Interprotomer motion-transmission mechanism for the hexameric AAA ATPase p97

Abstract

Multimeric AAA ATPases represent a structurally homologous yet functionally diverse family of proteins. The essential and highly abundant hexameric AAA ATPase p97 is perhaps the best studied AAA protein, playing an essential role in various important cellular activities. During ATP-hydrolysis process, p97 undergoes dramatic conformational changes, and these changes are initiated in the C-terminal ATPase domain and transmitted across the entire length of the molecule to the N-terminal effector domain. However, the detailed mechanism of the motion transmission remains unclear. Here, we report an interprotomer motion-transmission mechanism to explain this process: The nucleotide-dependent motion transmission between the two ATPase domains of one protomer is mediated by its neighboring protomer. This finding reveals a strict requirement for interprotomer coordination of p97 during the motion-transmission process and may shed light on studies of other AAA ATPases.

Protein p97, also called valosin-containing protein (VCP) in mammals and Cdc48 in yeast, is an essential and highly abundant type-II AAA ATPase, accounting for approximately 1% of the protein of the cell (1). The gene encoding p97 is one of the most conserved from archaebacteria to yeast and to humans (2–4). Protein p97 is involved in various important cellular activities including endoplasmic reticulum-associated degradation (ERAD), homotypic membrane fusion, programmed cell death, activation of B and T cells, the stress response, cell-cycle control and regulation of certain transcription factors (5–14). The diverse biological functions of p97 implicate it in a number of neurodegenerative diseases including Alzheimer’s (15), Parkinson disease, and cancer (16).

Protein p97 is a ring-shaped homohexameric protein. Each protomer consists of an unstructured C-terminal tail and three structured domains: a somewhat flexible N domain and two highly conserved AAA+ domains, termed the D1 and D2 domains (Fig. 1A) (17). The D1 domain is a degenerate ATPase domain that is responsible for hexameric-state stability (18), whereas the D2 domain is the major ATPase domain of p97 at physiological temperature (1). Protein p97 binds to different substrate-recruiting cofactors and substrates, resulting in various activities (16, 19–21). Structural studies of p97 in different nucleotide states showed that the protein undergoes conformation changes during nucleotide binding and hydrolysis [see review (17)]. In this manner, the free energy released from ATP hydrolysis may be utilized to drive the movement of its substrates from the lumen of the endoplasmic reticulum (ER) to the cytosol. However, it has been controversial how the ATP-hydrolysis-induced motion is transmitted through the length of the entire molecule; that is, from the D2 domain to the D1 domain and then to the distant N domain of p97 (22–26).

Fig. 1.

Crystal structure of p97 suggests three possible routes for motion transmission from the D2 to the D1 domain. (A) Domain sketch of p97 protein with domain boundary residues marked. (B) Cartoon presentation of p97 structure [Protein Data Bank (PDB) ID code 3CF2] with the domain color matched to the sketch in (A). Route 1 indicates intramolecular transmission via direct contact. Route 2 and 3 refer to motion transmission via D1–D2 linkers either within the same protomer or from the neighboring promoter, respectively. These routes are labeled as 1, 2, and 3 in red circles.

The D1 and D2 domains of p97 are connected by a highly conserved 20-residue linker (the D1–D2 linker), consisting of amino acid residues 461–480 (Fig. 1A). Structural observations suggested that this linker may play an important role in communication between the D2 and D1 domains (23, 27). However, direct experimental evidence pertaining to the physiological role of the linker has been lacking, and furthermore, it has not been known how the D1–D2 linker may transmit motion.

In the hexamer structure of p97, the D2 domain makes contact with the D1 domain by three interactions (Fig. 1B): the direct interaction between the regions near residue V559 on the D2 domain and residue R349 on the D1 domain (labeled 1 in Fig. 1B); interaction via the D1–D2 linker on the same protomer (labeled 2); and via the D1–D2 linker of the neighboring protomer (labeled 3). These three interactions implicate three possible routes for the motion transmission from D2 to D1. Routes 1 and 2 describe possible motion transmission pathways within the same protomer, whereas route 3 depicts an interprotomer pathway in which one protomer utilizes the D1–D2 linker from its neighboring protomer to transmit the motion. We asked if any or all of the routes is actually taken by p97 to propagate the motion. We reasoned that systematic mutagenesis at the D1–D2 interface, which disrupts the biological functions of p97 such as ERAD activity but changes neither the ATPase activity nor the oligomeric structure, may reveal the true motion transmission pathway. We found that the nucleotide-induced motion of one p97 protomer is transmitted from its D2 domain to the D1 domain via the linker residue L464, not on the same protomer, but on its neighboring protomer. Based on this, we propose an interprotomer mechanism for the nucleotide-dependent motion-transmission process between the two ATPase domains of p97.

