Conformational Dynamics Modulate Activation of the Ubiquitin Conjugating Enzyme Ube2g2.

The ubiquitin conjugating enzyme Ube2g2 together with its cognate E3 ligase gp78 catalyzes the synthesis of lysine-48 polyubiquitin chains constituting signals for the proteasomal degradation of misfolded proteins in the endoplasmic reticulum. Here, we employ NMR spectroscopy in combination with single-turnover diubiquitin formation assays to examine the role of the RING domain from gp78 in the catalytic activation of Ube2g2∼Ub conjugates. We find that approximately 60% of the Ube2g2∼Ub conjugates occupy a closed conformation in the absence of gp78-RING, with the population increasing to 82% upon gp78-RING binding. As expected, strong mutations in the hydrophobic patch residues of the ∼Ub moiety result in Ube2g2∼Ub populating only open states with corresponding loss of the ubiquitin conjugation activity. Less disruptive mutations introduced into the hydrophobic patch of the ∼Ub moiety also destabilize the closed conformational state, yet the corresponding effect on the ubiquitin conjugation activity ranges from complete loss to an enhancement of the catalytic activity. These results present a picture in which Ube2g2's active site is in a state of continual dynamic flux with the organization of the active site into a catalytically viable conformation constituting the rate-limiting step for a single ubiquitin ligation event. Ube2g2's function as a highly specific K48-polyubiquitin chain elongator leads us to speculate that this may be a strategy by which Ube2g2 reduces the probability of nonproductive catalytic outcomes in the absence of available substrate.

Introduction

Polyubiquitylation is a major post-translational mechanism by which eukaryotic cells respond to external stresses and stimuli. The nature of the response varies widely depending on the site of substrate ubiquitylation as well as the specific construction of the polyubiquitin signal.[1−3] Well-established examples include the involvement of lysine-63 (K63)-linked polyubiquitin chains in mediating DNA repair and endocytosis and the role of K48- and K11-linked polyubiquitin chains in the targeting of substrates for proteasomal degradation.[4,5] Proper function of the ubiquitylation system is an important bulwark against disease progression.[6,7]

The construction of the appropriate polyubiquitin chain and its attachment to the proper substrate is carried out by a cascade of enzymatic reactions. This process begins with the C-terminal adenylation of ubiquitin by an ubiquitin-activating enzyme (E1) followed by the formation of a thioester linkage to the E1 enzyme. This activated ubiquitin molecule is then transferred to the active site cysteine of one of the several dozen ubiquitin-conjugating enzymes (E2s) again making a thioester linkage. The charged E2 (denoted E2∼Ub) then typically goes on to promote the catalysis of an isopeptide bond between the C-terminal carboxylate of the ∼Ub moiety and the ε-amino group of a lysine residue residing on either substrate protein or another ubiquitin molecule. This step is carried out in conjunction with a ubiquitin ligase (E3), with the outcome depending on the identity of the involved E2 and E3 enzymes. A large majority of E3 ligases contain a so-called RING domain, which is capable of binding different E2 enzymes by means of a conserved interface. Given that there are many hundreds of different E3 enzymes, a major question is what factors govern the collaboration of the proper E2 and E3 enzymes to produce the desired outcome.

In addition to its role in the recruitment of the E2∼Ub conjugate, RING-E3s also strongly activate E2∼Ub reactivity in a manner that is not completely understood.[8−11] It is now generally recognized that many E2∼Ubs must be capable of assuming a closed conformation in which the hydrophobic patch of the ∼Ub moiety interacts with the E2 at an interface centered on its α2 helix. For a number of different E2s, including Ubc1, Ube2e3, Ube2e1, Ube2n, and Ube2g2,[12−14] it has been shown that the destabilization of this closed conformation by mutations either in the E2’s α2 helix or in the hydrophobic patch of ∼Ub abolishes or strongly diminishes the catalytic activity. Furthermore, there is evidence that binding of a RING-E3 promotes the formation of this closed conformation.[13,15,16] The recognition that the conformation of E2∼Ub in the presence of a cognate RING-E3 ligase is essential to function has driven the development of approaches for stabilizing the covalent linkage between Ub and E2 so that structural studies can be carried out. Strategies that have been employed include replacement of the thioester linkage with ester,[17−19] disulfide,[20] or isopeptide[16,21,22] linkages.

X-ray crystallographic structures of several E3/E2∼Ub ternary complexes have now been determined.[17−19,16,21,23] Among the RING-E3 structures a common theme is the observation of a closed E2∼Ub conformation, which is stabilized by direct interaction with the bound RING.[16,21,23] These studies confirm directly the ability of E2∼Ub conjugates to assume a closed conformation mediated by the hydrophobic patch of the donor ubiquitin and the α2 helix of the E2. Mutations designed to destabilize these interactions lead to reductions in activity, suggesting that these structures correspond to the activated conformation of the ternary complex. A similar ternary interaction has been reported in the case of Ube2g2 charged with Ub and bound to gp78-RING.[22] The existence of ternary E2, ∼Ub, and RING interactions is not universal, however, as it is not observed when E4B or BRCA1 RING domains interact with Ube2e3∼Ub.[13]

A dynamic picture of the E2∼Ub conformation that has emerged from the NMR studies complements the view obtained from these static structures. It appears that many E2∼Ub conjugates undergo extensive conformational fluctuations, with the nature of the fluctuations depending on the specific E2. The NMR studies carried out for Ubc1∼Ub[12] and Ubc13∼Ub[17] conjugates clearly show an substantial interaction between the ∼Ub moiety and the region surrounding the α2 helix of the corresponding E2 in a solution. Yet, the observed linebroadening indicates that these conjugates are in dynamic equilibrium with other presumably more open conformational states. The NMR studies of Ube2e3∼Ub,[17,24] on the other hand, conclude that the interaction between E2 and ∼Ub is relatively weak and nonspecific. Indeed, the predominant conformation is open enough so that UbcH5c∼Ub tends to form linear oligomers mediated by noncovalent interaction of the ∼Ub moiety of one conjugate with the E2’s backside Ub-binding site of another.[18,24] Similarly, oligomerization has been reported for Rad6∼Ub[25] conjugates. Alternative modes of interaction have also been reported, such as contacts between Ub and the α3 helix of UbcH8 in the UbcH8∼Ub conjugates.[20] Taken together, these studies suggest that E2∼Ub conjugates sample a wide variety of conformational states in addition to the putative closed state conformation required for catalytic activity.

