Multiple UBX proteins reduce the ubiquitin threshold of the mammalian p97-UFD1-NPL4 unfoldase

  1. Ryo Fujisawa
  2. Cristian Polo Rivera
  3. Karim PM Labib  Is a corresponding author
  1. The MRC Protein Phosphorylation and Ubiquitylation Unit, School of Life Sciences, University of Dundee, United Kingdom

Abstract

The p97/Cdc48 ATPase and its ubiquitin receptors Ufd1-Npl4 are essential to unfold ubiquitylated proteins in many areas of eukaryotic cell biology. In yeast, Cdc48-Ufd1-Npl4 is controlled by a quality control mechanism, whereby substrates must be conjugated to at least five ubiquitins. Here, we show that mammalian p97-UFD1-NPL4 is governed by a complex interplay between additional p97 cofactors and the number of conjugated ubiquitins. Using reconstituted assays for the disassembly of ubiquitylated CMG (Cdc45-MCM-GINS) helicase by human p97-UFD1-NPL4, we show that the unfoldase has a high ubiquitin threshold for substrate unfolding, which can be reduced by the UBX proteins UBXN7, FAF1, or FAF2. Our data indicate that the UBX proteins function by binding to p97-UFD1-NPL4 and stabilising productive interactions between UFD1-NPL4 and K48-linked chains of at least five ubiquitins. Stimulation by UBXN7 is dependent upon known ubiquitin-binding motifs, whereas FAF1 and FAF2 use a previously uncharacterised coiled-coil domain to reduce the ubiquitin threshold of p97-UFD1-NPL4. We show that deleting the Ubnx7 and Faf1 genes impairs CMG disassembly during S-phase and mitosis and sensitises cells to reduced ubiquitin ligase activity. These findings indicate that multiple UBX proteins are important for the efficient unfolding of ubiquitylated proteins by p97-UFD1-NPL4 in mammalian cells.

Editor's evaluation

The p97/Cdc48 AAA ATPase and its heterodimeric ubiquitin adapter Ufd1-Npl4 unfold ubiquitylated proteins, often to segregate large protein complexes such as the MCM helicase at the termination of replication. The authors now demonstrate that an important difference exists between the yeast and metazoan system. While Cdc48-Ufd1-Npl4 can target MCM7 with a relatively short ubiquitin chain (5-7 units), the metazoan p97 requires either much longer chains or one of a group of accessory proteins that have previously been connected to various p97 and Ufd1-Npl4 mediated processes. The paper presents multiple lines of experimental evidence in support of the proposed ubiquitin-chain length model.

https://doi.org/10.7554/eLife.76763.sa0

Introduction

The regulated unfolding of proteins is essential for many aspects of cell biology and is mediated by multiple ATPases of the AAA + protein family (Puchades et al., 2020; Seraphim and Houry, 2020; Zhang and Mao, 2020). These comprise hexameric rings with a central pore, in which loops from each subunit form a ‘spiral staircase’ around the unfolded polypeptide chain. ATP hydrolysis drives conformational changes within the ATPase hexamer, leading to translocation of the unfolded polypeptide along a ‘conveyor belt’ of pore loops.

p97/Cdc48/VCP is a AAA + ATPase with orthologues in all three domains of life (Bodnar and Rapoport, 2017a; van den Boom and Meyer, 2018). Eukaryotic p97/Cdc48/VCP unfolds highly stable proteins to prepare them for subsequent degradation by the proteasome (van den Boom and Meyer, 2018; Verma et al., 2011; Wang et al., 2021). In addition, eukaryotic p97/Cdc48/VCP plays an essential role in diverse areas of cell biology by disrupting highly stable protein complexes or extracting proteins from cellular membranes. For example, p97/Cdc48/VCP is required for disassembly of the replisome at the end of chromosome replication (Dewar et al., 2017; Franz et al., 2011; Maric et al., 2014; Moreno et al., 2014; Sonneville et al., 2017; Villa et al., 2021), release of newly synthesised protein phosphatase I from an inhibitory protein complex (van den Boom et al., 2021; Weith et al., 2018), release of DNA repair complexes from chromosomes (Kilgas et al., 2021; Meerang et al., 2011; van den Boom et al., 2016), and the proteasomal degradation of misfolded membrane proteins (Ravanelli et al., 2020; Wolf and Stolz, 2012; Wu and Rapoport, 2018). Consistent with the important roles of p97/Cdc48/VCP in eukaryotic cell biology, dominant mutations in human p97 are associated with a late-onset multisystem proteinopathy (Johnson et al., 2010; Meyer and Weihl, 2014; Watts et al., 2004), whilst p97 is upregulated in certain cancer types (Li et al., 2021; Tsujimoto et al., 2004) and is a target for anti-cancer therapies (Anderson et al., 2015; Deshaies, 2014; Magnaghi et al., 2013; Wang et al., 2021). p97/Cdc48/VCP associates with a wide range of adaptor proteins, many of which are still understood poorly (Buchberger et al., 2015; Hänzelmann and Schindelin, 2017). Some recognise substrates directly (van den Boom et al., 2021; Weith et al., 2018), but others contain ubiquitin-binding motifs that link p97/Cdc48/VCP to poly-ubiquitylated substrates (Bandau et al., 2012; Buchberger et al., 2015; Stach and Freemont, 2017). The best characterised ubiquitin adaptors of p97/Cdc48/VCP are Ufd1 and Npl4, which form a heterodimeric receptor for polyubiquitin chains that include linkages via lysine 48 (K48) of ubiquitin (Bodnar and Rapoport, 2017a; Pan et al., 2021b). Studies of yeast Cdc48-Ufd1-Npl4 have shown that the Ufd1–Npl4 complex not only helps to recruit Cdc48 to polyubiquitylated substrates but is directly involved in the first step of protein unfolding, which initiates within the ubiquitin chain rather than the substrate itself and is independent of ATP hydrolysis by Cdc48 (Bodnar and Rapoport, 2017a; Ji et al., 2022; Twomey et al., 2019). In this way, Ufd1-Npl4 underpin a potentially universal mechanism for the recognition and unfolding of those Cdc48 substrates that are conjugated to K48-linked ubiquitin chains, without any requirement for direct recognition of the folded substrate protein itself (Ji et al., 2022; Twomey et al., 2019). It is likely that this mechanism exploits the reduced structural stability of ubiquitin within polyubiquitin chains (Carrion-Vazquez et al., 2003; Morimoto et al., 2015).

Structural studies suggested that the yeast Ufd1–Npl4 complex uses multiple ubiquitin-binding sites (Park et al., 2005; Sato et al., 2019) to interact with up to five ubiquitins within a K48-linked chain (Twomey et al., 2019). The functional significance of polyubiquitin binding was further demonstrated via reconstituted assays for disassembly of the 11-subunit CMG helicase with purified proteins (Deegan et al., 2020), in which displacement of the ubiquitylated CMG subunit provides a readout for the number of conjugated ubiquitins that are required for substrate unfolding. Such assays showed that yeast Cdc48-Ufd1-Npl4 can only process substrates with at least five conjugated ubiquitins (Deegan et al., 2020). This ‘five-ubiquitin threshold’ represents a quality control mechanism that protects substrates from premature unfolding, by coupling Cdc48 activity to efficient ubiquitylation. Until now, the ubiquitin threshold for substrate unfolding by mammalian p97-UFD1-NPL4 had not been determined.

Although Ufd1-Npl4 are essential for yeast Cdc48 and metazoan p97/VCP to unfold ubiquitylated substrates (Blythe et al., 2017; Bodnar and Rapoport, 2017b; Deegan et al., 2020; Mukherjee and Labib, 2019), eukaryotic cells including yeast and metazoa also encode a range of other adaptor proteins, which combine ubiquitin-binding motifs with a UBX domain that associates with the amino terminus of p97/Cdc48/VCP (Alexandru et al., 2008; Hänzelmann and Schindelin, 2017; Kloppsteck et al., 2012; van den Boom and Meyer, 2018). Strikingly, studies of such ubiquitin-binding adaptors of human p97 found that UBXN7/UBXD7, FAF1, FAF2/UBXD8, and UBXN1/SAKS1 all co-purified from human cells with UFD1-NPL4 in addition to p97 (Alexandru et al., 2008; Hänzelmann et al., 2011; Lee et al., 2013; Raman et al., 2015). Moreover, in vitro studies of purified FAF1 and UBXN7 showed that both factors associate preferentially with p97-UFD1-NPL4 compared to p97 alone (Hänzelmann et al., 2011). These findings suggested that the mechanism or regulation of p97-UFD1-NPL4 have additional layers of complexity that remain to be explored.

Here, we show that the segregase activity of mammalian p97-UFD1-NPL4 is controlled by a combination of ubiquitin chain length and multiple UBX proteins. In contrast to yeast Cdc48-Ufd1-Npl4, reconstituted human p97-UFD1-NPL4 only unfolds substrates that are conjugated to extremely long ubiquitin chains. However, association of p97-UFD1-NPL4 with the UBX proteins UBXN7, FAF1, or FAF2 reduces the ubiquitin threshold to a comparable level to yeast Cdc48-Ufd1-Npl4, dependent upon conjugating five ubiquitins to the substrate. In this way, such UBX proteins ensure that p97-UFD1-NPL4 can unfold ubiquitylated proteins with high efficiency in mammalian cells.

Results

Human p97-UFD1-NPL4 has a high ubiquitin threshold

To investigate how the ability of mammalian p97-UFD1-NPL4 to unfold substrate proteins is controlled by the number of conjugated ubiquitins, we adapted a reconstituted assay for the ubiquitylation and disassembly of budding yeast CMG helicase (Deegan et al., 2020), using purified recombinant versions of human p97 and UFD-NPL4 (Figure 1A, B). Initially, we used highly efficient ubiquitylation conditions (Deegan et al., 2020) in which ~two to three polyubiquitin chains were conjugated to the CMG-Mcm7 subunit, linked largely but not entirely via lysine 48 (K48) of ubiquitin (Figure 1—figure supplement 1A). The ubiquitylated helicase was then bound to beads that were coated with antibodies to the Cdc45 subunit, before incubation with human p97 and UFD1-NPL4 in the presence of ATP. Unfolding of ubiquitylated Mcm7 led to its release into the supernatant, together with all the other CMG subunits except for Cdc45 that remained bound to the antibody-coated beads (Figure 1C, compare lanes 1–2 and 3–4). Helicase disassembly required UFD1-NPL4 (Figure 1C, lanes 5 and 6) and was dependent upon ATP binding by the D1 and D2 ATPase domains of p97 (Figure 1D, E, Walker A mutations) together with ATP hydrolysis by the D2 ATPase (Figure 1D, E, Walker B mutation), as predicted by previous studies of the Cdc48/p97 unfoldase enzyme (Blythe et al., 2017; Bodnar and Rapoport, 2017b).

