terminus due to differences in DHFR's local stability. Previous experiments with bacterial ATP‐dependent proteases indicate that local structures (α helix versus ß sheet) are important for unfolding and that sequences at the surface of a folded structure might be easier to “peel away” than buried sequences (Lee et al., 2001). Indeed, for selected proteasome substrates degraded from either terminus, the force needed to unfold from either terminus in directed molecular dynamics simulations could predict from which direction the substrate was degraded faster (Berko et al., 2012).

The Escherichia coli DHFR domain in our substrates appears to have a more accessible C‐terminus (Figure 5a) that might be easier to unfold. If unfolding occurs from the C‐terminus, the first β‐strand could be more easily peeled away from the body of DHFR and promote efficient global unfolding of the substrate. In contrast, the N‐terminus is noticeably buried, which would require a more dramatic unfolding of the overall structure to access, implying greater resistance to initial unfolding. This structural observation agrees with a study using Monte Carlo unfolding simulations of apo‐DHFR, which predicted that the C‐terminal β‐hairpin separates from the rest of the protein early in the unfolding process (Tian et al., 2015), and with molecular simulations of directional unfolding of DHFR, which predict that greater force is required to unfold DHFR from the N‐terminus than the C‐terminus (Varikoti et al., 2022).

FIGURE 5.

Substrate and proteasome properties that might influence unfolding differently based direction of degradation. (a) X‐ray crystal structure of Escherichia coli DHFR (PDB 1DRE) shown in cartoon representation with the N and C termini depicted in red and blue, respectively. The C‐terminus is located close to the outer surface, so could be “peeled” away from the rest of the protein to start the unfolding and degradation process. The N‐terminus reaches into the center of DHFR, which suggests that if pulled from this terminus, global unfolding of DHFR would be required to initiate degradation. (b) Cryo‐EM structure (PDB 6EF3) of yeast 26S proteasome highlighting the asymmetry of the aromatic paddles within the heterohexameric Rpt subunits. Each Rpt subunit is colored a different shade, and its corresponding aromatic paddle is shown in spheres in the same color. A side view shows the spiral architecture of the translocation pathway. The top view shows the orientations of the aromatic paddles. (c) Substrates at the region of the aromatic paddles depicting C‐to‐N degradation (left, orange = substrate, gray = aromatic paddles, PDB 6EF3) and N‐to‐C degradation (right, green = substrate, gray = aromatic paddles, PDB 6MSE). Substrate residues were mutated to poly‐alanine to aid in comparison. The vertical dashed line shows the orientation of a linear portion of the peptide in the RP pore. Horizontal lines positioned over the α‐carbons and carbonyl carbons are drawn to illustrate the orientations of sidechains and carbonyl oxygens. Together, these show how the RP might interact differently with N‐to‐C versus C‐to‐N substrates.

Based on these observations, we would predict that degradation from the C‐terminus would give a higher unfolding ability than from the N‐terminus. However, our results show the opposite. In addition, we examined the directional degradation of several circular permutants of DHFR. Lee et al. showed that cp38DHFR could be stabilized against degradation by either ClpAP or mammalian proteasome from the N‐terminus (where degradation proceeds from an internal ß strand), but could not be fully stabilized against degradation by ClpXP from the C‐terminus (where degradation proceeds from an α helix) (Lee et al., 2001). In agreement with these results, we found the proteasome could efficiently degrade methotrexate‐stabilized DHFR from the C but not the N‐terminus.

Lee et al. also showed that cp70DHFR (where degradation begins from an edge ß strand from the N‐terminus and short coil followed by an internal ß strand from the C‐terminus) was largely protected from degradation by ClpAP, but did not test with the proteasome (Lee et al., 2001). We reasoned that the C‐terminus might be easier to degrade because the first structural element encountered was random coil, although the next element (if unfolding this coil did not fully unfold DHFR) was a buried strand. However, we found that methotrexate‐stabilized DHFR was highly resistant to degradation from the C‐terminus, but was completely degraded from the N‐terminus, again suggesting a possible increased strength from the N‐terminus. Finally, we examined cp145DHFR, which structurally we would predict would be easier to unfold from the C‐terminus (edge strand) than the N‐terminus (internal strand). The proteasome was able to fully degrade methotrexate‐stabilized cp145DHFR from either direction (despite similar thermodynamic stability to cp38 and cp70 (Arai et al., 2003; Iwakura et al., ²⁰⁰⁰)), indicating that the local structure alone is also not sufficient to fully predict stability. Taken together, the overall better than expected degradation we observe from the N‐terminus suggests (although by no means definitively proves) that there might also be features inherent to the proteasome that contribute to stronger unfolding from the N‐terminus.