The relationship between E2∼Ub conformation and catalytic function is an important question. It has now been shown that binding of a RING-E3 can influence the conformational dynamics of some E2∼Ub conjugates. The NMR titration of the BIRC7 RING domain into a solution of Ube2e2∼[15N]-Ub shows chemical shift perturbations (CSPs) consistent with an increase in Ube2e2/∼Ub interactions as well as direct ∼Ub/RING interactions.[16] Upon addition of the U-box domain of the E3 ligase E4B (E4BU) to Ube2e3∼Ub, CSPs are observed corresponding to the α2 helix of UbcH5c and the hydrophobic patch of ubiquitin consistent with the E4BU-induced closure of Ube2e3∼Ub.[13,15] In addition, formation of a hydrogen bond between a conserved arginine residue on E4BU and the backbone of Ube2e3 is found to be essential for the activation of Ube2e3. Subsequent NMR spin relaxation studies demonstrate that E4BU binding restricts the breadth of the Ube2e3∼Ub dynamic ensemble toward a more compact distribution while still remaining in dynamic equilibrium between closed and open states. Moreover, the change in conformational distribution is independent of whether a wild-type or an activation-incompetent mutant of E4BU is employed, suggesting an intricate, multifactor mechanism for Ube2e3∼Ub activation.[15]

Here, we investigate the interaction of native thioester-linked Ube2g2∼Ub conjugates with the RING domain from the E3 enzyme gp78. This E2/E3 pair function as a part of the endoplasmic reticulum (ER) quality control pathway in which the conjugates specifically catalyze the synthesis of K48-polyubiquitin chains prior to transfer onto the substrate proteins.[26−29] We have employed NMR spectroscopic methods and single-turnover diubiquitin formation kinetic assays to specifically examine the relationship between the Ube2g2∼Ub conformational ensemble, gp78-RING binding, and catalytic efficiency.

Results

The catalytic activity of Ube2g2 depends on the integrity of the so-called hydrophobic patch of the donor ubiquitin molecule linked to the Ube2g2 active site.[30] This requirement has now been observed for a number of different E2 enzymes (UbcH5a,[21] UbcH5b,[16] UbcH5c,[13] Ube2r1,[30,31] and Ube2s[30]). Furthermore, the Ube2g2 activity is both highly dependent on the presence of a cognate RING domain and highly specific for the ligation of two ubiquitin molecules by means of a K48 linkage (Figure S3). When Ube2g2 is charged with a ubiquitin molecule bearing a K48R mutation (Ub* is used herein to denote Ub-K48R), it remains inactive provided no ubiquitin molecules bearing a native K48 residue are present. We have developed a protocol that allows Ube2g2 to be charged by the means of a native thioester linkage to a donor-specific ubiquitin molecule (Ub*) at the concentrations suitable for NMR spectroscopy. These conjugates (Ube2g2∼Ub*) can be maintained for at least 4 h at pH 8.3 in the presence of gp78-RING without significant hydrolysis, allowing two-dimensional (2D) chemical shift correlation experiments to be carried out. This approach enables the NMR spectroscopic investigation of the interaction of Ube2g2 with a covalently linked donor Ub* molecule in the absence and presence of the RING domain from the E3 ligase gp78.

Determination of the Ube2g2/gp78-RING Binding Interface

Our initial NMR investigations focused on investigating the interaction between gp78-RING and Ube2g2 by 2D-[15N,1H]-transverse relaxation-optimized spectroscopy-heteronuclear single-quantum correlation (TROSY-HSQC) spectroscopy. Addition of unlabeled gp78-RING into a solution of 70%-2H,15N-labeled Ube2g2 produced chemical shift perturbations of the resonances corresponding to residues residing in helix α1 as well as loops 4 and 7. These measured CSPs are depicted in Figure a, and the corresponding residues having a CSP > 0.3 ppm are shown in red in a surface representation of Ube2g2 in Figure b. The observed RING-binding site for Ube2g2 corresponds to the canonical RING recognition interface and agrees closely with the observations for other E2 enzymes.[29,32,33]

Figure 1

Chemical shift perturbations seen when RING binds to free Ube2g2. (A) Bar graph of CSPs for Ube2g2 (200 μM) in the presence gp78-RING (1000 μM). (B) Pymol surface representation of Ube2g2 (PDB: 2KLY) illustrating the effects of gp78-RING binding. Residues for which measured CSPs > 0.3 ppm are given in red.

NMR CSP Study of Ube2g2∼Ub* Thioester Formation

Initiation of the thioester formation reaction in the NMR tube between unlabeled donor Ub* and 70%-2H,15N-labeled Ube2g2 leads to extensive changes in the [15N,1H]-TROSY-HSQC spectrum of Ube2g2. Progression of the reaction is easily monitored as the active site C89 amide resonance in Ube2g2 is well resolved and experiences a large CSP of approximately 3 ppm upon thioester formation. Shown in Figure b are the expanded spectral regions illustrating the sort of CSPs observed upon thioester formation (black to red peaks). Significant CSPs resulting from Ube2g2∼Ub* formation are observed in several different regions of the Ube2g2 sequence (Figure a), and in addition, a number of resonances disappear in the vicinity of the residues 95–110 and 120–135. A surface plot of Ube2g2, given in cyan and blue to depict these perturbations, shows that they include residues that are both proximal to and remote from the site of covalent modification (Figure c). These perturbations form a contiguous patch that is consistent with an interaction with the covalently linked ∼Ub* moiety. No evidence of such an interaction is seen in the absence of a thioester linkage between Ube2g2 and Ub*. Notably, the surface affected upon thioester formation does not overlap with but rather nearly adjoins the RING-binding interface (compare Figures b and 2c), consistent with the direct interaction of gp78-RING and the ∼Ub moiety as reported by Das et al.[22]