Figure 1 with 1 supplement see all
Reconstituted disassembly of ubiquitylated CMG helicase by purified human p97-UFD1-NPL4.

(A) Budding yeast CMG helicase (with an internal FLAG-tag on the Cdc45 subunit) was ubiquitylated in vitro, bound to beads coated with anti-FLAG antibody and then incubated with recombinant human p97-UFD1-NPL4 (see Materials and methods). Disassembly of ubiquitylated CMG displaced all subunits except Cdc45 into the supernatant. (B) Purified human p97 and UFD1-NPL4. (C) CMG disassembly reactions carried out according to the scheme in (A), in the presence of the indicated factors. (D) Purified recombinant human p97 with the indicated mutations in the D1 and D2 ATPase modules. (E) CMG disassembly reactions equivalent to those in (C), in the presence of the indicated factors.

Subsequently, we compared the ability of human p97-UFD1-NPL4 and yeast Cdc48-Ufd1-Npl4 to disassemble CMG complexes that had a single K48-linked chain of up to ~12 ubiquitins conjugated to CMG-Mcm7 (Figure 1—figure supplement 1B and Figure 2A, B, 2.5 nM E2 + 1 nM E3). As described previously (Deegan et al., 2020), yeast Cdc48-Ufd1-Npl4 disassembled CMG complexes with five or more ubiquitins conjugated to CMG-Mcm7, though disassembly was most efficient when the attached ubiquitin chains were much longer (Figure 2B, compare lanes 9–10 and 11–12). In contrast, although p97-UFD1-NPL4 rapidly disassembled CMG complexes with very highly ubiquitylated CMG-Mcm7 (Figure 2B, lanes 17–18 and Figure 2—figure supplement 1, lanes 7–9), the human enzyme was almost inactive against CMG complexes with up to ~12 ubiquitins conjugated to CMG-Mcm7 (Figure 2B, lanes 15–16 and Figure 2—figure supplement 1, lanes 1–3; also see below, Figure 4C, lanes 3–4), even in the presence of a large excess of unfoldase (Figure 2—figure supplement 2). These findings indicated that the yeast and human unfoldases have important differences in their requirements for processing ubiquitylated substrates.

Figure 2 with 3 supplements see all
Human p97-UFD1-NPL4 has a much higher ubiquitin threshold than yeast Cdc48-Ufd1-Npl4.

(A) Reconstituted CMG ubiquitylation reactions were performed under conditions that produced a single chain of up to ~12 ubiquitins on CMG-Mcm7 (2.5 nM E2, 1 nM E3; see Figure 1—figure supplement 1B), or 2–3 chains per CMG-Mcm7 (25 nM E2, 10 nM E3; see Figure 1—figure supplement 1A). (B) The products of the ubiquitylation reactions in (A) were bound to beads coated with anti-FLAG antibodies, before incubating as indicated with yeast Cdc48-Ufd1-Npl4 or human p97-UFD1-NPL4. CMG disassembly was monitored via displacement of subunits from beads (B) to supernatant (S), except for FLAG-tagged Cdc45 that remained bound to the anti-FLAG beads.

We then tested whether the preference of human p97-UFD1-NPL4 for highly ubiquitylated CMG was dependent on the attachment of multiple ubiquitin chains to CMG-Mcm7, since branched K48-linked ubiquitin chains were previously found to enhance the activity of human p97-UFD1-NPL4 (Blythe et al., 2017). However, CMG disassembly was equally efficient under conditions where a single very long ubiquitin chain was attached to CMG-Mcm7 (Figure 1—figure supplement 1C; Figure 2—figure supplement 3A). We also considered that the ability of p97-UFD1-NPL4 to disassemble highly ubiquitylated CMG was dependent upon the formation of mixed ubiquitin chains, which have been found to associate with p97 complexes in human cells (Yau et al., 2017). However, both human p97-UFD1-NPL4 and yeast Cdc48-Ufd1-Npl4 were able to disassemble highly ubiquitylated CMG complexes when the ubiquitin chains were exclusively K48 linked (Figure 2—figure supplement 3B, compare release of Mcm6 and Sld5 into the supernatant, in reactions with ‘K48-only’ ubiquitin). Therefore, these data indicate that human p97-UFD1-NPL4 can disassemble CMG complexes with very long K48-linked ubiquitin chains attached to CMG-Mcm7, regardless of the presence of multiple ubiquitin chains or other forms of ubiquitin chain linkage. In contrast, human p97-UFD1-NPL4 fails to disassemble CMG complexes with a single K48-linked chain of up to about 15 ubiquitins on CMG-Mcm7, contrasting with the five-ubiquitin threshold of yeast Cdc48-Ufd1-Npl4.

The UFD1-NPL4 adaptor complex sets the high ubiquitin threshold of human p97

To explore whether the high ubiquitin threshold of human p97-UFD1-NPL4 reflects the properties of p97 or its UFD1-NPL4 adaptor complex, we took advantage of the high sequence conservation between human p97 and yeast Cdc48 (Figure 3—figure supplement 1A) and performed reconstituted CMG disassembly reactions in which the unfoldase enzyme from one species was mixed with the UFD1-NPL4 adaptor from the other. Strikingly, human UFD1-NPL4 supported the disassembly of heavily ubiquitylated CMG, not only by human p97 (Figure 3A, lanes 5–6), but also by yeast Cdc48 (Figure 3A, lanes 7–8). Similarly, yeast Ufd1-Npl4 promoted the disassembly of heavily ubiquitylated CMG by both yeast Cdc48 (Figure 3A, lanes 3–4) and human p97 (Figure 3A, lanes 9–10). These findings reflected the conserved interactions between human NPL4/yeast Npl4 and the amino terminus of human p97/yeast Cdc48 (Bodnar et al., 2018; Pan et al., 2021a; Pan et al., 2021b).

Figure 3 with 2 supplements see all
The Ufd1-Npl4/UFD1-NPL4 adaptor complex determines the ubiquitin threshold of Cdc48/p97.

(A) CMG helicase was ubiquitylated as in Figure 1—figure supplement 1A, then bound to anti-FLAG beads. Disassembly reactions were then performed as indicated in the presence of yeast Cdc48 or human p97, combined with yeast Ufd1-Npl4 or human UFD1-NPL4. (B) Analogous reactions with CMG ubiquitylated as in Figure 1—figure supplement 1B.

Figure 4 with 3 supplements see all
UBXN7, FAF1, and FAF2 reduce the ubiquitin threshold of p97-UFD1-NPL4.

(A) Domain organisation of the indicated UBX proteins. UBA = ‘UBiquitin-Associated’ domain that binds ubiquitin; UBX = ‘UBiquitin regulatory X’ domain that binds p97; UBL = ‘UBiquitin-Like’ domain that binds ubiquitin and NEDD8; UAS = domain of unknown function in FAF1/FAF2/UBXN7. (B) Purified proteins – the membrane anchoring domain of FAF2 was deleted in FAF2∆M to facilitate expression of a soluble protein. (C–E) CMG disassembly reactions in the presence of the indicated factors, performed as described in Figures 13. See also Figure 4—figure supplements 13.

We then performed similar reactions with a CMG substrate that had a single ubiquitin chain of up to ~10 ubiquitins conjugated to the Mcm7 subunit (Figure 3B). Importantly, yeast Ufd1-Npl4 supported the disassembly of CMG complexes with five or more ubiquitins conjugated to CMG-Mcm7, not only by yeast Cdc48 (Figures 3B and 4), but also in the presence of human p97 (Figure 3B, lanes 9–10; yeast Ufd1-Npl4 was less efficient in combination with human p97 compared to reactions with yeast Cdc48, yet in both cases was selective for CMG-Mcm7 with five or more conjugated ubiquitins). In contrast, neither yeast Cdc48 nor human p97 disassembled CMG with 5–10 ubiquitins in reactions containing human UFD1-NPL4 (Figure 3B, lanes 7–8 and 5–6). These findings indicated that human UFD1-NPL4 and yeast Ufd1-Npl4 determine the ubiquitin threshold of their cognate unfoldase enzymes.

It is likely that human UFD1-NPL4 and yeast Ufd1-Npl4 employ a similar mechanism to initiate the unfolding of ubiquitylated substrates by p97/Cdc48. Both adaptor complexes bind to the amino terminal face of human p97/yeast Cdc48, with Npl4 forming a tower adjacent to the central p97 pore within which unfolded peptides are translocated (Bodnar et al., 2018; Pan et al., 2021a; Pan et al., 2021b). Moreover, human UFD1-NPL4 and yeast Ufd1-Npl4 utilise conserved ubiquitin-binding modules to recognise K48-linked ubiquitin chains, via multiple contacts including the top of the NPL4 tower and the UT3 domain of UFD1 (Ji et al., 2022; Pan et al., 2021b; Park et al., 2005; Pye et al., 2007; Sato et al., 2019; Twomey et al., 2019; Ye et al., 2003). Studies of the yeast unfoldase showed that binding of K48-linked ubiquitin chains to Ufd1-Npl4 led to unfolding of a ubiquitin moiety, which then interacts with a groove on the surface of the Npl4 tower before entering the central pore of Cdc48 (Twomey et al., 2019). Correspondingly, mutations of conserved residues in the Npl4 groove blocked substrate unfolding by yeast Cdc48-Ufd1-Npl4 (Twomey et al., 2019). We found that mutation of the equivalent residues in human NPL4 (Figure 3—figure supplement 2A, B, NPL4-AAE) blocked the disassembly of ubiquitylated CMG helicase (Figure 3—figure supplement 2C, compare lanes 3–4 and 7–8), without affecting the association of NPL4 with UFD1 (Figure 3—figure supplement 2B), or the interaction of UFD1-NPL4 with p97 (Figure 3—figure supplement 2D) or folded K48-linked polyubiquitin (Figure 3—figure supplement 2E). These data suggested that human p97-UFD1-NPL4 also initiates substrate unfolding via an unfolded ubiquitin intermediate, analogous to the action of yeast Cdc48-Ufd1-Npl4.