Results and Discussion

Improved ERAD Ativity Assay.

First, we optimized a previously described ERAD-assay method (28). In that method, cDNA fragments for p97 and the cell-surface protein ΔCD4 mutant were constructed in the same plasmid with an internal ribosome entry site (IRES) element inserted between them (28). Based on the assumption that both p97 and the ΔCD4 mutant are expressed simultaneously, expression of p97 was measured by expression of the ΔCD4 mutant, which recognizes the dye-conjugated anti-CD4 antibody and therefore can be detected using fluorescence-activated cell sorting (FACS). The drawbacks of this system are: (i) the expression of p97 mutant is not detected directly and (ii) the expression of ΔCD4 mutant must be detected by FACS using dye-conjugated antibody. To overcome these disadvantages, in this study the ERAD-assay method was modified to ensure the expression of p97 and simplify the FACS-recognition process. Briefly, full-length p97 was in-frame fused to the C terminus of the fluorescent protein mCherry and all mutations were made on the template of pmCherry-p97. The plasmids of p97 wild type and mutants were transfected into a reporter-cell line TG12 (HEK293) cells stably expressing the ERAD substrate TCR-α-GFP (28). The modified vectors ensured us that the mCherry expressing cells (red cells) also express the recombinant p97. The effects of the mutations were examined by measuring the density of the GFP fluorescence in the red cells; namely, the mcherry positive cells simultaneously expressing p97, using the fluorescence-activated cell sorting (FACS) technique.

We tested the validity of this experimental system by preparing mutants of the well-studied Walker-A/B motifs and SRH domains. It has been shown that the Walker-A/B motifs and SRH domains of p97 play critical roles in mediating ATP binding and hydrolysis (1, 29). Consistent with an earlier report (28), dramatic accumulation of the ERAD substrate TCR-α-GFP was observed in the pairs of mutations on the Walker-A/B and SRH domains (Fig. 2A), indicating this optimized system is suitable for functional assays of p97. Interestingly, when we compared the mutants in a pairwise manner, we found significant differences between the D1 and D2 mutants on the degradation of TCR-α-GFP: In the pairs of mutants in the Walker-A/B and SRH domains, substrate accumulation in mutants in the D2 domain was always much higher than that in mutants in the D1 domain. Our observation is consistent with the earlier in vitro finding that the defective ubiquitin-proteasome system in the p97-depleted situation was partially restored by adding back the mutant containing Walker-A mutation in the D1 domain, but it was not restored by adding back the mutant in D2 domain (1). All these results indicated that certain mutations in the D2 domain caused more significant substrate accumulation than that in the D1 domain, in agreement with the previous conclusion that D2 is the major ATPase domain in p97 (1).

Fig. 2.

Mutation of route 1 residues does not abolish the p97 function. (A) Mutational effects of p97 in the Walker-A/B and SHR regions on degradation of the ERAD substrate TCR-α-GFP. Reporter cells stably expressing TCR-α-GFP were transfected with pmCherry-p97 variants. Data were normalized to wild-type p97 and shown as the fold increase in mean fluorescence. Error bars were calculated based on three separate experiments. (B) Block mutation of residues 349–351 has no effect on degradation of the ERAD substrate TCR-α-GFP.

Route 1 Is Not Responsible for the Motion Transmission from the D2 to the D1 Domain.

Next, we investigated if the direct interaction, route 1 in Fig. 1B, is responsible for the motion transmission from the D2 to the D1 domain. Route 1 utilizes the potential direct interactions through the region near amino acid residue R559 (on the D2 domain) and amino acid residue R349 (on the D1 domain) (Fig. 1B). Because the region near residue R559 may also be involved in interaction with the D1–D2 linker of the adjacent protomer, we made an alanine block mutation of residues 349–351 (349–351AAA) and examined the role of this possible direct interaction in the ERAD activity of p97. As shown in Fig. 2B, we found that there is no defect in ERAD activity upon mutation at the region near residue R349, suggesting route 1 in Fig. 1B is not responsible for motion transmission.