Figure 2

Chemical shift perturbations observed for Ube2g2 due to gp78-RING binding and charging with a donor ubiquitin (Ub*). (A) Overlaid bar graph of CSPs for Ube2g2 upon charging with ubiquitin (Ube2g2∼Ub*) (red) and in the presence of gp78-RING (1250 μM gp78-RING/170 μM Ube2g2∼Ub*) (black). The CSPs for the active site amide resonance of C89 are off scale. Measured values are 3 and 2.8 ppm with and without RING, respectively. (B) Overlaid 2D-[15N,1H]-TROSY-HSQC subspectra acquired at 800 MHz for 200 μM Ube2g2 (black), 170 μM Ube2g2∼Ub* (red), and 1000/140 μM gp78-RING/Ube2g2∼Ub* (green). Black arrows indicate the trajectory of the resulting CSPs. (C) Pymol surface representation of Ube2g2 (PDB: 2KLY) illustrating the effects of charging of Ube2g2 with Ub* in the presence of gp78-RING. Residues for which measured CSPs > 0.3 ppm, from panel (A), are given in cyan, whereas Ube2g2 residues for which the corresponding resonances either broaden out or are too weak to be unambiguously assigned are given in blue.

Subsequent addition of unlabeled gp78-RING to Ube2g2∼Ub* produces CSPs encompassing residues corresponding to the union of CSPs observed separately upon gp78-RING binding and thioester formation (compare Figures a and 2a). However, many residues that are perturbed upon thioester formation exhibit an increase in the measured CSP upon subsequent RING binding while preserving the same spectral trajectory (resonances shown in green in Figure b). Furthermore, the resonances that disappear upon the formation of Ube2g2∼Ub* are not recovered upon gp78-RING binding. These results suggest strongly that Ube2g2∼Ub* is in dynamic equilibrium between the states in which the ∼Ub* interacts with Ube2g2 (closed state) and the “open” states in which it does not, with gp78-RING binding causing an increase in the closed state population.

NMR CSP Study of Ube2g2∼Ub*-L8AI44A Thioester Formation

The identical experiment was carried out utilizing a donor ubiquitin molecule with a doubly mutated hydrophobic patch (i.e., Ube2g2∼Ub*-L8AI44A). Notably, the observed spectral perturbations (Figure b) are reduced substantially relative to the results for the Ub* donor molecule. Shown in Figure a are the measured CSPs upon thioester formation and with gp78-RING bound as well. Perturbed resonances are localized to residues in the region surrounding the active site and at the gp78-RING binding interface (Figure c). Notably, no CSPs were observed for regions remote from the active site. Furthermore, the measured CSPs for gp78-RING/Ube2g2∼Ub*-L8AI44A correspond closely to the vectorial sum of the separately measured CSPs for Ube2g2/RING and Ube2g2∼Ub*-L8AI44A. These results suggest that the donor moiety of Ube2g2∼Ub*-L8AI44A is incapable of interacting with Ube2g2 by means of its hydrophobic patch and thus exclusively occupies an open, functionally incompetent, conformation as reported for other E2s.[21,16,13,30,31,30]

Figure 3

Chemical shift perturbations observed for Ube2g2 due to gp78-RING binding and charging with a hydrophobic patch defective donor ubiquitin (Ub*-L8AI44A). (A) Overlaid bar graph of CSPs for Ube2g2 upon charging with a mutant donor ubiquitin (Ube2g2∼Ub*-L8AI44A) (red) and in the presence of gp78-RING (1250 μM gp78-RING/170 μM Ube2g2∼Ub*-L8AI44A) (black). The CSPs for the active site amide resonance of C89 are off scale. Measured values are 2.7 and 2.6 ppm with and without RING, respectively. (B) Overlaid 2D-[15N,1H]-TROSY-HSQC subspectra acquired at 800 MHz for 200 μM Ube2g2 (black), 170 μM Ube2g2∼Ub*-L8AI44A (red), and 1000/140 μM gp78-RING/Ube2g2∼Ub*-L8AI44A (green). (C) Pymol surface representation of Ube2g2 (PDB: 2KLY) illustrating the effects of charging of Ube2g2 with Ub*-L8AI44A in the presence of gp78-RING. Residues for which measured CSPs > 0.3 ppm, from panel (A), are given in cyan, whereas Ube2g2 residues for which the corresponding resonances either broaden out or are too weak to be unambiguously assigned are given in blue.

Determination of gp78-RING Binding Affinities

We propose a simple model for RING-induced closure of Ube2g2∼Ub* in Figure . In this model, the observed increase in closed state population results from a difference in the affinity for gp78-RING between the open and closed state conformations of Ube2g2∼Ub*. This change in affinity results from a direct contact between the gp78-RING domain and the ∼Ub* moiety in the closed state and potentially other factors as well. As the affinity of Ube2g2 for gp78-RING depends on both pH and Mg++ concentration, a titration was carefully carried out such that the affinities of Ube2g2 and Ube2g2∼Ub are measured on the same sample differing only by the addition of E1 to initiate thioester formation. Unlabeled Ub* and Ub*-L8AI44A were utilized as donors for 2H,15N-labeled Ube2g2 with calibrated aliquots of unlabeled gp78-RING added to independent NMR samples. This allows for the determination of any relative changes in the affinity to be established with a much higher confidence. The affinities resulting from a combined fit of all residues exhibiting a maximum CSP > 0.15 ppm are summarized in Table . The results show that the affinity of Ube2g2 for gp78-RING increases approximately 3-fold upon thioester formation with Ub* as the donor molecule. When Ub*-L8AI44A is utilized as the donor, the affinity change is marginally outside our estimated precision of measurement. We speculate that this may be due to weak noncovalent interaction between gp78-RING and free Ub in a solution. Representative fits to the gp78-RING titration data are shown in Figure . These fits underscore the general observation that the nature of gp78-RING/Ube2g2 binding remains unchanged upon thioester bond formation with a Ub*-L8AI44A donor ubiquitin, unlike the case of an unmutated donor ubiquitin (Ub*).

Figure 4

Hypothetical model for gp78-RING activation of Ube2g2∼Ub function. The Ube2g2∼Ub conjugates dynamically fluctuate between closed and open conformations populated according to the equilibrium constant K. This equilibrium shifts toward the closed state upon RING binding as a result of an increased affinity of gp78-RING for Ube2g2∼Ub when it is closed rather than open potentially due to direct ∼Ub/RING contacts.