However, yeast Ufd1-Npl4 and human UFD1-NPL4 are considerably more divergent in primary sequence, compared to yeast Cdc48 and human p97 (Figure 3—figure supplement 1A, C), suggesting some degree of functional diversification. Moreover, human NPL4 has a ubiquitin-binding Zinc finger (‘NPL4-like Zinc Finger’ or NZF) at the carboxy-terminal end of the protein, which in vitro represents the highest affinity binding site for ubiquitin in human NPL4 (Meyer et al., 2002; Tsuchiya et al., 2017) and is not present in the yeast orthologue. Consistent with previous studies (Meyer et al., 2002), deletion of the NPL4-NZF did not impair association with UFD1 of p97 (Figure 3—figure supplement 2B, D) but largely abrogated the interaction of UFD1-NPL4 with K48-linked ubiquitin chains in vitro (Figure 3—figure supplement 2D). Nevertheless, the NPL-NZF was previously found to be dispensable in human cells for the degradation of misfolded proteins in the endoplasmic reticulum (Ye et al., 2003), a process known as ERAD that is dependent upon ubiquitylation and the p97-UFD1-NPL4 unfoldase. This suggested that the evolutionarily conserved low affinity binding sites for K48-linked ubiquitin chains in human UFD1-NPL4, at the top of the NPL4 tower and in the UT3 domain of UFD1 (Ji et al., 2022; Pan et al., 2021b; Park et al., 2005; Pye et al., 2007; Sato et al., 2019; Twomey et al., 2019; Ye et al., 2003), are sufficient to initiate the unfolding of ubiquitylated substrates of p97. Indeed, we found that human p97-UFD1-NPL4-∆NZF still supported the disassembly of heavily ubiquitylated CMG helicase in reconstituted in vitro assays, albeit with reduced efficiency compared to wild-type NPL4 (Figure 3—figure supplement 2C, compare lanes 1–6). Considered together with the data and previous observations discussed above, these findings indicated that the dynamic association of the NPL4 tower and UFD1-UT3 domain of human UFD1-NPL4 with very long K48-linked ubiquitin chains is sufficient to initiate substrate unfolding, dependent upon the NPL4 groove. Additional ubiquitin binding by the NPL4-NZF domain is not an essential part of the mechanism of substrate unfolding, but likely increases the efficiency of substrate engagement by human p97-UFD1-NPL4, as does increased length of the K48-linked ubiquitin chains.

UBXN7, FAF1, and FAF2 all reduce the ubiquitin threshold of mammalian p97-UFD1-NPL4

To process substrate proteins that are conjugated to shorter K48-linked ubiquitin chains, it is possible that p97-UFD1-NPL4 is aided in mammalian cells by additional adaptor proteins. In this regard, it is interesting that the UBX proteins FAF1, FAF2, UBXN7, and UBXN1 were previously found to co-purify with ubiquitin chains, p97 and UFD1-NPL4 from human cell extracts (Alexandru et al., 2008). Moreover, we found that all four factors bound directly to p97-UFD1-NPL4 in vitro (Figure 4—figure supplement 1; the association of each UBX protein with UFD1-NPL4 was dependent upon the presence of p97). Nevertheless, the interaction of UBXN1 with p97-UFD1-NPL4 was relatively weak, both in vitro with recombinant proteins (Figure 4—figure supplement 1) and in human cells (Alexandru et al., 2008).

To monitor the impact of UBXN1, FAF1, FAF2, and UBXN7 on the unfoldase activity of human p97-UFD1-NPL4, we purified the four UBX proteins (Figure 4A, B) and added them to reconstituted CMG disassembly assays. Reactions were performed under conditions where most CMG complexes had more than five ubiquitins conjugated to the Mcm7 subunit (Figure 4C, D), principally as K48-linked ubiquitin chains (Figure 1—figure supplement 1D) that were too short to support CMG disassembly by p97-UFD1-NPL4 alone (Figure 4C, lanes 1–4). UBXN1 did not significantly alter the in vitro activity of the p97-UFD1-NPL4 unfoldase (Figure 4C), consistent with the weak interaction of UBXN1 with p97-UFD1-NPL4. In contrast, FAF1, FAF2, and UBXN7 all stimulated the disassembly of ubiquitylated CMG by human p97 (Figure 4D, lanes 1–2, 5–6, and 9–10). In each case, disassembly was still dependent upon the presence of UFD1-NPL4 (Figure 4D, lanes 3–4, 7–8, and 11–12). These data indicated that FAF1, FAF2, and UBXN7 stimulate the unfoldase activity of p97-UFD1-NPL4 when the K48-linked ubiquitin chains attached to substrate are of limiting length.

Strikingly, helicase disassembly by human p97-UFD1-NPL4 in the presence of the three UBX proteins required five or more ubiquitins on the CMG-Mcm7 subunit (Figure 4E), thereby matching the minimal ubiquitin threshold of yeast Cdc48-Ufd1-Npl4 (Deegan et al., 2020) that is thought to reflect the presence of multiple conserved ubiquitin-binding domains within yeast Ufd1-Npl4 (Park et al., 2005; Sato et al., 2019; Twomey et al., 2019). This suggested that the UBX proteins function by stimulating productive interactions between the multiple ubiquitin-binding modules of human p97-UFD1-NPL4 and ubiquitylated substrate proteins. The initial engagement of UFD1-NPL4 with ubiquitylated substrates in the presence of UBX proteins should then lead to the trapping of an unfolded ubiquitin intermediate on the NPL4 groove. Consistent with this view, CMG disassembly in the presence of UBXN7, FAF1, and FAF2 was still dependent upon conserved residues in the NPL4 groove (Figure 4—figure supplement 2B; compare lanes 5–6 and 13–14).

The UBX proteins suppressed the partial defect in disassembling heavily ubiquitylated CMG that was observed in the absence of the ubiquitin-binding Zinc finger at the carboxyl terminus of NPL4 (Figure 4—figure supplement 2, compare release of Mcm6 and Sld5 into the supernatant in lanes 5–10). We also found that the human UBX proteins restored the ability of yeast Cdc48 to disassemble CMG complexes in the presence of human UFD1-NPL4, when the K48-linked ubiquitin chains on CMG-Mcm7 were otherwise too short to allow disassembly (Figure 4—figure supplement 3, lanes 1–6). Overall, these findings indicated that UBXN7, FAF1, and FAF2 augment the ability of human UFD1-NPL4 to initiate the unfolding of ubiquitylated substrates by p97.

The UBX, UBA, and UIM domains of UBXN7 all contribute to stimulation of p97-UFD1-NPL4 activity

To explore further how UBXN7 stimulates the unfolding of ubiquitylated CMG-Mcm7 by p97-UFD1-NPL4, we purified a series of truncated versions (Figure 5A, B). As predicted, all the truncated proteins still associated with p97-UFD1-NPL4, except for the version lacking the UBX domain (Figure 5—figure supplement 1A, B) that binds to the amino terminus of p97 (Li et al., 2017). We then tested the ability of the truncated UBXN7 proteins to support CMG disassembly by p97-UFD1-NPL4 (Figure 5C). The UBX domain of UBXN7 was essential for p97-UFD1-NPL4 to disassemble ubiquitylated CMG (Figure 5C, compare lanes 1–4), indicating that UBXN7 functions in direct association with p97-UFD1-NPL4. Moreover, efficient CMG disassembly was dependent upon both the ubiquitin-binding UBA domain (Figure 5C, lanes 5–6) and also the UIM domain (Figure 5C, lanes 7–8) that binds to both ubiquitin and NEDD8 (den Besten et al., 2012; Fisher et al., 2003; Young et al., 1998). Since the reconstituted CMG disassembly reactions lack NEDD8, these findings indicated that UBXN7 uses its UBA and UIM domains to stabilise productive interactions between the p97-UFD1-NPL4 unfoldase and its ubiquitylated substrate, thereby promoting the initiation of substrate unfolding via the NPL4 groove.

Figure 5 with 3 supplements see all
Mapping domains of UBXN7, FAF1, and FAF2 that stimulate the unfoldase activity of p97-UFD1-NPL4.

(A) Truncations of UBXN7. (B) Purified proteins. (C) CMG was ubiquitylated as in Figure 1—figure supplement 1D and then bound to anti-FLAG beads. Disassembly reactions were performed as above, in the presence of the indicated factors. (D–F) Equivalent analysis for FAF1. (G–I) Analogous truncations of FAF2 – ‘∆M’ indicates alleles that contain the amino terminus of the protein but lack the membrane anchoring domain. For (A), (D), and (G), the numbers correspond to residues in the full-length proteins. Domains were predicted using the SMART algorithm (http://smart.embl-heidelberg.de/) and Alphafold (Jumper et al., 2021).

The UBX and coiled-coil domains of FAF1 and FAF2 are required to stimulate p97-UFD1-NPL4 unfoldase activity

We then purified an equivalent series of truncated versions of FAF1 (Figure 5D, E), all of which still associated with p97-UFD1-NPL4 except for FAF1-∆UBX (Figure 5—figure supplement 1C, F). As above, the UBX domain was essential to stimulate segregase activity (Figure 5F, compare lanes 1–4), indicating that FAF1 functions in a complex with p97-UFD1-NPL4, similar to UBXN7. Surprisingly, however, the UBA domain of FAF1 was dispensable for stimulation of p97-UFD1-NPL4 (Figure 5F, FAF1-∆268, lanes 5–6). Moreover, a carboxy-terminal fragment of FAF1 that comprised the coiled-coil and UBX domains was just as effective as the full-length protein in promoting disassembly of ubiquitylated CMG by human p97-UFD1-NPL4 (Figure 5F, FAF1-∆480, lanes 7–8). In contrast, the UBX domain on its own was inactive (Figure 5F, FAF1-UBX, lanes 9–10). We note that the coiled coil is not required for the UBX domain of FAF1 to interact with p97-UFD1-NPL4 (Figure 5—figure supplement 1E, F). Furthermore, human p97 associates with UFD1-NPL4 in the absence of FAF1 or other UBX proteins (Figure 3—figure supplement 2D), as shown previously (Meyer et al., 2002), and FAF1 does not associate detectably with UFD1-NPL4 in the absence of p97 (Figure 5—figure supplement 1F, lane 1). These findings indicated that the coiled-coil domain of FAF1 has a previously unanticipated role in stimulating the unfoldase activity of p97-UFD1-NPL4.

We also generated analogous truncations of FAF2 (Figure 5G, H) and again found that the UBX domain was essential for stimulation of the p97-UFD1-NPL4 unfoldase (Figure 5I, compare lanes 1–4), whereas the UBA domain was dispensable (Figure 5I, FAF2-∆137, lanes 5–6). Furthermore, a FAF2 allele that lacked the coiled coil was inactive (Figure 5I, FAF2-∆CC∆M, lanes 9–10), whilst a fragment comprising the coiled-coil and UBX domains retained activity (Figure 5I, FAF2-∆297, lanes 7–8). These findings indicated that FAF2 and FAF1 share a common mechanism by which they stimulate the disassembly of ubiquitylated CMG by p97-UFD1-NPL4.