Where might such a direction preference come from? It is possible that the Rpt subunit aromatic paddles have a better hold on the substrate in one direction versus the other, and thus are better able to apply an unfolding force. Several factors might contribute to this asymmetry. First, the environment around the substrate at the aromatic paddles has intrinsic handedness. The Rpt subunits are sequentially tilted in a clockwise manner to form a staircase configuration (Figure 5b) that couples translocation movement to proteolysis (Bard et al., 2018). Second, these aromatic paddles sterically pull on the substrate by interleaving between side chains and backbone groups and the orientations of the side chains and backbone carbonyls suggest directional differences. Both the translocation channel and the substrate are asymmetrical, so it stands to reason that pulling interactions might be different when a substrate is degraded from the N‐ versus the C‐terminus. Indeed, comparison of substrates oriented in the N‐ and C‐ terminal directions reveals differences in the angles of side chains and carbonyl oxygens relative to both the N and C‐termini of the substrate and to the tyrosines that presumably sterically interact with the substrate to drive translocation (Figure 5c). Although there is no specific interaction that is obviously structurally superior in one orientation versus the other, we hypothesize that the sum of multiple asymmetric interactions might drive more efficient force generation for substrates that are unfolded from the N‐terminus. To truly address this hypothesis, unnatural substrates where the same feature is presented to the proteasome with only the direction of degradation initiation different would be required.

Analysis of the kinetic contributions to U suggests that different features of the proteasome along the translocation pathway contribute differentially to proteasomal processivity (Figure 6). In agreement with previous studies (Kraut et al., 2012), polyG insertions that are situated near the aromatic paddles lower the unfolding ability or processivity principally by slowing the rate of unfolding, presumably by reducing the ability of the proteasome to apply an unfolding force to the substrate while interacting with featureless glycine residues. In contrast, we observe that polyG, when positioned further downstream of the aromatic paddles, impairs processivity by increasing the rate at which partially degraded proteins are released from the proteasome. What could explain these differences? We hypothesize that the α subunits of the CP are important for maintaining grip on the substrate, and that when polyG insertions interfere with this interaction, a partially degraded fragment is more susceptible to release by slipping out of the translocation channel. Experiments with open‐channel proteasome mutants could help to further validate this finding. Finally, the reduction in C‐terminal unfolding when polyG is inserted 53 amino acids away from DHFR might be caused by polyG interacting with the ß subunit active sites. This is in agreement with our previous study (Ragwan et al., 2024), where we observed that with N‐terminal degradation, polyG insertions between 58 and 67 amino acids away from DHFR resulted in impaired unfolding, suggesting that sequences that map onto the ß subunit active sites also influence unfolding.

FIGURE 6.

Features within the proteasomal translocation pathway contribute to unfolding ability using different mechanisms. Model of the proteasomal translocation pathway based on cryo‐EM structures (Ragwan et al., 2024). In black (ball‐and‐stick, representation) are substrate residues within the channel, which are attached to DHFR (red, surface representation) at the entrance of the channel. Residues on the proteasome within 5 Å of the model substrate at the aromatic paddle region of the 19S RP (orange) and α subunit region of the 20S CP (gray) are depicted in surface view. PolyG tracts inserted at the aromatic paddles reduce k _deg ^frag , which is the rate that describes unfolding and translocation. In contrast, polyG tracts inserted at the α subunits primarily increases k _frag ^rel, which is the rate that describes fragment release, so are implicated in gripping or holding the substrate in place. The model was developed using structures of the proteasome from PDB 6EF0, 6EF1, 6EF2, and 6EF3 (De La Peña et al., 2018), and of DHFR from 1DRE (Sawaya & Kraut, 1997).