Table 1

Measured Affinities for gp78-RING Binding to Ube2g2

	Ube2g2 + RING		Ube2g2∼Ub + RING
donor Ub	KD	# residues	KD	# residues
Ub*	479 ± 21	38	157 ± 14	44
Ub*-L8AI44A	417 ± 15	41	329 ± 18	41

Figure 5

Representative binding curves resulting from the titration of gp78-RING into the solutions containing 200 μM Ube2g2 and 170 μM Ube2g2∼Ub*. Data shown are CSPs measured for amide resonances of L16 (black) and L66 (red) corresponding to free Ube2g2 (circles) and Ube2g2∼Ub conjugates (squares). The result of a best fit to all CSP data are depicted by solid lines (Ube2g2∼Ub conjugates) and dashed lines (Ube2g2). (A) Ube2g2 charged with donor Ub-K48R. (B) Ube2g2 charged with donor Ub-K48RL8AI44A.

Estimation of Closed State Populations

It is tempting to speculate that one of the features of Ube2g2 activation upon gp78-RING binding is the transition of Ube2g2∼Ub conjugates from a dynamically populated ensemble of conformations to a fully closed conformation, which is catalytically viable. On their own, the NMR CSP data do not allow one to determine whether the RING-bound state of Ube2g2∼Ub* is entirely closed. Yet, assuming that the free Ube2g2 affinity for gp78-RING is an adequate proxy for the affinity operative for the open state of Ube2g2∼Ub* and the model of Figure holds, the measured change in the affinity between charged and uncharged states of Ube2g2 can be related to the ratio, n, of equilibrium constants describing the relative populations of closed and open states of Ube2g2∼Ub* in the presence and absence of bound gp78-RING. This allows the equilibrium constant, K, relating the populations of the closed and open states of Ube2g2∼Ub* to be estimated according toin which r is the experimentally determined ratio of CSPs for Ube2g2∼Ub* in the presence and absence of RING. We have identified 8 residues (G23, M42, G43, E45, D46, T47, F51, and G52) that are perturbed by Ube2g2∼Ub closure, yet they do not directly experience gp78-RING binding and are remote from the active site. Taking the average of CSP measurements corresponding to these residues (Table S4), a value of r = 1.37 ± 0.09 is obtained. In combination with the measured ratio of RING affinities (n = 3.05 ± 0.30), a value for K of 1.5 ± 0.8 is obtained. This equilibrium constant K is related to the corresponding population according to K/(K + 1). On the basis of this analysis, we estimate that in the absence of gp78-RING, 60% of Ube2g2∼Ub conjugates populate the closed state. Binding of gp78-RING causes Ube2g2∼Ub* to become more closed (82%), yet open conformational states are still dynamically populated.

Single-Turnover Kinetics Study of Diubiquitin Formation for Different Donor Mutants

To further explore the role of the interaction between the hydrophobic patch of the donor ubiquitin in Ube2g2∼Ub activation, four additional donor Ub hydrophobic patch mutants were created (Ub*-V70A, Ub*-L8A, Ub*-I44A, and Ub*-I44V) in addition to Ub*-L8AI44A. The rate of diubiquitin formation for each mutant was then quantified by means of the single-turnover kinetics assay shown in Figure A. We have found that the addition of ethylene diamine tetra-acetic acid and NEM to prevent the recharging of Ube2g2 renders a fraction of the Ube2g2∼Ub conjugates inactive. Instead, we quench the reaction by the addition of 15 U/mL hexokinase followed by the addition of gp78-RING and an acceptor ubiquitin molecule bearing an Oregon Green chromophore (OG-Ub) to allow quantification of diubiquitin formation by fluorescence. The initial rates of diubiquitin formation were measured at different concentrations of gp78-RING (Figure B) and then the kinetic parameters were determined according to a standard Michaelis–Menten analysis. This enabled the estimation of diUb formation velocities corresponding to gp78-RING saturated conditions. The resulting kinetic parameters are listed in Table . For three of the mutants (Ub*-L8AI44A, Ub*-L8A, and Ub*-I44A), the diubiquitin formation activity was effectively abolished. On the other hand, the most conservative mutant (Ub*-I44V) resulted in an unchanged diubiquitin formation activity relative to the wild-type donor Ub*. Unexpectedly, the donor ubiquitin bearing a V70A mutation exhibited a doubling in the rate of diubiquitin formation under gp78-RING saturated conditions.

Figure 6

Rates of diubiquitin formation depend strongly on the specific Ub mutant that Ube2g2 is charged with. (A) Diubiquitin formation assay. 10 μM Ube2g2 is preloaded with a specific donor ubiquitin by the addition of 1 mM adenosine 5′-triphosphate (ATP), 10 mM Mg++, and 250 nM E1. The reaction is then quenched followed by the addition of gp78-RING and 1 μM ubiquitin modified to bear an Oregon Green chromophore to allow the quantification by fluorescence. (B) Measured rates of diubiquitin formation for the specified donor Ub mutant and as a function of RING concentration. Lines represent best fit to the equation Vinit = Vmax*[RING]/(KM + [RING]).

Table 2

Kinetic Parameters Measured for Different Donor Ubiquitin Mutants

donor Ub	Km (μM)	Vmax (μM/min)
Ub*	65 ± 20	0.1 ± 0.03
Ub*-I44V	86 ± 28	0.1 ± 0.01
Ub*-L8A	94 ± 30	0.004 ± 0.0005
Ub*-I44A	32 ± 8	0.005 ± 0.0004
Ub*-V70A	136 ± 49	0.2 ± 0.03
Ub*-L8AI44A	139 ± 53	0.003 ± 0.0004