The D1–D2 Linker Residue L464 Relays Motion Transmission from the D2 to the D1 Domain.

The D1–D2 linker, which connects the D1 and D2 domains, is highly conserved from yeast to human (Fig. 3A). Because the putative routes 2 and 3 both utilize the D1–D2 linker in the motion transmission process, we then verified the functional importance of the D1–D2 linker of p97. Indeed, the D1–D2 linker has been proposed in the past to mediate the motion transmission from the D2 domain to the D1 domain (23, 25, 27). However, this proposal was speculative because there was no supporting experimental evidence. Here, we first demonstrated the functional importance of the D1–D2 linker of p97 by preparing individual or block mutations encompassing the whole linker region and examining the mutational effect on the degradation of the ERAD substrate TCR-α-GFP. As shown in Fig. 3B, mutant L464A exhibited a significant accumulation of the substrate, indicating a dramatic loss of ERAD activity. For comparison, we also prepared a D2 domain-deletion (ΔD2) mutant by introducing a stop codon at position 478. We found that the mutation of leucine residue at position 464 to alanine impaired the ERAD activity of p97 to an extent that was comparable to the deletion of the entire D2 domain, suggesting the vital role of this linker residue. In contrast, aside from the mutant L464A, all the other mutants showed marginal or no substrate accumulation when compared to the WT p97. In a previous structural analysis, it has been suggested that the central portion of the D1–D2 linker, especially the residue E470, may relay the motion from the D2 to the D1 domain (27). However, our data do not support the above proposal because mutant E470A did not cause any defect in ERAD activity (Fig. 3B). We also mutated this hydrophobic-leucine residue to a charged-residue glutamate (L464E) to change its polarity and to the cyclic residue proline (L464P) to interrupt its local structural pattern. As shown in Fig. 3B, similar to L464A, both L464E and L464P also exhibited dramatic accumulation of the ERAD substrate. These results suggest that both the hydrophobic property and a strict requirement for the structural arrangement at position 464 are crucial to maintain a properly functional protein.

Fig. 3.

L464 in the D1–D2 linker is essential to the p97 function. (A) Sequence alignment of the D1–D2 linker from different eukaryotic species, showing that the D1–D2 linker, especially the N-terminal half, is highly conserved. The identical residues are marked as +, whereas the conserved replacements are labeled as asterisk *. (B) Effects of mutation in the D1–D2 linker region degradation of the ERAD substrate TCR-α-GFP. (C) Spotting assay of Cdc48 mutant strains showing the effects of D1–D2 linker mutations of yeast Cdc48. The same amount of L126 carrying Cdc48 variants were collected at log phase. The cells was spotted on -Ura-His or -His supplemented with FOA plates and incubated at 25, 30, or 37 °C.

As noted, the D1–D2 linker of protein p97 is highly conserved from yeast to human. Therefore, we asked whether the deleterious effect of residue L464 mutation in p97 is conserved in its yeast orthologue, Cdc48. As shown in Fig. 3C, mutation of L474 to proline in yeast Cdc48 resulted in severe growth defects at 25, 30, and 37 °C. This observation clearly demonstrates that the physiological importance of residue L464 on the D1–D2 linker is conserved in yeast. It is of interest to note that for yeast Cdc48, mutation of L474 to glutamate resulted in only a mild growth defect, whereas mutant L474A had a similar growth rate to wild type. The significance of the difference in these mutational effects of mammalian p97 and yeast Cdc48 remains to be clarified.

To rule out the possibilities that the residue L464 may play a structural role or a role in ATPase activity, we performed extensive biochemical and biophysical characterization of the L464E mutant. We demonstrated that mutation on the D1–D2 linker residue L464 does not affect the oligomeric state, nucleotide-binding property, ATPase activity or secondary structure of p97 (see SI Results and Discussion for details). These results, combined with the observation that mutation of L464 led to a dramatic defect in ERAD activity of p97, led us to propose that residue L464 on the D1–D2 linker plays a vital role of in the motion transmission process.

Motion Transmission Between the D2 and D1 Domain Is Mediated by the D1–D2 Linker, Not from the Same Protomer, but from Its Neighboring Protomer.