4. MATERIALS AND METHODS

4.1. Substrate design

Starting with a previously constructed plasmid, His‐SUMO‐R‐N2D‐ACTR‐C‐DHFR (Ragwan et al., 2024) (herein N‐ACTR_cntrl), a C‐terminally degraded control substrate, C‐ACTR_cntrl, was generated in which the order of domains was reconfigured using PCR and Gibson Assembly to encode His‐SUMO‐R‐C‐DHFR‐ACTR‐N2D. This construct included a His‐SUMO tag for purification, a short linker containing a unique cysteine for labeling, a hard‐to‐unfold DHFR domain, an easy‐to‐unfold ACTR domain, and an N‐terminal unstructured region derived from the Neh2 domain of Nrf2 (R‐N2D) which contains the only lysines within the substrate. As described previously (Ragwan et al., 2024), 10 amino acid segments of the ACTR domain were replaced with glycine residues using synthetic dsDNA (IDT or Twist Biosciences) and restriction cloning. The mutant ACTR domain sequences were amplified by PCR, and inserted into BamHI and XhoI (N‐terminal substrates) or BamHI and NheI (C‐terminal substrates) restriction sites of the control substrate, replacing the native ACTR. Site‐directed mutagenesis was used to generate C‐ACTR_10gly1 from C‐ACTR_10gly3 by removing the two‐residue linker between DHFR and ACTR. The number following 10gly represents the distance between the 10 glycine repeat and DHFR and is defined as the amino acid position that the glycine tract begins. Substrates with the sequence of ACTR reversed (ACTR_rev) were also generated. N‐terminally degraded substrates were described previously (Ragwan et al., 2024), with the exception of N‐ACTR_10gly1, which was made by PCR‐directed mutagenesis of the Cys linker between DHFR and ACTR and contains 9 glycines along with the Cys used for labeling. All sequences are presented in Supplementary Table S1.

Circular permutants of DHFR were synthesized as double‐stranded DNA (Twist Biosciences) with a 5‐gly linker between the N‐ and C‐terminus of wildtype DHFR and a “break” in the original sequence at the position indicated and inserted using restriction cloning. For example, N‐ACTR‐DHFR_CP38 contains His‐SUMO‐R‐N2D‐ACTR‐C‐DHFR_38‐159‐GGGGG‐DHFR_1‐37.

4.2. Protein expression, purification, labeling, and ubiquitination

Protein substrates were prepared as previously described (Hurley & Kraut, 2021; Ragwan et al., ²⁰²⁴). Proteins were overexpressed in BL21*(DE3) cells by autoinduction, purified by NiNTA chromatography, and the SUMO tag was then cleaved with Ulp1 protease and removed by subtractive purification on NiNTA. Substrates were incubated with sulfo‐cyanine5 maleimide (Cy5) (Lumiprobe), labeling the single cysteine, and separated from free dye on a Superdex S200 column (Cytiva). Labeled substrates were ubiquitinated with the UbcH7/Rsp5 ubiquitination system, producing K63‐linked chains conjugated to the degron region of the substrate as described previously (Hurley & Kraut, 2021). Ubiquitinated substrates were purified using spin size‐exclusion chromatography in assay buffer.

PI31/Fub1 inhibitor was cloned into a pE‐SUMO vector and purified by NiNTA chromatography similarly to the substrates, but following removal of the SUMO tag purified protein was buffer‐exchanged into 50 mM Tris–HCl (pH 7.5), 10% glycerol.

4.3. Proteasome purification

26S Proteasome was from the YYS40 strain, and was initially expressed and purified using a 3x‐FLAG tag on the Rpn11 subunit of the RP as previously described (Hurley & Kraut, 2021; Ragwan et al., ²⁰²⁴). This proteasome was used for initial experiments and DHFR circular permutant comparisons. 20S proteasome was similarly purified from strain yMDC11 using a 3×‐FLAG tag on the Pre11 subunit of the CP, with the addition of a 500 mM NaCl wash and without the use of ATP.

For all other experiments, the prep was modified to use an elution buffer that more closely matched the degradation assay buffer (100 μM 3X‐Flag peptide, 50 mM Tris‐Cl pH 7.5, 5 mM MgCl₂, 2.5% glycerol, 2 mM DTT, and 1 mM ATP) to keep the degradation assay buffer consistent when varying the amount of proteasome between experiments.