NMR CSP Study of Ube2g2∼Ub*-X Donor Mutants

Two-dimensional-[15N,1H]-TROSY-HSQC NMR spectroscopy was performed on Ube2g2 conjugated, respectively, to the different donor ubiquitin mutants to examine the relationship between the population of the closed conformational state and the ability to catalyze diubiquitin formation. In these experiments, only Ube2g2 was isotopically labeled, whereas the specific donor ubiquitin and gp78-RING remained unlabeled. The question was whether mutations in the hydrophobic patch destabilize the closed state conformation in a manner that correlates with the observed loss or gain of catalytic competence. Figure shows an overview of the results for the four donor mutants Ub*-I44V, Ub*-I44A, Ub*-L8A, and Ub*-V70A. Most striking is that all of the mutants (compare NMR overlaid subspectra shown in Figures , 3, and 7) exhibit significant weakening or disappearance of the hydrophobic patch mediated binding between the donor ubiquitin molecule and Ube2g2. This is best seen by consideration of the measured CSPs for the region furthest from the active site, which includes residues such as G23, G43, and T47. From this perspective, the Ub*-I44A, Ub*-V70A, and Ub*-L8AI44A mutants are not observed to populate the closed state conformation at all. On the other hand, it could be argued that the Ub*-I44V and Ub*-L8A mutants do still populate the closed state, albeit to a greatly diminished extent. This is based on the CSPs observed for the amide N–H of residue T47 (Figure b,c, upper right panel), which is very remote from the active site and thus a reasonable probe for closing. However, upon a close examination of all measured CSPs, we conclude that only Ub*-I44V displays a CSP signature similar to Ub*, whereas Ub*-L8A exhibits notable differences. Although it is not possible to draw more detailed conclusions from these data, what is striking is that these NMR spectroscopic observations are totally discordant with the measured rates of diubiquitin formation among the different donor ubiquitin mutants. That is, all of the mutants studied have small to nonexistent closed state populations, whereas the observed catalytic rates range from negligible to twice that observed for Ub* in the case of Ub*-V70A.

Figure 7

Chemical shift perturbations observed for Ube2g2 due to gp78-RING binding and charging with specific ubiquitin mutants Ub*-X. Shown at the top of each panel are overlaid 2D-[15N,1H]-TROSY-HSQC subspectra acquired at 800 MHz for 200 μM Ube2g2 (black), 160 μM Ube2g2∼Ub*-X (red), and 975/130 μM gp78-RING/Ube2g2∼Ub*-X (green). At the bottom of each panel are pymol surface representations of Ube2g2 (PDB: 2KLY) illustrating the effects of charging of Ube2g2 with the operative donor mutant Ub*-X in the presence of gp78-RING. Residues for which measured CSPs > 0.3 ppm are given in cyan, whereas Ube2g2 residues for which the corresponding resonances either broaden out or are too weak to be unambiguously assigned are given in blue. Each panel shows the results for a specific donor mutant: (A) Ub*-V70A. (B) Ub*-I44V. (C) Ub*-L8A. (D) Ub*-I44A.

In addition to the resonances that shift upon thioester formation, there are a number of those that broaden out and disappear. Corresponding residues are located either proximal to the active site or in the region of the α2 helix, which is the center of the donor ubiquitin binding site. The observation of a significant line broadening can be considered to be indicative of significant conformational fluctuations occurring on the microsecond to millisecond time scale. We have carried out a close examination of the TROSY-NMR data to identify which resonances experience substantial conformational exchange broadening, manifesting as either complete disappearance or significant reduction in intensity. These regions are displayed in dark blue on a surface representation of Ube2g2 in Figures , 3, and 7. The two mutants with the least conformational exchange broadening are Ub*-I44A and Ub*-L8AI44A, suggesting that they are strongly biased toward the occupation of an array of open conformations with minor interactions with Ube2g2. Notably, all three functional Ube2g2∼Ub conjugates (i.e., bearing Ub*, Ub*-I44V, and Ub*-V70A) show quite similar regions of exchange broadening, which one might propose as corresponding to the conformational fluctuations associated with the reorganization of the functionally relevant active site. The Ub*-L8A donor mutant, although exhibiting a larger region of exchange broadening relative to I44A-containing mutants, still exhibits a less extensive region than the active mutants. We speculate that the interaction is sufficiently different to preclude fruitful active site reorganization. Viewed from this vantage point, the results are in better alignment with the measured rates of diubiquitin formation and lend support to a view of Ube2g2 function in which certain contacts must be present between Ube2g2 and the ∼Ub moiety to allow for constructive active site rearrangement. Formation of these essential contacts is likely driven by the closure of Ube2g2∼Ub conjugates, but apparently a priori interaction of the hydrophobic patch of the ∼Ub moiety with the α2 region of Ube2g2 is not strictly necessary.

Discussion

Our results demonstrate that, when charged, Ube2g2 interacts directly with the hydrophobic patch of its covalently linked ∼Ub moiety. This closed conformation is transient and in dynamic equilibrium with other open conformations. Furthermore, NMR CSP analysis shows that the conformational distribution of Ube2g2∼Ub shifts toward the closed state upon binding of the RING domain from the E3 ligase gp78. Sufficient disruption of the closed conformation by means of mutations of the hydrophobic patch of the donor ubiquitin abolishes Ube2g2’s RING-dependent catalytic activity as observed for a number of other E2 enzymes.[13,16,21,30,31] These observations support the crystallographic structures of RING/E2∼Ub ternary complexes[16,21] as well as a model of gp78-RING/Ube2g2∼Ub,[22] suggesting that RING binding stabilizes a catalytically competent state in which the donor ubiquitin is in direct contact with the E2 enzyme. These results are also in accordance with the NMR-based studies showing that E2∼Ub conjugates are in dynamic equilibrium between open and closed states with the binding of RING causing a bias in the population toward the closed state.[13,15]

We measured a 3-fold increase in the affinity of gp78-RING for Ube2g2∼Ub relative to Ube2g2 alone, which agrees very closely with that reported by Byrd and co-workers.[22] Preferential binding of RING domains to E2 enzymes in their Ub-charged state appears to be a common theme, although there are some exceptions.[13] Affinity increases upon charging have been reported for the BIRC7/UbcH5b∼Ub[16] and Rbx1/Roc1/CDC34∼Ub[34] systems, both of which exhibit ternary contacts between the RING, E2, and Ub components. In the BIRC7/UbcH5b study, it was shown that destabilization of the closed conformation by hydrophobic patch mutations of the conjugated donor ubiquitin reduced the affinity for BIRC7 RING.[16] Likewise, we find here that Ube2g2 charged with a Ub-L8AI44A mutant has nearly the same affinity for gp78-RING as for Ube2g2 alone. As such, we have considered a simple model in which the observed increase in RING affinity for Ube2g2∼Ub conjugates is energetically linked to the stabilization of the closed conformation upon RING binding. On the basis of this analysis, we conclude that gp78-RING binding to Ube2g2∼Ub increases the fractional population of the closed state from approximately 60 to 82%. Our conclusion that Ube2g2∼Ub remains in dynamic equilibrium between closed and open states even while gp78-RING is bound is in agreement with the conclusion of Soss et al. in the case of the E4BU/UbcH5∼Ub complex.[15]