Finally, we tested whether the novel role for the coiled-coil domain of mammalian FAF1 and FAF2 is also conserved in the single C. elegans orthologue UBXN-3. In reconstituted reactions analogous to those described above, we found that a carboxy-terminal fragment of UBXN-3 comprising the coiled-coil and UBX domains was sufficient to promote the disassembly of ubiquitylated CMG helicase by C. elegans CDC-48_UFD-1_NPL-4 (Figure 5—figure supplement 2A, B, UBXN-3-∆435). Moreover, the UBX domain was essential for UBXN-3 to stimulate CDC-48_UFD-1_NPL-4 (Figure 5—figure supplement 2A, B, UBXN-3-∆UBX), as reported recently (Xia et al., 2021), but was insufficient in the absence of the coiled-coil domain (Figure 5—figure supplement 2A, B, UBXN-3-∆527). Therefore, the coiled-coil domain is essential for UBXN-3 to stimulate C. elegans CDC-48_UFD-1_NPL-4, mirroring the importance of the coiled-coil domains of mammalian FAF1 and FAF2. These findings indicate that metazoan FAF1/FAF2/UBXN-3 stimulate p97-UFD1-NPL4 activity by a common mechanism, involving direct binding to the unfoldase and an essential role for the coiled-coil domain of the UBX cofactor.

Cells lacking FAF1 and UBXN7 show defects in CMG helicase disassembly and are sensitive to inhibition of cullin ligase activity

Our data show that human p97-UFD1-NPL4 has a very high ubiquitin threshold, below which the unfoldase is dependent upon one of multiple UBX proteins, including UBXN7, FAF1, and FAF2. Such factors might act redundantly in mammalian cells, or else might stimulate p97-UFD1-NPL4 activity in distinct subcellular localisations. Previous work indicated that UBXN7 is largely nuclear in human cells (Raman et al., 2015). In contrast, FAF1 is present in both nucleus and cytoplasm (Raman et al., 2015), whereas FAF2 is a membrane anchored protein that localises to the endoplasmic reticulum, lipid droplets and the nuclear outer membrane (Go et al., 2021; Raman et al., 2015; Zehmer et al., 2009).

The function of human p97 and its cofactors is likely to be very highly conserved in other mammalian species, and we note that the mouse orthologues of p97-UFD1-NPL4, UBXN7, FAF1, and FAF2 are 97–100% identical to their human equivalents (Figure 6—figure supplements 12). To begin to assay for functional redundancy amongst the set of UBX proteins that stimulate mammalian p97-UFD1-NPL4 activity, we took advantage of the fact that mouse embryonic stem cells (mouse ES cells) were recently established (Villa et al., 2021) as a model system for studying mammalian CMG helicase disassembly. As in other metazoan species, mouse CMG is ubiquitylated on its MCM7 subunit during DNA replication termination, by the cullin ubiquitin ligase CUL2LRR1 (Villa et al., 2021). Ubiquitylated CMG is then disassembled very rapidly by p97-UFD1-NPL4, but the ubiquitylated helicase can be stabilised by treating cells with a small molecule inhibitor of the p97 ATPase (Villa et al., 2021).

CRISPR-Cas9 was used to make deletions in exon 1 of the Ubxn7 and Faf1 genes (Figure 6—figure supplement 3), using mouse ES cells in which both copies of the Gins4 (Sld5) gene had previously been modified to incorporate a GFP tag (Villa et al., 2021). The CMG helicase was then isolated from cell extracts, by immunoprecipitation of the GFP-tagged GINS4/SLD5 subunit. In extracts of control or FAF1∆ cells, CMG-MCM7 ubiquitylation was scarcely detectable (Figure 6A, lanes 5–6), reflecting the rapid and efficient disassembly of the ubiquitylated helicase by p97 (Villa et al., 2021). However, ubiquitylated CMG-MCM7 was observed in extracts of UBXN7∆ cells (Figure 6A, lane 7) and accumulated slightly in cells that lacked both FAF1 and UBXN7 (Figure 6A, lane 8). The observed defects could be rescued by expression of full-length UBXN7 or FAF1 in UBXN7∆ FAF1∆ cells, though FAF1-∆UBX or FAF1-∆Coiled Coil were unable to rescue (Figure 6—figure supplement 4A–D). These data indicated that nuclear UBXN7 contributes to the disassembly of ubiquitylated CMG helicase during DNA replication termination in mouse ES cells, with a minor contribution from FAF1. Nevertheless, ubiquitylated CMG further accumulated when control or UBXN7∆ FAF1∆ cells were treated with p97 inhibitor (Figure 6B, lanes 5–8; the short ubiquitin chains on CMG-MCM7 under such conditions might reflect the reduction of free ubiquitin pools upon p97 inhibition), suggesting that CMG disassembly in the absence of UBXN7 and FAF1 was not blocked completely. It is possible that additional UBX proteins not characterised in this study contribute to CMG disassembly in the absence of UBXN7 and FAF1. Alternatively, or in addition, our data indicate that extensive ubiquitylation of CMG-MCM7 should bypass the requirement for UBX proteins (Figure 1).

Figure 6 with 5 supplements see all
Partial redundancy between UBXN7 and FAF1 for CMG helicase disassembly during S-phase and mitosis in mouse ES cells.

(A) CMG was isolated from extracts of mouse embryonic stem cells (mES) with the indicated genotypes, by immunoprecipitation of GFP-tagged SLD5 subunit of the helicase. (B) Equivalent experiment in which cells were treated as indicated with 5 µM CB-5083 (p97i = p97 inhibitor) for 3 hr before harvesting. (C) Time-lapse video analysis of mitotic entry in mouse ES cells expressing GFP-SLD5, following inhibition of CUL2LRR1. Scale bars correspond to 10 µm. (D) Quantification of the data in (C). The individual data points from three independent experiments (>300 cells in total) are depicted as black dots in a scatter plot. The mean values from each experiment are shown in red circles, whilst red bars and error bars represent the average of the mean values and the associated standard deviations (n = 3). The samples were then compared by a Kruskal–Wallis test followed by Dunn’s test, yielding the indicated p values. See also Figure 6—figure supplements 35 (note that Figure 6—figure supplement 4E confirms that the difference between FAF1∆ UBXN7∆ and cells lacking either FAF1 or UBXN7 is statistically significant, despite the weaker significance of the data in (D)).

A second pathway for the disassembly of ubiquitylated CMG helicase acts during mitosis in metazoan cells and is important to process sites of incomplete DNA replication (Deng et al., 2019; Sonneville et al., 2019; Sonneville et al., 2017; Villa et al., 2021). This pathway is independent of CUL2LRR1 and instead requires the TRAIP ubiquitin ligase (Deng et al., 2019; Sonneville et al., 2019; Sonneville et al., 2017; Villa et al., 2021), which is mutated in human patients with a form of primordial dwarfism syndrome (Harley et al., 2016). To test whether FAF1 and UBXN7 also contribute to the mitotic disassembly of ubiquitylated CMG, we inactivated CUL2LRR1 in mouse ES cells and monitored the presence on mitotic chromatin of GFP-SLD5. Time-lapse video analysis showed that CMG disassembly upon mitotic entry was blocked in the absence of TRAIP (Figure 6—figure supplement 5) as shown previously (Villa et al., 2021) and was slower in FAF1∆ UBXN7∆ compared to FAF1∆, UBXN7∆, or control cells (Figure 6C, D). This defect was efficiently rescued by expression of either UBXN7 or FAF1, but not by FAF1-∆UBX or FAF1-∆CC (Figure 6—figure supplement 4E). These data indicated that TRAIP-dependent CMG disassembly by p97 was delayed in cells that lacked both FAF1 and UBXN7, which act redundantly in the mitotic CMG disassembly pathway in mouse ES cells.

These findings suggested that the unfolding of other ubiquitylated substrates of p97-UFD1-NPL4 should also be impaired in the absence of FAF1 and UBXN7, particularly when ubiquitin chain formation on substrates of p97-UFD1-NPL4 is suboptimal. Cullin ligases represent the largest family of E3 enzymes directing K48-linked ubiquitin chain formation in eukaryotic cells (Harper and Schulman, 2021) and their activity is dependent upon neddylation of the cullin scaffold. Correspondingly, we found that FAF1∆ UBXN7∆ cells showed enhanced sensitivity to the neddylation inhibitor MLN4924 (Soucy et al., 2009), compared to control cells (Figure 7A, B), likely reflecting the defective processing of multiple targets of p97-UFD1-NPL4 in diverse aspects of cell biology. This defect was rescued by expression of wild type FAF1 protein but not FAF1-∆UBX or FAF1-∆CC (Figure 7C). Moreover, expression of UBXN7 restored the sensitivity of FAF1∆ UBXN7∆ cells to the level to FAF1∆ cells (compare Figure 7B, D). Overall, these findings suggest that the unfolding of ubiquitylated substrates by mammalian p97-UFD1-NPL4 is modulated both by the extent of ubiquitylation and by the action of multiple UBX proteins.

Cells lacking FAF1 and UBXN7 are sensitive to global inhibition of cullin ligase activity.

(A) Cells of the indicated genotypes were treated as shown and then grown on 10 cm plates for 6 days. Surviving colonies were fixed and stained with crystal violet. (B) Viability of cells treated with 0, 3, or 6 µM MLN-4924 for 6 hr. The data represent the mean values and standard deviation for three independent experiments (n = 3). The mean survival ratios of wt and FAF1∆ UBXN7∆ after treatment with 6 µM MLN-4924 condition were compared via a two-tailed t-test. (C) Analogous experiment involving FAF1∆ UBXN7∆ mouse ES cells expressing the indicated versions of human FAF1 (as in Figure 6—figure supplement 4B) at the Rosa26 locus. (D) Similar experiment to that in (C), comparing wt mouse ES cells, FAF1∆ UBXN7∆ cells, and FAF1∆ UBXN7∆ cells expressing human UBXN7 from the Rosa26 locus.

Discussion

Together with previous studies of the conserved ubiquitin-binding modules of human UFD1-NPL4 (Pan et al., 2021b; Pye et al., 2007; Ye et al., 2003), our finding that the NPL4 groove is essential for CMG helicase disassembly in reconstituted reactions (Figure 3—figure supplement 2) indicates that human p97-UFD1-NPL4 initiates substrate unfolding by a fundamentally similar mechanism to yeast Cdc48-Ufd1-Npl4 (Ji et al., 2022; Park et al., 2005; Twomey et al., 2019). It is likely that the engagement of UFD1-NPL4 with K48-linked poly-ubiquitin leads to initial ubiquitin unfolding and translocation through the central pore of p97/Cdc48, followed by unfolding of the ubiquitylated polypeptide.

However, human p97-UFD1-NPL4 is dependent upon the conjugation of very long ubiquitin chains to the substrate, in the absence of other factors (Figure 2; Figure 2—figure supplements 1 and 2). Our data indicate that this high ubiquitin threshold is set by UFD1-NPL4 rather than by p97 itself (Figure 3). Although the NPL4-NZF domain is responsible for the majority of detectable ubiquitin binding by NPL4 in our assays (Figure 3—figure supplement 2E), as seen in past studies (Meyer et al., 2002; Tsuchiya et al., 2017), it is largely dispensable for CMG disassembly and thus for substrate unfolding by p97-UFD1-NPL4 (Figure 3—figure supplement 2C), consistent with previous findings that the NPL4-NZF is not required for ERAD in human cells (Ye et al., 2003). These findings indicate that the dynamic binding to ubiquitylated substrate of conserved ubiquitin-binding modules within the NPL4 tower and the UFD1-UT3 domain (Pan et al., 2021b; Pye et al., 2007), favoured by the presence of very long K48-linked ubiquitin chains, is sufficient to initiate substrate unfolding by p97.