In our hypothesis, as shown in Fig. 1B, both putative routes 2 and 3 utilized the D1–D2 linker to mediate the nucleotide-induced motion from the D2 to the D1 domains of p97. In route 2, the D1–D2 linker of the same protomer propagates the motion transmission; in route 3, the D1–D2 linker of the adjacent protomer transmits the conformational change. To determine which route is involved in motion communication between the D2 and D1 domain, we carried out further structure-based mutations to the corresponding regions and examined the mutational effects on the ERAD activities.

Route 2 is not responsible for motion transmission from the D2 to the D1 domain.

In the crystal structure of p97, L464 is not in close contact to any residues in the D2 domain within the same protomer. However, L464 is in proximity to three regions of the D1 domain: the first region near amino acid residues P246 and P247, the second region between a V367 and I371, and the third region near V407 (Fig. 4A) (23). We mutated these three regions in p97 and found that only marginal substrate accumulation were observed in mutants of I371A and V407E, whereas no substrate accumulation was observed for the other mutants (Fig. 4B). Therefore, the disruption of the interaction between the D1–D2 linker and D1 domain from the same protomer have no effect on the function of p97, suggesting route 2 is unlikely to be the pathway for nucleotide-induced motion transmission from the D2 to the D1 domain.

Fig. 4.

The essentiality of the interprotomer but not the intraprotomer Leu464-interacting residues demonstrates that only route 3 is functionally important. (A) Residues that are in close proximity to L464 of the same p97 protomer. (B) Mutations of p97 in L464-interacting regions in the same protomer show no effects on degradation of the ERAD substrate TCR-α-GFP. (C) Regions that are in close proximity to L464 of the neighboring p97 protomer. (D) Mutational effects in L464-interacting regions in the neighboring protomer on degradation of the ERAD substrate TCR-α-GFP.

Route 3 is the motion transmission pathway from the D2 to the D1 domain.

All experimental results described so far point to route 3 as the likely pathway; that is, the amino acid residue L464 does not functionally interact with the D1 or D2 domain from the same protomer, but with the D1 or D2 domain from an adjacent protomer. In the p97 hexamer crystal structure, L464 is in close proximity to the region around residue R358 in D1 domain, and to the region between amino acid residues N558 and D564 of the D2 domain of the adjacent protomer, as shown in Fig. 4C. Therefore, we carried out block mutagenesis in these regions and determined the mutational effect on ERAD. As shown in Fig. 4D, the alanine block mutations of residues 358–360AAA and 562–564AAA caused dramatic defects, which are comparable to the mutational effects of L464. With mutations at the D2 domain, we observed that the I562A and especially F563A mutations led to dramatic ERAD substrate accumulation, whereas there was no significant effect with the other mutations (Fig. 4D). This result suggests that the hydrophobic region of I562–F563 is involved in passing nucleotide information to the D1–D2 linker residue L464 of its neighboring protomer, likely via hydrophobic interactions. For mutations of the L464 interacting-region in the D1 domain, the mutation of R359A resulted in a marginal defect, whereas the other individual mutations did not cause any ERAD substrate accumulation. However, the block mutation of residues 358–360AAA showed a remarkable defect (Fig. 4D). Thus the region of 358–360 as a whole appears to be involved in the motion transmission, and it likely receives the motion information from the residue L464 of the neighboring protomer. The interaction between the region 358–360 and L464 appears rather intriguing because both R358 and R359 are positively charged. One explanation is that L464 interacts with F360 via hydrophobic interactions, whereas R358 and/or R359 might form hydrogen bonds with the linker residues N460 and S462 and thereby strengthen the interaction between the D1 domain and the D1–D2 linker of its neighboring protomer. Indeed, this hypothesis is supported by the result that block mutation of N460–S462 caused accumulation of about 40% more ERAD substrate than the wild type (Fig. 3B). In this scenario, individual mutations of R358, R359, or F360 would not disrupt the interaction between the D1 domain and the D1–D2 linker of its neighboring protomer, in agreement with the virtually intact ERAD activity of the single mutations (Fig. 4D). In contrast, the block mutation of the whole region of R358–F360 is expected to completely disrupt its interaction with the D1–D2 linker of the neighboring protomer, which is consistent to the observed striking defect in ERAD (Fig. 4D).

Interprotomer Motion Transmission Between the Two ATPase Domains of p97.