The enhancement of an E2 enzyme’s catalytic activity resulting from RING binding is typically many orders of magnitude. Clearly, the change in the closed state population of Ube2g2∼Ub upon RING binding is not a significant factor for activation. The primary factor for activation is likely a conserved hydrogen bond present in several E3/E2 pairs as proposed by Klevit and co-workers.[13] Indeed, a study of Ube2g2 activation by gp78 has established the presence of a salt bridge between the Ube2g2 residue E108 and the gp78 residue R379, as well as direct RING/∼Ub interactions.[22] Our results are fully consistent with the existence of ternary contacts between Ube2g2, gp78-RING, and the ∼Ub moiety in the closed state. The significant feature of our results is not the modest change in the closed state population upon RING binding, but rather the fact that Ube2g2∼Ub remains dynamic even when RING is bound.

To probe the role of Ube2g2∼Ub closure on function, we examined five different donor ubiquitin molecules bearing mutations to their hydrophobic patch. Our results are in agreement with the reports of the loss of activity for donor Ub molecules with L8A and I44A mutations.[13,14,16,21,30,31] Our observed increase in the activity in the case of Ub*-V70A is a surprising result at odds with most reports to date. However, a study of the yeast Ubc7 along with the cognate RING domains Doa10 and Hrd1 found that in the case of Hrd1 (but not Doa10), Ubc7∼Ub*-V70A was fully active, whereas Ubc7∼Ub*-L8A retained some activity. Ubc7 is the yeast ortholog of Ube2g2, exhibiting 79% sequence homology,[14] whereas Hrd1 is the ortholog of gp78.[35] Apparently, there are features of the gp78-mediated activation of Ube2g2 that are not shared with other E2/E3 pairs.

Our results demonstrate no relationship between the population of a closed Ube2g2∼Ub conformation as observed by NMR spectroscopy and catalytic function as quantified by our single-turnover kinetics assay. We propose that the results be viewed simply in terms of whether a specific donor mutant is able to undergo mutual reorganization of its own C-terminal tail with the Ube2g2 active site into the catalytically important conformation. The apparent importance of the Ube2g2∼Ub closed state then arises due to the reduction of the conformational rearrangement space available. In essence, closure of the Ube2g2∼Ub conjugate acts to guide the required active site rearrangement. From this vantage point, the donor Ub mutants that exhibit loss of function (Ub*-I44A, Ub*-L8A, and Ub*-L8AI44A) do so because they are sampling mostly open conformations, or alternative closed state conformations from which the catalytically active conformation is not accessible. The Ub*-I44V donor mutant has diubiquitin ligation activity, which is essentially identical to the wild-type because it does sample the closed conformation, although with a much lower fractional population. This lack of correlation between the closed state population and catalytic activity suggests a process governed by the operative kinetic rates of opening and closing. This suggests a possible explanation for the surprising enhancement in activity observed for the Ub*-V70A mutant. Further experiments are required to test this possibility. Preliminary results indicate that the introduction of a V70A mutation is not capable of rescuing a catalytically inactive donor ubiquitin (specifically Ub*-L8AI44AV70A as donor) (Figure S4).

This work did not explore in detail aspects of the catalytic mechanism involving binding and recognition of the acceptor Ub molecule. It has been presumed that the aspects relating to acceptor Ub recognition are independent and catalytically downstream of gp78-RING binding, Ube2g2∼Ub closure, and subsequent active site reorganization. Yet, it remains that the acceptor ubiquitin could play an important role in this active site reorganization as seen in Ube2s,[30] or when it is covalently linked to the active site of a second Ube2g2 molecule.[36] Although the NMR experiments were carried out with a stoichiometric excess of ubiquitin present, the diubiquitin kinetics assays were performed at a 1:10 ratio of OG-Ub/Ube2g2∼Ub* in large part because the reaction proceeds too fast to be followed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) at higher Ub concentrations. Indeed, we observed that the rate of reaction is approximately linear with the acceptor Ub concentration and have estimated an affinity of 1 mM for the acceptor Ub binding based on the measurement of Ki with Ub* as the competitive inhibitor (data not shown). Considered together with the lack of correlation between the kinetics results and the observed closed state populations, we propose that the active site rearrangement, facilitated by Ube2g2∼Ub closure, is the rate-limiting step for Ub-to-Ub ligation by Ube2g2.

Our conclusion that Ube2g2∼Ub remains in a dynamic equilibrium between open and closed states even when gp78-RING is bound could have a significant functional relevance in that it could substantially reduce the rate of hydrolysis or the undesired side reactions by maintaining the active site in a state of dynamic reorganization. Recently, it has been shown that Ube2g2, in collaboration with oligomeric assemblies of gp78, elongates polyUb chains from the distal end of the chain, which is, in turn, positioned by interactions with the CUE domain of gp78.[37,38] Furthermore, the donor Ube2g2∼Ub conjugates, but not the growing acceptor chain, are recruited by means of high-affinity binding to the G2BR domain of gp78.[29,37] Although we cannot exclude the possibility that G2BR promotes further closure of the Ube2g2∼Ub conjugates, we propose that the inherent dynamics of the Ube2g2∼Ub conjugates coupled with the dependence of the rate of catalysis on the rate of active site organization enables gp78 oligomers to effectively be preloaded with the Ube2g2∼Ub conjugates, which can efficiently elongate the distal end of a growing polyUb chain in proximity.