Our findings indicate that UBX proteins not only help to target p97 to substrates in a variety of subcellular locations but also play a direct role in the unfolding reaction, particularly when the ubiquitin chains conjugated to the substrate are short. Stimulation of p97-UFD1-NPL4 by UBX proteins is dependent not only on the presence of UFD1-NPL4 (Figure 4) but on conserved residues in the NPL4 groove (Figure 4—figure supplement 2, NPL4-AAE), indicating that substrate processing under such conditions is initiated by the trapping of an unfolded ubiquitin intermediate, as for yeast Cdc48-Ufd1-Npl4. UBXN7, FAF1, and FAF2 all bind directly to p97-UFD1-NPL4 (Figure 4—figure supplement 1; Alexandru et al., 2008), in each case using their UBX domain to associate with the amino terminal face of the p97 hexamer (Kim et al., 2011; Li et al., 2017). We propose that these UBX proteins promote the formation of productive complexes between p97-UFD1-NPL4 and polyubiquitylated substrates, to facilitate the initiation of substrate unfolding. Consistent with this view, the ubiquitin threshold of human p97-UFD1-NPL4 in the presence of the UBX proteins is very similar to yeast Cdc48-Ufd1-Npl4 (Figure 4E), likely reflecting the engagement of the K48-linked ubiquitin chain by the multiple conserved ubiquitin-binding modules within human UFD1-NPL4.

In the case of UBXN7, stimulation of p97-UFD1-NPL4 in reconstituted assays is dependent not only on the UBX domain but also on the UBA and UIM motifs that bind ubiquitin (Figure 5A–C). This indicates that UBXN7 functions by binding simultaneously to p97-UFD1-NPL4 and the polyubiquitin chain on the substrate, thereby promoting the trapping of an unfolded ubiquitin intermediate on the NPL4 groove. It is possible that FAF1 and FAF2 function in an analogous manner, but we were unable to detect stable binding of the FAF1 coiled coil to K48-linked ubiquitin chains in vitro (Figure 5—figure supplement 3). However, the same is true for UFD1-NPL4-∆NZF (Figure 3—figure supplement 2E), which nevertheless contains conserved low affinity ubiquitin-binding modules (Pan et al., 2021b; Pye et al., 2007) and supports the disassembly of ubiquitylated CMG in reconstituted assays (Figure 3—figure supplement 2C). Therefore, we cannot exclude that the FAF1 coiled coil also functions by binding to both p97-UFD1-NPL4 and ubiquitin. Structural studies of early intermediates in the unfolding reaction will be important in future studies, to elucidate further the underlying mechanism.

Our data indicate some degree of functional redundancy between UBXN7 and FAF1 in mammalian cells (Figures 6 and 7), which is likely to be mitigated by differences in their subcellular localisation (Raman et al., 2015), together with additional unique features of each factor such as the preferential association of the UBXN7-UIM domain with neddylated cullins (den Besten et al., 2012). The growth of UBXN7∆ FAF1∆ cells is further impaired in the absence of FAF2 (Figure 6—figure supplement 3F), which likely regulates p97-UFD1-NPL4 on the surface of the endoplasmic reticulum, lipid droplets, and the nuclear outer membrane (Go et al., 2021; Raman et al., 2015; Zehmer et al., 2009). The remaining activity of p97-UFD1-NPL4 in UBXN7∆ FAF1∆ FAF2∆ cells might involve additional UBX proteins that associate with p97-UFD1-NPL4 (Raman et al., 2015) and remain to be characterised in vitro. However, our findings also indicate (Figure 1) that substrates with very long ubiquitin chains can be unfolded by p97-UFD1-NPL4 in the absence of UBX proteins, consistent with previous in vitro assays with an artificial model substrate (Blythe et al., 2017). Nevertheless, Ubxn7, Faf1, and Faf2 all become essential for viability during mouse development or soon after birth (Adham et al., 2008; Koscielny et al., 2014), indicating their importance for the biology of mammalian p97-UFD1-NPL4. Similarly, the ubxn-3 orthologue of mammalian FAF1 is important for embryonic viability and adult fertility during C. elegans development (Xia et al., 2021). Our data suggest that UBX proteins such as UBXN7, FAF1, and FAF2 all increase the efficiency of substrate unfolding by p97-UFD1-NPL4, in diverse areas of eukaryotic cell biology. We suggest that such factors are particularly important whenever the length of the ubiquitin chain on the substrate is below the inherently high ubiquitin threshold of the mammalian unfoldase.

Materials and methods

Reagents and resources used in this study, including all purified proteins, are listed in Supplementary file 1.

Plasmid DNA construction

Request a detailed protocol

The plasmids generated in this study are shown in Supplementary file 2. PCRs were conducted with Phusion High-Fidelity DNA Polymerase (New England Biolabs, M0530), and amplified DNA fragments were cloned via the Gibson Assembly Cloning Kit (New England Biolabs, E2611) or with restriction enzymes and T4 ligase. All new constructs were verified by Sanger sequencing. Site-specific mutagenesis was performed using the Phusion polymerase, according to the manufacturer’s protocol.

Coding sequences of human p97 (Uniprot identifier P55072-1), UFD1 (Uniprot identifier Q92890-2), NPL4 (Uniprot identifier Q8TAT6-1), FAF1 (Uniprot identifier Q9UNN5-1), and UBXN7 (Uniprot identifier O94888-1) were amplified from XpressRef Universal Total human RNA (QIAGEN, 338112) by RT-PCR (TaKaRa, RR014) and cloned by Gibson assembly into the vector pK27SUMO (expressing a 14His-tagged version of the yeast SUMO protein Smt3) or pET28c. The coding sequences of human UBXN1 (Uniprot identifier Q04323-1) and FAF2 (Uniprot identifier Q96CS3-1) were synthesised by GenScript Biotech. For FAF2, the residues encoding the membrane anchoring domain (amino acids 90–118) were removed by PCR to improve solubility of the expressed protein.

Expression of proteins in bacterial cells

Request a detailed protocol

Each plasmid was transformed into Rosetta (DE3) pLysS (Novagen, 70956), which was grown in LB medium supplemented with kanamycin (50 µg/ml) and chloramphenicol (35 µg/ml). Subsequently, a 100-ml culture was grown overnight at 37°C with shaking at 200 rpm. The following morning, the culture was diluted fivefold with 400 ml of LB medium supplemented with kanamycin (50 µg/ml) and chloramphenicol (35 µg/ml) and then left to grow at 37°C until an OD600 of 0.8 was reached. Protein expression was then induced overnight at 18°C in the presence of 1 mM IPTG (Isopropyl β- d-1-thiogalactopyranoside). Cells were harvested by centrifugation for 15 min in a JLA-9.1000 rotor (Beckman) at 5180 × g and the pellets were stored at −20°C.

Protease inhibitors

Request a detailed protocol

One tablet of ‘Roche cOmplete EDTA-free protease inhibitor Cocktail’ (Roche, 11873580001) was either added directly to 20 ml of protein purification buffer or else dissolved in 1 ml of water to make 20× stock (EDTA = ethylenediaminetetraacetic acid).

Buffers

Lysis buffer: 50 mM Tris–HCl (pH 8.0), 0.5 M NaCl, 5 mM Mg(OAc)2, and 0.5 mM TCEP (Tris Carboxy Ethyl Phosphene).

Gel filtration buffer: 20 mM HEPES–KOH (pH 7.9; HEPES = (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid)), 0.15 M NaCl, 5 mM Mg(OAc)2, 0.3 M sorbitol, and 0.5 mM TCEP.

Reaction buffer: 25 mM HEPES–KOH (pH 7.9), 10 mM Mg(OAc)2, 0.02% (wt/vol) IGEPAL CA-630, 0.1 mg/ml bovine serum albumin (BSA), and 1 mM DTT (dithiothreitol).

Wash buffer: 100 mM HEPES–KOH (pH 7.9), 100 mM KOAc, 10 mM Mg(OAc)2, 2 mM EDTA, 0.1% (wt/vol) IGEPAL CA-630.

Cell Lysis buffer: 100 mM HEPES–KOH (pH 7.9), 100 mM KOAc, 10 mM Mg(OAc)2, 2 mM EDTA, 10% (wt/vol) glycerol, 0.1% (wt/vol) Triton X-100, 2 mM sodium fluoride, 2 mM sodium β-glycerophosphate pentahydrate, 10 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1 µg/ml LR-microcystin, 1 mM DTT, and Roche cOmplete EDTA-free protease inhibitor Cocktail.

Purification of human p97 and its ATPase mutants

Request a detailed protocol

Bacterial pellets expressing 14His-Smt3-p97 from 500 or 1000 ml cultures were resuspended in 20 ml Lysis buffer supplemented with 40 mM imidazole, 0.1 mM ATP, Roche cOmplete EDTA-free protease inhibitor Cocktail. The samples were lysed by incubation with 1 mg/ml lysozyme on ice for 30 min, followed by sonication for 90 s (15 s on, 30 s off) at 40% on a Branson Digital Sonifier, then clarified by centrifugation at 10,000 × g for 30 min in an SS-34 rotor (Sorvall).

The supernatant was mixed with 1 ml resin volume of Ni-NTA beads (Qiagen, 30210) which was pre-equilibrated with Lysis buffer supplemented with 40 mM imidazole and 0.1 mM ATP. After 1-hr incubation at 4°C, beads were recovered in a disposable gravity flow column and washed with 150 ml of Lysis buffer containing 40 mM imidazole and 0.1 mM ATP, followed by 10 ml of Lysis buffer containing 60 mM imidazole, 2 mM ATP, and 10 mM MgOAc. 14His-Smt3p97 was eluted with 5 ml of Lysis buffer containing 400 mM imidazole and 0.1 mM ATP. Ulp1 protease (10 µg/ml) was added to the eluate and incubated at 4°C overnight, to cleave the 14His-Smt3 tag from p97. The sample was then loaded onto a 24 ml Superose 6 column equilibrated in Gel filtration buffer containing 0.1 mM ATP. Fractions containing hexameric p97 were pooled, concentrated with an Amicon Ultra-4 Centrifugal Filter Unit (Merck, UFC803024) according to the manufacture’s protocol, aliquoted and snap-frozen with liquid nitrogen.