4.4. Degradation assays

Degradation experiments on DHFR circular permutants were done as described previously (Bragança & Kraut, 2020). 25 nM proteasome was incubated with 20 nM substrate at 30°C in 50 mM TrisCl, 5 mM added MgCl₂, 5% (v/v) glycerol, 1 mM added ATP, 10 mM creatine phosphate, 0.1 mg/mL creatine kinase, 500 μM NADPH, 100 μM methotrexate, 1 mg/mL BSA, 0.1% Tween‐20, and 1% DMSO, pH 7.5. All other experiments were done similarly, but lacked methotrexate and were supplemented with additional MgCl₂ and ATP such that the final concentration was 20 mM MgCl₂ and 2 mM ATP (including the concentration of ATP and MgCl₂ coming from proteasome), and the glycerol concentration was lowered to 2.5% (v/v).

We included 1 μM yeast Fub1/PI31 (Chu‐Ping et al., 1992; McCutchen‐Maloney et al., ²⁰⁰⁰; Rawson et al., ²⁰²²), a 20S proteasome inhibitor, to prevent clipping by free 20S, which efficiently inhibited 20S but had no effect on 26 peptidase activity (Supplementary Figure S13).

Time points of the assay were taken and quenched in SDS‐PAGE sample buffer. SDS‐PAGE gels were imaged on a Typhoon FLA 9500 using Cy5 fluorescence and quantification of fluorescence intensity was done using ImageQuant (Cytiva). At each time point, the fluorescence corresponding to full‐length and DHFR fragment band intensities was quantified and plotted as a percentage of the full‐length substrate at the start of the reaction (10 s time point). Experiments at high proteasome concentration resulted in substantial degradation at the 10 s time point. For these experiments, a zero time point was obtained before the start of the reaction. When appropriate, data were globally fit to single exponentials, and an unfolding ability was determined using Equation (2).

4.5. Kinetic modeling

Kinetic modeling was carried out using COPASI software (Hoops et al., 2006) to the scheme shown in Figure 2a with several modifications. N‐ and C‐terminally targeted substrates were modeled separately, but within each, all data were globally fit. Initial substrate binding to the proteasome was modeled with a forward rate that was constrained to be the same for all substrates and a reverse rate fixed at 1.5 min⁻¹. An additional step (labeled DUB in the model; see Supplementary Information) was added to account for the fraction of full‐length protein not successfully engaged by the proteasome as described previously (Kraut et al., 2012). k _deg ^FL and k _deg ^frag (degstep1 and degstep2 in the model output; see Supplementary Information) were allowed to vary by both substrate and NADPH concentration. k _rel ^frag was allowed to vary by substrate. All other parameters were constrained to be the same for all substrate and NADPH concentrations. Unfolding abilities were calculated using Equation (1) and are given at 500 μM NADPH.

$$U=\frac{k_{\mathrm{deg}}^{\text{frag}}}{k_{\mathrm{rel}}^{\text{frag}}}$$ (1)

$$U=\frac{\text{amplitude full length disappearance}-\text{amplitude fragment formation}}{\text{amplitude fragment formation}}$$ (2)

4.6. Peptidase assays

2 nM 20S proteasome, with and without 1 μM PI31/Fub1, was mixed with 50 μM substrate (Suc‐LLVY‐AMC) in a buffer consisting of 50 mM Tris‐Cl pH 7.5, 5 mM MgCl₂, 2.5% glycerol, 2 mM DTT, and 1% DMSO. The appearance of AMC was monitored using excitation at 350 nm and emission at 445 nm on a FluoroMax® Plus‐C spectrofluorometer, and the linear portion of the curve was fit to a line to determine rates of reaction. The experiment was repeated with 26S proteasome in the presence of 1 mM ATP, 10 mM creatine phosphate, and 0.1 mg/mL creatine kinase.

4.7. Statistical analysis

When making comparisons between two substrates a student t‐test was used where a p‐value below 0.05 was considered significant. When making comparisons between more than two substrates, 1‐way ANOVA analysis was conducted (One‐way Analysis of Variance, n.d.).

4.8. Structural figures

Cryo‐EM and x‐ray images were generated using PyMol (The PyMOL Molecular Graphics System, Version 3.0 Schrödinger, LLC., n.d.) and ChimeraX (Meng et al., 2023).

AUTHOR CONTRIBUTIONS

Edwin R. Ragwan: Writing – original draft; conceptualization; investigation; writing – review and editing; formal analysis; methodology; data curation. Kristi M. Pham: Investigation; writing – review and editing. Arushi Palta: Investigation; writing – review and editing. Destiny N. McWilliams: Investigation; methodology; writing – review and editing. Daniel A. Kraut: Conceptualization; funding acquisition; writing – review and editing; supervision.