If the maintenance of Ube2g2∼Ub dynamics is functionally relevant, then the question arises as to what factors prevent stabilization of a single RING bound catalytic complex. Although there could very well be Ube2g2-specific factors, an attractive possibility is that the conformational dynamics of ubiquitin itself may be the governing factor. Using NMR spectroscopic techniques, it has been shown that ubiquitin undergoes microsecond time scale fluctuations in the region surrounding its hydrophobic patch.[39] Subsequent studies have shown that these conformational fluctuations are important aspects pertaining to ligand recognition and energetics of binding.[40−43] Presuming that only a subset of ∼Ub conformations are competent to bind to Ube2g2 and the binding energies involved are small, one could expect that Ube2g2∼Ub may remain in a dynamic equilibrium with the kinetics of closing mediated by the associated conformational fluctuations of ubiquitin.

Experimental Section

Expression of E1, Ube2g2, gp78-RING, and Ubiquitin

A codon-optimized insert for the RING domain of the human gp78 gene (amino acids 227–298) was synthesized by Integrated DNA Technologies (Coralville, IA) and inserted into a pET28b vector with a His-tag removable by TEV protease. Ube2g2 was cloned as described in Ju et al.[44] The ubiquitin plasmid was a gift from Prof. Cecile Pickart. The plasmid for E1, with an N-terminal His-tag encoded, was a gift from Prof. Cynthia Wolberger. Ubiquitin mutants were generated by the polymerase chain reaction site-directed mutagenesis using Invitrogen oligonucleotides and Pfu Turbo (Stratagene) at annealing temperatures between 45 and 65 °C. Incorporation of mutations was verified at the JHU sequencing facility.

Plasmids for gp78-RING, Ube2g2, and E1 were transformed into Escherichia coli strain BL21 (DE3) cells, whereas the ubiquitin plasmids were transformed in BL21 (DE3) cells containing the PJY2 accessory plasmid. The drug screen used for Ube2g2 and gp78 RING was kanamycin (50 μg/mL), whereas ampicillin (100 μg/mL) and chloramphenicol (25 μg/mL) screens were used for ubiquitin. 10 mL LB cultures were inoculated with a colony from a given plate and grown overnight at 37 °C. The small cultures were diluted into 1 L cultures and grown at 37 °C until they reached an OD600 of 0.6–0.8. Perdeuteration and 15N-labeling of proteins was achieved by pelleting the cells after reaching the target OD followed by resuspension in the M9 minimal media containing deuterium oxide, 1 g/L 15NH4Cl, and 6 g/L unlabeled glucose. Expression of E1, Ube2g2, and gp78-RING was induced by the addition of 0.5 mM IPTG and allowed to continue overnight at 18 °C. Expression of ubiquitin was induced by the addition of 0.4 mM IPTG followed by incubation at 37 °C for 6 h. Cells were pelleted by centrifugation at 6000 rpm for 30 min and stored at −80 °C.

Purification of Ube2g2 and gp78-RING

The cells that expressed gp78-RING were lysed in Bugbuster Protein Extraction Reagent (EMD Millipore) with Benzonase nuclease (EMD Millipore), and the lysate was clarified by centrifugation at 12 000 rpm for 30 min. The supernatant was loaded onto a pre-equilibrated 5 mL gravity flow nickel column and washed with 25 mM Tris buffer at pH 7.5. The proteins were eluted in 1 mL fractions with 25 mM Tris buffer at pH 7.5 containing 200 mM imidazole. Fractions containing the proteins (assayed by SDS page) were loaded onto a size-exclusion column (Superdex 75, GE Healthcare Life Sciences) and washed with 10 mM sodium phosphate buffer pH 7.5 containing 1 mM tris(2-carboxyethyl)phosphine (TCEP) and 0.05% NaN3. Fractions with pure protein were pooled together and incubated with TEV protease for 2 h at 30 °C. The cleaved protein was dialyzed against 25 mM Tris buffer at pH 7.5 before loading onto the nickel column. The pure protein was collected in a single fraction and dialyzed against 10 mM sodium phosphate buffer at pH 7.5 containing 1 mM TCEP and 0.05% NaN3. The protein was concentrated by centrifugation at 4000g in a 3k MWCO centrifugal tube (Millipore). Ube2g2 was purified in a similar method as described in Ju et al.[44] E1 was purified as described in Berndsen.[45]

Purification of Ubiquitin and Ubiquitin Mutants

Cells were resuspended in 50 mM Tris buffer at pH 7.6 containing 0.02% Triton, 1 tablet Roche protease inhibitors, 10 mM MgCl2, 10 μg/mL DNaseI, and 1 mM phenylmethylsulfonyl fluoride. Lysis was performed using a French Press at 1000 PSI with the precipitate removed by centrifugation at 12 000 rpm for 30 min. A 1% v/v solution of perchloric acid (70%) was added dropwise to the stirring supernatant. The precipitate formed upon acid addition was removed by centrifugation. The supernatant was dialyzed against 50 mM ammonium acetate at pH 4.5, after which the protein was loaded onto a cation-exchange column pre-equilibrated with 50 mM ammonium acetate at pH 4.5 and then eluted with a 500 mM NaCl gradient. The fractions containing ubiquitin were run through a 30k MWCO centrifugal tube at 4000g to remove high-molecular-weight impurities. The pure protein was then concentrated and exchanged into 10 mM sodium phosphate buffer at pH 7.5 with 1 mM TCEP and 0.05% NaN3 using a 3k MWCO centrifugal tube at 4000g. Concentrations for all of the proteins were determined by absorbance at UV 280 nm in the presence of 6 M guanidinium chloride.

NMR Spectroscopy of Ube2g2∼Ub* Conjugates

Charging of Ube2g2 with Donor Ubiquitin for NMR Experiments

To produce Ube2g2∼Ub* conjugates in the NMR tube, a 200 μM solution of [70%-2H,U-15N]-labeled Ube2g2 was prepared in 50 mM Tris–HCl buffer at pH 8.3 (NMR buffer) containing 1 mM MgCl2, 1 mM TCEP, 4 mM ATP, 0.05% NaN3, 10% D2O, 10 mM phosphocreatine, 0.6 U/mL creatine phosphokinase, 0.6 U/mL inorganic pyrophosphate, 300 nM C-terminal hydrolase, and 1.25 mM ubiquitin-K48R (denoted Ub*). Thioester formation was initiated by the addition of 1 μM E1, with the progress monitored by [15N,1H]-HSQC NMR spectroscopy. The thioester formation reaction was typically complete after 15 min.