Purification of human UFD1-NPL4

Request a detailed protocol

Human untagged NPL4 (or 6HisNPL4) and 14His-Smt3UFD1 were expressed individually in 750 and 250 ml cultures (at a 3:1 ratio of NPL4 to UFD1). The respective bacterial pellets were then resuspended in 15 and 5 ml of Lysis buffer containing 30 mM imidazole, Roche protease inhibitor tablets. The samples were then mixed and lysed by incubation with 1 mg/ml lysozyme on ice for 30 min, followed by sonication for 90 s (15 s on, 30 s off) at 40% on a Branson Digital Sonifier, before clarification by centrifugation at 10,000 × g for 30 min in an SS-34 rotor.

The supernatant was mixed with 1 ml resin volume of Ni-NTA beads which had been pre-equilibrated with Lysis buffer containing 30 mM imidazole. After 1-hr incubation at 4°C, beads were recovered in a disposable gravity flow column and washed with 150 ml of Lysis buffer containing 30 mM imidazole, followed by 10 ml of Lysis buffer containing 50 mM imidazole, 2 mM ATP, and 10 mM MgOAc. The complex of 14His-Smt3UFD1 with NPL4 was eluted with 5 ml of Lysis buffer containing 400 mM imidazole.

The yeast Ulp1 protease (10 µg/ml) was then added to cleave the 14His-Smt3 tag from UFD1, and the mixture was dialyzed into 500 ml of Lysis buffer containing 30 mM imidazole at 4°C overnight. The dialyzed mixture was mixed with 0.5 ml Ni-NTA beads equilibrated with Lysis buffer containing 30 mM imidazole. After 15 min of rotation at 4°C, the flow-through fraction was collected and concentrated with an Amicon Ultra-4 Centrifugal Filter Unit according to the manufacture’s protocol and then loaded onto a 24-ml Superdex 200 column equilibrated in Gel filtration buffer. Fractions containing UFD1-NPL4 were pooled, concentrated, aliquoted, and snap-frozen with liquid nitrogen.

For purification of 14His-Smt3UFD1-NPL4 proteins in Figure 3—figure supplement 2, the tag cleavage by Ulp1 and dialysis steps were omitted and purified similarly.

Purification of metazoan UBX proteins

Request a detailed protocol

Bacterial pellets expressing 14His-Smt3-tagged UBX proteins from 250 to 1000 ml cultures were resuspended in 20 ml of Lysis buffer containing 30 mM imidazole and Roche cOmplete EDTA-free protease inhibitor Cocktail. The samples were lysed by incubation with 1 mg/ml lysozyme on ice for 30 min, followed by sonication for 90 s (15 s on, 30 s off) at 40% on a Branson Digital Sonifier, before clarification by centrifugation at 10,000 × g for 30 min in an SS-34 rotor.

The supernatant was mixed with 1 ml resin volume of Ni-NTA beads which was pre-equilibrated with Lysis buffer containing 30 mM imidazole. After 1-hr incubation at 4°C, beads were recovered in a disposable gravity flow column and washed with 150 ml of Lysis buffer containing 30 mM imidazole, followed by 10 ml of Lysis buffer containing 50 mM imidazole, 2 mM ATP, and 10 mM MgOAc. 14His-Smt3-tagged UBX proteins were eluted with 5 ml of Lysis buffer containing 400 mM imidazole. Ulp1 protease (10 µg/ml) was then added to cleave the 14His-Smt3 tag from UBX proteins during an overnight incubation. Subsequently, the samples were concentrated with an Amicon Ultra-4 Centrifugal Filter Unit according to the manufacture’s protocol and loaded onto a 24-ml Superdex 200 column in Gel filtration buffer. In the case of the fragments comprising the UBX domain of FAF1 and UBXN-3, a Superdex75 column was used instead of Superdex 200. The peak fractions containing each UBX protein were pooled, aliquoted, and snap-frozen with liquid nitrogen.

For purification of 14His-Smt3FAF1 proteins shown in Figure 5—figure supplement 3, eluted proteins from the Ni-NTA beads were dialysed into Gel filtration buffer, and snap-frozen with liquid nitrogen.

Generation of a budding yeast strain expressing CMG-Mcm7-KR

Request a detailed protocol

A yeast strain coexpressing the 11 subunits of budding yeast CMG was generated as described previously (Deegan et al., 2020; Zhou et al., 2017), with the following modification. The original plasmid expressing MCM7 and MCM6 genes from the bi-directional GAL1,10 promoter at the leu2 locus (pJF4, Frigola et al., 2013) was replaced with a modified version, which introduced mutations in surface lysines of Mcm7 (pCPR037, see Supplementary file 1, Supplementary file 2) that restricted CMG ubiquitylation by SCFDia2 to a single chain per CMG-Mcm7 subunit (Figure 1—figure supplement 1C). In the resulting strain (yCPR337, see Supplementary file 1) protein expression for each CMG subunit was monitored by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblotting.

Purification of wild-type yeast CMG and CMG-Mcm7-KR

Request a detailed protocol

Budding yeast cells (yTDK20 for wt CMG, yCPR337 for CMG-Mcm7-KR) were grown at 30°C in rich medium (1% [wt/vol] yeast extract, 212750, Becton Dickinson and 2% [wt/vol] bacteriological peptone, LP0037B, Oxoid) supplemented with 2% raffinose. Once the cultures reached a cell density of 2–3 × 107 cells/ml, galactose was added to 2% to induce expression of CMG subunits from the GAL1,10 promoter, for 3 hr at 30°C. Subsequently, cells were collected by centrifugation at 3315 × g for 10 min and washed once in buffer containing 25 mM HEPES–KOH (pH 7.6), 10% glycerol, 0.02% Tween-20, 2 mM MgOAc, 1 mM DTT, and 300 mM KCl (CMG buffer/300 mM KCl). After the wash, cell pellets were resuspended in 0.3 volumes of CMG buffer/300 mM KCl supplemented with 1× Complete Protease Inhibitor Cocktail (11873580001, Roche; one tablet dissolved in 1 ml water makes 25× stock solution). The cell suspension was then frozen dropwise in liquid nitrogen to make ‘yeast popcorn’. Subsequently, the frozen yeast popcorn was ground in a freezer mill (SPEX CertiPrep 6850 Freezer Mill) using four cycles of 2 min at a rate of 15 cycles per second. The resulting powder was stored in −80°C until required.

Thawed yeast cell powder was resuspended with an equivalent volume of CMG buffer/300 mM KCl containing protease inhibitors and the sample was centrifuged at 235,000 × g at 4°C for 1 hr. The soluble extract was recovered and mixed with 3 ml of anti-FLAG M2 affinity resin (A2220, Sigma), to bind CMG via an internal FLAG tag on the Cdc45 subunit, and the mixture was then incubated at 4°C for 2 hr on a rotating wheel. The resin was then recovered and washed extensively with 200 ml of CMG buffer/300 mM KCl lacking protease inhibitors. Elution was then performed by mixing the beads with 5 ml of CMG buffer/300 mM KCl containing 0.5 mg/ml 3XFLAG peptide (F4799, Sigma), and with another 5 ml of CMG buffer/300 mM KCl containing 0.25 mg/ml 3XFLAG peptide. The recovered 10 ml eluate was supplemented with 2 mM CaCl2 and mixed with 1 ml Calmodulin Sepharose 4B (7052901, Cytiva) for 1 hr at 4°C to capture purified CMG via CBP-tagged Mcm3 subunit. The resin was then collected and washed extensively with 100 ml of CMG buffer/300 mM KCl supplemented with 2 mM CaCl2 and Roche protease inhibitors. CMG complexes were then eluted in 15 ml of CMG buffer/300 mM KCl containing 2 mM EDTA and 2 mM EGTA (Ethyleneglycol Bis(2-Aminoethyl Ether)-N,N,N’,N’ Tetraacetic Acid). Eluate fractions were loaded onto a 0.24 ml MiniQ column equilibrated in CMG buffer/300 mM KCl and the helicase complex was eluted using a 4 ml gradient ranging from 300 to 600 mM KCl in CMG buffer. Fractions containing CMG complexes were pooled and dialysed against CMG buffer/200 mM KOAc for 4 hr at 4°C, recovered, aliquoted and snap frozen, before storage at −80°C.

Purification of budding yeast proteins

Request a detailed protocol

The budding yeast proteins used in this study (replisome factors, ubiquitylation factors, and Cdc48 plus its Ufd1-Npl4 adaptors) are described in Supplementary file 1. The various factors were expressed using plasmids and strains described in Supplementary file 1 and were purified as described previously (Deegan et al., 2020).

Purification of C. elegans proteins

Request a detailed protocol

The C. elegans proteins used in this study (replisome factors, ubiquitylation factors, neddylation factors, and CDC-48 plus adaptors) are described in Supplementary file 1. The various factors were expressed using plasmids and strains described in Supplementary file 1 and were purified as described previously (Xia et al., 2021), except for UBXN-3-∆435 and UBXN-3-∆527 that are depicted in Figure 5—figure supplement 2 and were expressed as fusions to 14His-Smt3 in E. coli using the plasmids pRF056 and pRF057, as described above for other UBX proteins.

Preparation of IR Dye-labelled K48-linked ubiquitin chains

Request a detailed protocol

Fluorescently labelled K48-linked ubiquitin chains were synthesised as described previously (Abdul Rehman et al., 2021; Abdul Rehman et al., 2016) with slight modifications. K48-linked ubiquitin chains were synthesised in a reaction mix containing 2500 μM wt ubiquitin, 250 μM 6His-ubiquitin (K48R, K63C), 0.5 μM UBE1, 15 μM UBE2R1, 10 mM ATP, 50 mM Tris–HCl (pH 7.5), 10 mM MgCl2, and 0.6 mM DTT with overnight incubation at 30°C. Chains with successful incorporation of 6His-ubiquitin (K48R, K63C) at their distal end were purified using a HisTrap FF 5 ml column (17525501, Cytiva). Subsequently, ubiquitin chains were fractionated using a Superdex 200 16/60 column. Fractions containing chains with 7–20 ubiquitins were pooled and concentrated, before buffer exchange into phosphate-buffered saline (PBS). The purified ubiquitin chains were then reacted with a threefold molar excess of IRDye-800CW Maleimide (929-80020, Li-Cor) for 3 hr at 22°C with gentle agitation. The reaction was quenched with 50 mM β-mercaptoethanol. Subsequently, 500 µl of the labelled ubiquitin chains (5.5 mg/ml) were incubated with 20 µl of Precission protease (2.5 mg/ml) at 4°C overnight to cleave the 6His-tag off ubiquitin, then incubated with 500 µl slurry of Ni-NTA beads at 4°C for 30 min. The unbound fraction was then concentrated and buffer exchanged into 50 mM Tris–HCl (pH 7.5), before being aliquoted and snap-frozen with liquid nitrogen.