Supporting information

Data S1: Supplementary data containing substrate sequences, unfolding ability values, SDS‐PAGE gels, degradation assay graphs, kinetic modeling parameters, and statistical analysis are provided.

Supplementary Table S1. Design and amino acid sequences of the substrate constructs.

Supplementary Table S2. Similar impact on unfolding ability of PolyG tracts observed between K48‐ and K63‐linked substrates with N‐terminal degradation.

Supplementary Table S3. Unfolding ability values of K63‐linked substrates containing polyG tracts with C‐terminal degradation.

Supplementary Table S4. 1‐way ANOVA comparison of unfolding abilities of K63‐ubiquitinated N‐terminal substrates as outlined in the Methods.

Supplementary Table S5. 1‐way ANOVA comparison of unfolding abilities of K63‐ubiquitinated C‐terminal substrates as outlined in the Methods.

Supplementary Table S6. 1‐way ANOVA comparison of k _deg ^frag rates obtained from kinetic modeling of K63‐ubiquitinated C‐terminal substrates as outlined in the Methods.

Supplementary Table S7. 1‐way ANOVA comparison of k _rel ^frag rates obtained from kinetic modeling of K63‐ubiquitinated C‐terminal substrates as outlined in the Methods.

Supplementary Table S8. 1‐way ANOVA comparison of k _deg ^frag rates obtained from kinetic modeling of K63‐ubiquitinated N‐terminal substrates as outlined in the Methods.

Supplementary Table S9. 1‐way ANOVA comparison of k _rel ^frag rates obtained from kinetic modeling of K63‐ubiquitinated N‐terminal substrates as outlined in the Methods.

Supplementary Figure S1. Rsp5 ligase improves ubiquitination of substrates to be degraded from the C‐terminus. Representative SDS‐PAGE gel showing the substrate, C‐ACTR_cntrl, was ubiquitinated with (a) K48‐linked chains using Ubc2 (E2) and Ubr1 (E3) and (b) K63‐linkages using UbcH7 (E2) and Rsp5 (E3). Gels show time zero (0), after 4 h (4), and after purification of the ubiquitinated substrate by spin‐size exclusion chromatography (P).

Supplementary Figure S2. Ubiquitination of N‐to‐C degraded substrates with K63‐linkages. Representative SDS‐PAGE gels showing ubiquitination with K63‐linked chains using UbcH7 (E2) and Rsp5 (E3). Gels show time zero (0), after 2 h (2), and after purification of the ubiquitinated substrate by spin‐size exclusion chromatography (P).

Supplementary Figure S3. Ubiquitination of C‐to‐N degraded substrates with K63‐linkages. Representative SDS‐PAGE gels showing ubiquitination with K63‐linked chains using UbcH7 (E2) and Rsp5 (E3). Gels show time zero (0), after 4 h (4), and after purification of the ubiquitinated substrate by spin‐size exclusion chromatography (P).

Supplementary Figure S4. Degradation is sensitive to Mg²⁺ and ATP but not proteasome concentrations. Representative controls showing the effect of Mg²⁺, ATP and proteasome concentration. (a) Increasing the ATP concentration from 1.2 to 2.6 mM inhibits degradation and decreases the overall unfolding ability of N‐ACTR_cntrl. (b) Increasing the Mg²⁺ concentration from 10 to 20 mM, recovers the diminishing effect of high ATP concentrations. (c) Unfolding abilities calculated from the amplitude of single‐exponential fits. For (a)–(c), 25 nM proteasome was used. (d) Experiments with C‐ (left) and N‐ (right) terminal degradation with increasing concentrations of proteasome, from 25 to 100 mM. Results show similar endpoints for the percents of full‐length and fragment remaining. Curves were fit to single or double exponentials, as appropriate. Unfolding abilities were calculated from the amplitude of single‐exponential fits or the endpoint of double‐exponential fits. Experiments were conducted as outlined in Figure 2 of the main text, and errors represent SEM of 4–14 experiments.

Supplementary Figure S5. Representative gel images of K63‐ubiquitinated polyglycine‐containing substrates degraded from the N‐terminus. Experiments were conducted as outlined in Figure 2 of the main text, with 25 nM proteasome and 20 nM substrate. Non‐ubiquitinated substrate is shown in the first lane of each gel as a size reference. For the first gel, disappearance of ubiquitinated substrate is shown in black, with DHFR‐containing fragment in red. Substrates include N‐ACTR_cntrl, N‐ACTR_rev, and substrates with 10 aa long polyG tracts at varying distances from DHFR.