Assignment of Ube2g2∼Ub* Conjugates

Resonance assignments for Ube2g2 have been determined previously.[44] Assignment of Ube2g2∼Ub* resonances were confirmed based on three-dimensional (3D) 1HN-15N-13Cα correlation (HNCA) spectroscopy of 500 μM [70%-2H,U-15N]-labeled Ube2g2∼Ub* in 3-(N-morpholino)propanesulfonic acid buffer at pH 7 containing 1 mM MgCl2, 1 mM TCEP, 4 mM ATP, 0.05% NaN3, 10% D2O, 10 mM phosphocreatine, 0.6 U/mL creatine phosphokinase, 0.6 U/mL inorganic pyrophosphate, 300 nM C-terminal hydrolase, 2.5 mM ubiquitin-K48R, and 2 μM E1. Under these lower pH conditions, which also require higher E1 and ubiquitin concentrations, the rate of thioester hydrolysis is slowed, permitting 15 h data acquisition times.

NMR Data Collection

All of the NMR data were collected at 20 °C on a Varian Inova 800 MHz spectrometer. The 2D-[15N,1H]-TROSY-HSQC spectra were acquired with 2100 and 328 complex data points and spectral widths of 14 005 and 2593 Hz in the 1H and 15N dimensions, respectively. The total experimental acquisition time was approximately 2 h for each dataset. For the 3D HNCA datasets, 840, 32, and 68 complex data points were acquired with the spectral widths of 14 000, 5430 and 2066 Hz corresponding to 1H, 13C, and 15N dimensions, respectively. The total experimental duration was approximately 15 h. All of the datasets were Fourier-transformed using NMRPipe[46] and peak positions determined using Sparky[47] software.

Determination of gp78-RING Binding Affinities

Binding affinities of gp78-RING for Ube2g2 and Ube2g2∼Ub* were determined by [15N,1H]-TROSY-HQSC NMR spectroscopy for 200 μM solutions of [70%-2H,U-15N]-labeled Ube2g2 in the Tris pH 8.3 NMR buffer with 1 mM ubiquitin-K48R (Ub*) and the specified amount of gp78-RING present. Separate samples were prepared in parallel for each gp78-RING titration point, with the Ube2g2 and Ube2g2∼Ub* datasets collected for the same sample differing only by the addition of E1 to form the Ube2g2∼Ub* species when appropriate. Composite CSPs were calculated from individual displacements measured separately in the 1H and 15N dimensions of the TROSY-HSQC spectra as followsand then fit to the following equationwhere Kd represents the dissociation constant and Δδmax the residue-specific CSP expected when gp78-RING is fully bound. All of the CSP data corresponding to residues with an observed maximum CSP value >0.15 ppm were included in a fit to eq . Errors in the fitted parameters were then estimated from the standard deviation observed from 10 000 replicate fits, which included only 50% of the eligible residues (duplicates allowed). The fitted maximum CSP parameters were subsequently utilized for the purposes of estimating the Ube2g2∼Ub closed state population in the presence of gp78-RING.

Diubiquitin Single-Turnover Kinetics Assays

To quantify the catalytic activities of Ube2g2 charged with different donor ubiquitin mutants, a single turnover kinetics assay was designed in which a fluorescent tag was introduced onto a wild-type acceptor ubiquitin molecule. A 10-fold stoichiometric excess of precharged Ube2g2∼Ub*-X (where X denotes a specific mutant) conjugates were then added and the reaction was followed by fluorescence. This approach enabled the rate of diubiqutin formation to be quantified independently of the typically slow thioester loading rate.

Ube2g2∼Ub*-X conjugates were prepared in 50 mM Tris pH 7.5, 10 mM MgCl2, 40 mM glucose, 5 mM TCEP, 1 mM ATP, 275 nM Ubiquitin C-Terminal Hydrolase, 250 nM E1, 10 μM Ube2g2, and 20 μM of the specific donor ubiquitin mutant (Ub*-X). Donor ubiquitin mutants employed were Ub*, Ub*-L8A, Ub*-I44A, Ub*-I44V, Ub*-V70A, and Ub*-L8AI44A. The charging reaction was incubated at 20 °C for 5 min and then quenched with 15 U/mL hexokinase (Sigma) for an additional 5 min at 20 °C. The hexokinase-based quenching scheme, which operates by depleting ATP, was found to be very effective and benign with respect to Ube2g2 function. The yield for thioester preparation was estimated at >95% by SDS-PAGE. The wild-type acceptor ubiquitin was labeled with Oregon Green (OG-Ub) by reacting Oregon Green 488 Maleimide (Thermo Fischer Scientific) with a ubiquitin mutant bearing a cysteine residue inserted between the first (methionine) and second residue (glutamine). The diubiquitin formation reaction was initiated by the addition of an appropriate amount of gp78-RING followed by the addition of 1 μM of the acceptor OG-labeled ubiquitin (OG-Ub). Reactions were allowed to proceed in parallel at 20 °C, 1 atm, with individual reactions quenched by the addition of reducing SDS-PAGE loading buffer at intervals of 1, 2, 3, 4, 5, 7, 10, 15, 20, and 30 min after initiation. Each quenched reaction time point was then run on an 8% Bis–Tris SDS-PAGE gel. The OG-Ub was then visualized in situ using a 488 nM laser on a Typhoon 9410 Variable Mode Imager (GE Healthcare). The bands were quantified using the ImageQuant program (GE Healthcare).

The fractional progress toward reaction completion, fP, was computed for each time point based on the measured fluorescence intensities for the Ub and diUb bands according toThe initial rates of product formation were then obtained by fitting to a straight line in the case of mutants with low activity (i.e., Ub*-L8AI44A, Ub*-L8A, and Ub*-I44A) and in the other cases to the exponential functionwhere fP,0, fP,max, and v are allowed to float with the initial velocity, vinit, obtained according toIt was necessary to account for an incomplete diUb formation (i.e., fP,max ≠ 1) due to the competition with thioester hydrolysis. The relative error in the determined values of vinit was estimated at 11.8% based on the observed variance over 9 replicated experiments performed for the Ub*-I44A, Ub*-V70A, and Ub* mutants. This relative error was then propagated into the final reported Michaelis–Menten parameters using a bootstrap-type analysis with 5000 replicate fits performed.