Reconstituted ubiquitylation and disassembly of yeast CMG helicase

Request a detailed protocol

Yeast CMG was ubiquitylated in reconstituted in vitro reactions as described previously (Deegan et al., 2020). Reactions were set up using ‘Reaction buffer’ containing 150 mM KOAc and 2 mM ATP, in a final volume of 10 µl and a final concentration of 15 nM yeast CMG, 30 nM Uba1, the concentrations of the E2 Cdc34 and E3 SCFDia2 that are discussed below, 30 nM Ctf4 and 6 µM ubiquitin. Concentrations of E2 and E3 were as specified in the figures and figure legends. Ubiquitylation reactions were conducted at 30°C for 20 min and then stopped by the addition of KOAc to a final concentration of 700 mM.

The ubiquitylated yeast CMG was then incubated for 30 min at 4°C with 2.5 µl (per 10 µl ubiquitylation reaction mix) magnetic beads (Dynabeads M-270 Epoxy; Life Technologies, 14302D) that had been conjugated to anti-FLAG M2 antibodies. After the incubation, the bead-bound protein complexes were washed twice with 190 µl of Reaction buffer containing 100 mM KOAc and then resuspended in Reaction buffer containing 100 mM KOAc and 5 mM ATP. For the disassembly reactions, 50 nM each of human p97 (or yeast Cdc48) hexamers, human UFD1-NPL4 (or yeast Ufd1-Npl4), and the indicated human UBX proteins were added to the suspension of beads bearing ubiquitylated yeast CMG, before incubation at 30°C for 20 min with shaking at 1000 rpm on an Eppendorf ThermoMixer. The supernatants were collected and the beads washed twice with 190 µl of Reaction buffer containing 700 mM KOAc, before elution of bound proteins by the addition of LDS-PAGE sample loading buffer (Invitrogen, NP0007; LDS = lithium dodecyl sulphate) and heating for 3 min at 95°C.

Reconstituted ubiquitylation and disassembly of C. elegans CMG helicase

Request a detailed protocol

For the experiment in Figure 5—figure supplement 2B, worm CMG was ubiquitylated in reconstituted in vitro reactions as described previously (Xia et al., 2021), using proteins described in Supplementary file 1. Reactions were set up using ‘Reaction buffer’ containing 100 mM KOAc and 5 mM ATP, in a final volume of 10 µl and a final concentration of 15 nM worm CMG, 50 nM UBA-1, 300 nM LET-70, 300 nM UBC-3, 50 nM ULA-1_RFL-1, 300 nM UBC-12, 100 nM DCN-1, 500 nM NED-8, 15 nM CUL-2LRR-1, 60 nM TIM-1_TIPIN-1, 30 nM POLε, 30 nM CLSP-1, 30 nM MCM-10, 60 nM CTF-18_RFC, and 20 nM CTF-4. Ubiquitylation reactions were conducted at 20°C for 20 min.

The ubiquitylated worm CMG was then incubated for 30 min at 4°C with 2 µl (per 10 µl ubiquitylation reaction mix) magnetic beads (Dynabeads M-270 Epoxy; Life Technologies, 14302D) that had been coupled to anti-worm SLD-5 antibodies. After the incubation, the bead-bound protein complexes were washed twice with 190 µl of Reaction buffer containing 100 mM KOAc and then resuspended in Reaction buffer containing 5 mM ATP. For the disassembly reactions, 200 nM of C. elegans CDC-48.1 hexamer, 50 nM of C. elegans UFD-1_NPL-4.1 and 50 nM of the indicated form of C. elegans UBXN-3 were added to the suspension of beads bearing ubiquitylated worm CMG, in a final volume of 10 µl. The reactions were then incubated at 20°C for 20 min, whilst shaking at 1000 rpm on an Eppendorf ThermoMixer. The supernatants were then collected and the beads washed twice with 190 µl of Reaction buffer containing 700 mM KOAc. The bound proteins were eluted by addition of LDS-PAGE sample loading buffer (Invitrogen, NP0007) and heating for 3 min at 95°C.

Monitoring the interaction of UFD1-NPL4 with p97

Request a detailed protocol

For the experiment in Figure 3—figure supplement 2, 3 µg of 14His-Smt3UFD1-NPL4 and 13 µg of p97 were mixed as indicated with 2 µl slurry of Ni-NTA beads in a total of 15 µl using ‘Gel filtration buffer’ containing 30 mM imidazole and 2 mM ATP. After 15-min incubation on ice, the Ni-NTA beads were washed four times with 1 ml of ‘Wash buffer’ containing 20 mM imidazole and 0.5 mM ATP. Bead-bound proteins were eluted in LDS-PAGE sample loading buffer containing 50 mM EDTA, by heating for 3 min at 95°C.

Monitoring the interaction of UBX proteins with p97-UFD1-NPL4

Request a detailed protocol

For the experiments in Figure 4—figure supplement 1 and Figure 5—figure supplement 1, 3 µg of 6HisNPL4-UFD1, 3 µg of each UBX protein, and 13 µg of p97 were mixed as indicated with 2 µl slurry of Ni-NTA beads in a total of 11 µl, using ‘Gel filtration buffer’ containing 30 mM imidazole and 2 mM ATP. After 15-min incubation on ice, the Ni-NTA beads were washed three times with 200 µl of ‘Wash buffer’ containing 20 mM imidazole and 0.5 mM ATP. Bead-bound proteins were eluted in LDS-PAGE sample loading buffer containing 50 mM EDTA, by heating for 3 min at 95°C.

Monitoring the interaction of UFD1-NPL4 or FAF1 with K48-linked ubiquitin chains

Request a detailed protocol

For the experiments in Figure 3—figure supplement 2E and Figure 5—figure supplement 3, 15 µg of 14His-Smt3UFD1-NPL4 or 15 µg of 14His-Smt3FAF1, plus 2 µg of labelled K48-linked ubiquitin chains were mixed with 5 µl slurry of Ni-NTA beads, in a total of 15 µl using ‘Gel filtration buffer’ containing 30 mM imidazole. After 15-min incubation on ice, the Ni-NTA beads were washed four times with 1 ml of ‘Wash buffer’ containing 20 mM imidazole. Bead-bound proteins were eluted in LDS-PAGE sample loading buffer containing 50 mM EDTA, by heating for 3 min at 95°C. After SDS–PAGE, gels were stained with InstantBlue (Expedeon, ISB1L) and scanned with an Odyssey CLx imaging System (Li-Cor) to detect the labelled ubiquitin chains.

SDS–PAGE and immunoblotting

Request a detailed protocol

Protein samples were resolved by SDS–PAGE in NuPAGE Novex 4–12% Bis-Tris gels (Thermo Fisher Scientific, NP0321 and WG1402A) either with NuPAGE MOPS SDS buffer (Thermo Fisher Scientific, NP0001) or NuPAGE MES SDS buffer (Thermo Fisher Scientific, NP0002), or with NuPAGE Novex 3–8% Tris–Acetate gels (Thermo Fisher Scientific, EA0375BOX and WG1602BOX) using NuPAGE Tris-Acetate SDS buffer (Thermo Fisher Scientific, LA0041). Resolved proteins were either stained with InstantBlue (Expedeon, ISB1L) or were transferred to a nitrocellulose membrane with the iBlot2 Dry Transfer System (Invitrogen, IB21001S). Antibodies used for protein detection in this study are described in Supplementary file 1. Conjugates to horseradish peroxidase of anti-sheep IgG from donkey (Sigma-Aldrich, A3415), or anti-mouse IgG from goat (Sigma-Aldrich, A4416) were used as secondary antibodies before the detection of chemiluminescent signals on Hyperfilm ECL (cytiva, 28906837, 28906839) with ECL Western Blotting Detection Reagent (cytiva, RPN2124).

Preparation of antibody-coated magnetic beads

Request a detailed protocol

Firstly, 300 mg of Dynabeads M-270 Epoxy (Thermo Fisher Scientific, 14302D) were resuspended in 10 ml of dimethylformamide. Subsequently, 425 µl slurry of activated magnetic beads, which corresponded to ~1.4 × 109 beads, were washed twice with 1 ml of 1 M NaPO3 (pH 7.4). The beads were then incubated with 300 µl of 3 M (NH4)2SO4, 300 µg of anti-FLAG M2 antibody (Sigma-Aldrich, F3165), and 1 M NaPO3 (pH 7.4) up to a total volume of 900 µl. The mixture was then incubated at 4°C for 2 days with rotation. Finally, the beads were treated as follows: four washes with 1 ml PBS; 10 min in 1 ml PBS containing 0.5 % (wt/vol) IGEPAL CA-630 with rotation at room temperature; 5 min in 1 ml PBS containing 5 mg/ml BSA with rotation at room temperature; one wash with 1 ml PBS containing 5 mg/ml BSA. The beads were then resuspended with 900 µl PBS containing 5 mg/ml BSA and stored at 4°C.

Cell lines

Request a detailed protocol

This study utilised E14tg2A mouse cells that were described previously (Villa et al., 2021) and validated by DNA sequencing of genomic loci.

Growth of mouse ES cells

Request a detailed protocol

Mouse ES cells were grown as described previously (Villa et al., 2021). E14tg2a cells (Supplementary file 1) were cultured in a humidified atmosphere of 5% CO2, 95% air at 37°C under feeder-free conditions with Leukemia Inhibitory Factor (LIF; MRC PPU Reagents and Services DU1715) in serum-containing medium. All culturing dishes were precoated with PBS containing 0.1% (wt/wt) gelatin (Sigma-Aldrich, G1890) for at least 5 min prior to seeding. The medium was based on Dulbecco’s modified Eagle medium (DMEM; Thermo Fisher Scientific, 11960044), supplemented with 10% fetal bovine serum (FCS-SA/500, Labtech), 5% knockout serum replacement (Thermo Fisher Scientific, 10828028), 2 mM L-glutamine (Thermo Fisher Scientific, 25030081), 100 U/ml penicillin–streptomycin (Thermo Fisher Scientific, 15140122), 1 mM sodium pyruvate (Thermo Fisher Scientific, 11360070), a mixture of seven non-essential amino acids (Thermo Fisher Scientific, 11140050), 0.05 mM β-mercaptoethanol (Sigma-Aldrich, M6250) and 0.1 µg/ml LIF. For passaging, cells were released from dishes using 0.05% trypsin–EDTA (Thermo Fisher Scientific, 25300054).

To determine doubling times, 200,000 or 500,000 cells (initial cell number N1) were grown for 48 hr on a 6-well plate, before recovery with 0.05% trypsin/EDTA. The total cell number in the final suspension (final cell number N2) was determined using a CellDrop BF (DeNovix). The doubling time (G) was then calculated via the formula: G = (48 × log(2))/(log(N2) − log(N1)). The experiments were performed in triplicate and mean values were determined, together with the standard deviation (SD).