Based on the above data, we propose an interprotomer mechanism to delineate the motion transmission process between the D2 and D1 domains of p97, as shown in Fig. 5. In this process, the ATP hydrolysis-induced motion is transmitted from the D2 domain to the D1 domain through the D1–D2 linker of its neighboring protomer, rather than that of the same protomer. In another words, the actual motion transmission process does not occur in a straightforward perpendicular direction from the D2 domain via the D1–D2 linker to the D1 domain within the same protomer. Instead, it takes a zigzag pattern from one protomer to its neighboring protomer and then back to the original protomer. In this process, the D1–D2 linker serves as an interprotomer communicating “bridge,” receiving and transmitting the motion from the D2 to the D1 domain of its neighboring protomer.

Fig. 5.

Interprotomer motion transmission between the D2 and D1 domains of p97. AAA ATPase p97 is shown as a hexamer. The interprotomer motion-transmission pathway is shown for two protomers, which are highlighted in yellow and blue colors with domains/regions labeled. Red arrows indicate the direction of interprotomer motion transmission: From the D2 domain (yellow) to the D1–D2 linker of its neighboring protomer (blue) and then back to the D1 domain (yellow). The detailed motion transmission between the D1 and N domains remains unclear, as marked with a question mark.

Another prominent feature is that the α subdomain of the D1 domain is not directly involved in the motion-transmission pathway. This is different from the previously proposed preactive–active model, in which it was proposed that the ATP-hydrolysis-induced motion was transmitted from the α/β subdomain of the D2 to the D1–D2 linker of the same protomer, then through the α subdomain to the α/β subdomain of the D1 domain, and finally reached the remote N domain (25). In the model presented in this study, because both L464 interacting regions reside in the α/β subdomain of the D2 and D1 domains, respectively, the nucleotide-induced motion is transmitted directly from the α/β subdomain of the D2 domain to the α/β subdomain of the D1 domain via the D1–D2 linker of its neighboring linker, omitting the step of transmission to the α subdomain of D1 domain. We postulate that by taking this “shortcut,” the motion can be transmitted in a more efficient manner, facilitating the rapid action of p97.

It has been shown that the physiologically active form of p97 is a hexamer and that the monomeric p97 is functionally inactive (30), implying a key role of communication between neighboring protomers. The requirement of oligomerization is understood to be related to the transacting “arginine finger,” which senses the γ-phosphate of the adjacent promoter and promotes its hydrolysis (31–34). The interprotomer motion-transmission model proposed in this study provides another explanation why the activity of p97 is so critically dependent on its hexamerization. For many AAA ATPases, their strict symmetries may be abolished via evolutionary diversification, yet their oligomeric forms are almost invariantly preserved. Transacting mechanisms, be it ATPase activity or the motion transmission as found here in p97, is likely at the core of such extraordinary conservation of their oligomeric forms. It remains to be determined, however, whether the motion-transmission mechanism we propose for p97 can be generalized to the other AAA family proteins, especially the type-II AAA ATPases that contain tandem AAA ATPase domains.

Materials and Methods

Details of the materials and experimental procedures for characterization of p97 variants used are given in SI Materials and Methods, and additional references are also included in SI Results and Discussion.

Generation of p97 Mutants.

All mammalian p97 mutants were derived from the wild-type mouse-p97 (mp97) fusion constructs with pmcherry-p97 described as earlier (35). The pmCherry vector was a generous gift of Michael Frohman (Department of Pharmacology, Stony Brook University). Full-length p97 was fused to the C terminus of mCherry using the NotI and BamHI sites. All the site-directed mutants were generated using the QuickChange site-directed mutagenesis kit (Stratagene) and confirmed by DNA sequencing. All the mutated p97 fragments were amplified from pmcherry-p97 and ligated into the pET28a vector (Novagen).

FACS to Test GFP Fluorescence Density.

A two-color analysis of the transfected reporter cells was performed on a FACSCalibur flow cytometer (Becton Dickinson). Briefly, a reporter-cell line TG12 (HEK293 cells stably expressing the ERAD substrate TCR-α-GFP) (28) was used to transfect p97. After 24 h, the cells were resuspended in PBS buffer and analyzed by FACS to determine the effects of the mutation. The GFP fluorescence was determined in the red cells, which are indicators of the p97 positive cells. FACS gates were set to measure GFP fluorescence in at least three sets of 20,000 red cells and the mean GFP fluorescence was determined for each set.