Supplementary Figure S6. Degradation of polyglycine substrates from the N‐terminus. Experiments were conducted as outlined in Figure 2 of the main text, with 25 nM proteasome and 20 nM substrate. Individual data points are shown as dots, averages are solid symbols, error bars represent the SEM of 4–10 experiments (example gels given in Supplemental Figure S5), and fits are global fits to single exponentials.

Supplementary Figure S7. AlphaFold structures of N‐ and C‐ termini of DHFR circular permutants. AlphaFold 3 (Abramson et al., 2024) was used to predict the structures of DHFR circular permutants. Structures are depicted in cartoon representation with the N‐ and C‐ termini colored orange and cyan, respectively.

Supplementary Figure S8. Representative gel images of DHFR circular permutants degraded from N‐ and C‐ termini. Experiments were conducted as outlined in Figure 2 of the main text, except that reactions were supplemented with 100 μM methotrexate to further stabilize DHFR. Non‐ubiquitinated substrate is shown in the first lane of each gel as a size reference. Disappearance of ubiquitinated substrate is shown in black, with DHFR‐containing fragment in red.

Supplementary Figure S9. Quantification of degradation of circularly permuted substrates. Experiments were conducted as outlined in Figure 2 of the main text, except that reactions were supplemented with 100 μM methotrexate to further stabilize DHFR. Individual data points are shown as dots, averages are solid symbols, error bars represent the SEM of 4 experiments, and fits are global fits to singe exponentials. (a)–(c) are experiments done with DHFR permutants CP38, CP70, and CP145, respectively, with degradation initiated from the N‐ (left) and C‐ (right) terminus.

Supplementary Figure S10. Representative gel images of K63‐ubiquitinated polyglycine‐containing substrates degraded from the C‐terminus. Experiments were conducted as outlined in Figure 2 of the main text, with 25 nM proteasome and 20 nM substrate. Non‐ubiquitinated substrate is shown in the first lane of each gel as a size reference. For the first gel, disappearance of ubiquitinated substrate is shown in black, with DHFR‐containing fragment in red. Substrates include C‐ACTR_cntrl, C‐ACTR_rev, and substrates with 10 aa long polyG tracts at varying distances from DHFR.

Supplementary Figure S11. Degradation of polyglycine substrates from the C‐terminus. Experiments were conducted as outlined in Figure 2 of the main text, with 25 nM proteasome and 20 nM substrate. Individual data points are shown as dots, averages are solid symbols, error bars represent the SEM of 4–8 experiments (example gels given in Supplemental Figure S8), and fits are global fits to single exponentials.

Supplementary Figure S12. Kinetic modeling. Degradation experiments using varying concentrations of NADPH and proteasome, monitoring the disappearance of substrate (open circles) and accumulation of fragment (crosses) at various time points from individual experiments. Lines are global fits to all the data sets from either the N‐ or the C‐terminus to the kinetic model in Figure 2a. (a), (b) Substrates degraded from the N‐terminus (a) and C‐terminus (b) using 25‐100 nM proteasome and 25‐500 μM NADPH.

Supplementary Figure S13. Fub1/P131 inhibits DHFR fragment clipping and 20S activity but not 26S activity. (a) Representative SDS‐PAGE gels of degradation assays on control substrates carried out with (bottom) and without (top) Fub1 inhibitor. On the right, a doublet of fragment bands is observed for C‐ACTR_cntrl (orange box) in the absence of Fub1, while a single band is seen when Fub1 is included in the assay (green box). This extra band is not present with N‐ACTR_cntrl with or without Fub1 (left). (b) Peptidase activity assay, monitoring the proteolysis of SucLLVY‐AMC. Assays were done with 2 nM 20S and 26S proteasome, with and without 1 μM Fub1, showing that only 20S is inhibited. The average rates were normalized to the rate obtained from 20S activity. Experiments were done in triplicate, and errors represent the SEM.

Ragwan ER, Pham KM, Palta A, McWilliams DN, Kraut DA. Pulling and gripping: Proteasomal translocation pathway elements differentially contribute to substrate unfolding. Protein Science. 2026;35(11):e70814. 10.1002/pro.70814

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in a public LabArchives repository with a doi of 10.25833/7564-5y29.