CRISPR-Cas9 genome editing

Request a detailed protocol

To design a pair of guide RNAs (gRNAs) to target a specific site in the mouse genome, we used ‘The CRISPR Finder’ provided by the Welcome Sanger Institute (https://wge.stemcell.sanger.ac.uk//find_crisprs). Annealed oligonucleotides containing the homology region were phosphorylated with T4 polynucleotide kinase (New England Biolabs, M201) and then ligated into the Bbs1 site of the vectors pX335 and pKN7 (Supplementary file 1) in the presence of T4 DNA ligase (New England Biolabs, M202), as previously described (Pyzocha et al., 2014).

To create small deletions at a particular locus, mouse ES cells were transfected with two plasmids expressing the chosen pair of gRNAs together with the Cas9-D10A ‘nickase’ mutant and the Puromycin resistance gene, as described below. FAF1Δ UBXN7Δ cells were produced by targeting the Ubxn7 locus in FAF1∆ cells. Similarly, FAF1Δ UBXN7Δ FAF2-∆UBX cells were produced by targeting the Faf2 locus in FAF1Δ UBXN7Δ cells.

To express human FAF1 or UBXN7 from the CAG promoter in FAF1Δ UBXN7Δ mouse ES cells, we made donor DNA constructs in which the CAG promoter-FAF1 or the CAG promoter-UBXN7 were flanked with 0.5–1 kb of homology to either side of the Rosa26 target locus. FAF1Δ UBXN7Δ cells were then transfected with 1.5 µg donor plasmid DNA together with 1 µg each of the plasmid pKN88 that expresses Cas9 nickase, a gRNA targeting Rosa26 locus and the Puromycin resistance marker.

Transfection of mouse ES cells and selection of clones

Request a detailed protocol

A stock solution of 1 mg/ml linear polyethylenimine (PEI; Polysciences, Inc, 24765-2) was prepared in a buffer containing 25 mM HEPES, 140 mM NaCl, 1.5 mM Na2HPO4 adjusted to pH 7.0 with NaOH. The solution was sterilised by passing through a 0.2-µm filter and stored at −20°C. 1 µg of each plasmid DNA (two plasmids expressing Cas9, gRNAs and the Puromycin resistance gene) were mixed in 100 µl of reduced-serum medium OPTI-MEM (Thermo Fisher Scientific, 31985062) then mixed with 15 µl of 1 mg/ml PEI, vortex well and left for 15 min. Subsequently, 1.0 × 106 cells were aliquoted and centrifuged at 350 × g for 5 min before resuspension in 200 µl OPTI-MEM. The cell suspension and DNA-PEI mix were gently mixed and incubated at room temperature for 30 min in a 1.5 ml microfuge tube. Cells were then transferred to a single well of a 6-well plate that contained 2 ml of complete DMEM medium. Cells were incubated for 24 hr after transfection, followed by two 24-hr rounds of selection with fresh medium containing 2 µg/ml Puromycin (Thermo Fisher Scientific, A1113802). The surviving cells were then released from the wells with 0.05% trypsin/EDTA, diluted 5/25/125 times with complete DMEM medium and then plated on 10 cm plates precoated with PBS containing 0.1% (wt/wt) gelatin. Subsequently, colonies were expanded and monitored as appropriate by immunoblotting, PCR, and DNA sequencing of the target locus.

Genotyping of mouse ES cells by PCR and DNA sequencing

Request a detailed protocol

Cells from a single well of a 6-well plate were resuspended in 100 µl of 50 mM NaOH and the sample was then heated at 95°C for 15 min. Subsequently, 11 µl of 1 M Tris–HCl (pH 6.8) was added to neutralise the pH. A 0.5 µl aliquot of the resulting genomic DNA solution was used as the template for genotyping purposes in 25 µl PCR reactions with Ex Taq DNA polymerase (TaKaRa, RR001). PCR products were subcloned with the TOPO TA cloning Kit (Invitrogen, K457540) and sequenced with T3 primer (5′-AATTAACCCTCACTAAAGGG-3′).

Extracts of mouse ES cells and immunoprecipitation of protein complexes

Request a detailed protocol

1.2 × 107 mouse ES cells were plated on a 15-cm Petri dish and grown for 24 hr at 37°C. Cells were released from the dishes by incubation for 10 min with 10 ml of PBS containing 1 mM EGTA, 1 mM EDTA and then harvested by centrifugation at 350 × g for 3 min, before snap-freezing with liquid nitrogen and storage at −80°C. Cell pellets from two 15-cm Petri dishes were resuspended with an equal volume of ‘Cell Lysis buffer’ supplemented with 5 μM Propargyl-Ubiquitin (MRC PPU Reagents and Services, DU49003) to inhibit deubiquitylase activity, and chromosomal DNA was then digested for 30 min at 4°C with 1600 U/ml of Pierce Universal Nuclease (Thermo Fisher Scientific, 88702). The extracts were centrifuged at 20,000 × g for 30 min at 4°C. The supernatant was then mixed with 100 μl slurry of Protein A/G Sepharose beads (Expedeon, AGA1000) that had been pre-washed in cell Lysis buffer. This step was used to remove proteins that bound non-specifically to the beads. After 15 min of rotation at 4°C, the samples were centrifuged at 800 × g for 30 s at 4°C and the supernatant was recovered. A 20 μl aliquot of the supernatant was added to 40 μl of 1.5 × LDS sample loading buffer.

The rest of the extract was then incubated for 90 min with 10 μl slurry of GFP-Trap Agarose beads (gta-100, Chromotek). The beads were washed four times with 1 ml of Wash buffer and bound proteins were eluted at 95°C for 3 min in 30 μl of 1× LDS sample loading buffer.

Inactivation of CUL2LRR1 in mouse ES cells

Request a detailed protocol

Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific, 13778075) was used to introduce siRNA into mouse ES cells, according to the manufacturer’s protocol. LRR1 siRNA (Horizon Discovery, J-057816-10) were transfected at a final concentration of 25 nM. After 24 hr, the transfection was repeated and incubation continued for a further 24 hr. Subsequently, cells were treated with 5 µM MLN-4924 (Activebiochem, A1139) for 5 hr before imaging. MLN-4924 inhibits the E1 enzyme of the protein neddylation pathway and thereby inhibits cullin ligase activity (Soucy et al., 2009).

Analysis of mitotic CMG helicase disassembly by spinning disk confocal microscopy

Request a detailed protocol

Confocal images of live cells were acquired with a Zeiss Cell Observer SD microscope with a Yokogawa CSU-X1 spinning disk, using a HAMAMATSU C13440 camera with a PECON incubator, a 60 × 1.4 NA Plan-Apochromat oil immersion objective, and appropriate excitation and emission filter sets. Images were acquired using the ‘ZEN blue’ software (Zeiss) and processed with ImageJ software (National Institutes of Health) as previously described (Sonneville et al., 2017).

For time-lapse live cell imaging, mouse ES cells expressing GFP-SLD5 from the endogenous locus were grown on ‘µ-Dish 35 mm, high’ (Ibidi, 81156) with ‘no phenol red DMEM medium’ (Thermo Fisher Scientific, 21063029) supplemented as described above. Cells were treated as above with LRR1 siRNA for a total of 48 hr, followed by MLN-4924 for 5 hr, to inhibit CUL2LRR1 so that CMG would persist on chromatin until mitosis. Datasets were then acquired every 6 min for 14 hr (2 × 2 binning, 26% power of 0.26-s exposure at 488 nm laser to excite GFP) at approximately 30 different multiple stage positions, with heating to 37°C and 5% CO2.

Subsequently, cells entering mitosis were identified by manual examination of time-lapse movie data. The time of nuclear envelope breakdown was indicated by dispersion of the nuclear GFP-SLD5 signal into the cytoplasmic area as shown in Figure 6C and observed previously (Villa et al., 2021). The persistence of GFP-SLD5 foci after nuclear envelope breakdown was monitored in each time-lapse series, until the time point at which the foci disappeared (signal no longer visible above the background fluorescence of the mitotic cytoplasm).

Only one cell line could be recorded per day, with each recording session typically yielding 50–150 cells that entered mitosis during the experiment. remained in focus throughout and could therefore be quantified subsequently. For each of the genotypes shown in Figure 6D and Figure 6—figure supplement 4E, three independent experiments were performed. Mean values were then calculated for each genotype, together with the average of the mean values and the associated standard deviation, as indicated in Figure 6D and Figure 6—figure supplement 4E. The samples were then compared via a Kruskal–Wallis test followed by Dunn’s test, in order to assess statistical significance, using Prism 9 software (GraphPad, https://www.graphpad.com/scientific-software/prism/).

Cell survival assay for mouse ES cells

Request a detailed protocol

Five hundred cells (except 1000 cells for FAF1∆ UBXN7∆ in Figure 7A, B) were plated into 10-cm Petri dishes. On the next day, 0, 3, or 6 µl of 10 mM MLN-4924 or DMSO (dimethyl sulfoxide) were added to 10 ml medium to give a final concentration of 0, 3, or 6 µM, before incubation for 6 hr. The drug-containing medium was then replaced with fresh medium, and incubation proceeded for a further 6 days. Cells were then washed with PBS and fixed with methanol. The fixed cells were stained with 0.5% crystal violet solution (Sigma-Aldrich, HT90132), and images of the plates were captured with a scanner. The number of colonies was counted manually, and the survival rate was then calculated.

Statistics and reproducibility

Request a detailed protocol

No statistical method was employed to predetermine sample size and investigators were not blinded to allocation during the performance of experiments and assessment of results.

Figure 6D, Figure 6—figure supplement 3F, Figure 6—figure supplement 4E, and Figure 7B–D were repeated three times. All other experiments were repeated at least twice, except for Figure 3—figure supplement 2C, Figure 4—figure supplement 1, Figure 5—figure supplement 1, Figure 6—figure supplement 3D, E, and Figure 6—figure supplement 4C, which were performed once.

Graphs were generated and statistical tests performed using Prism 9 software (GraphPad, https://www.graphpad.com/scientific-software/prism/), as described above and in the figure legends. A Kruskal–Wallis test followed by Dunn’s test was used to analyse the data in Figure 6D and Figure 6—figure supplement 4E, which represent multiple non-parametric samples. A two-tailed t-test was used in Figure 7, to assess statistical significance between two groups of data.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files; Source Data files have been provided .

References

Decision letter

  1. Wade Harper
    Reviewing Editor; Harvard Medical School, United States
  2. David Ron
    Senior Editor; University of Cambridge, United Kingdom

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Multiple UBX proteins reduce the ubiquitin threshold of the mammalian p97-UFD1-NPL4 unfoldase" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by David Ron as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Summary:

The p97/Cdc48 AAA ATPase and its heterodimeric ubiquitin adapter Ufd1-Npl4 unfolds ubiquitylated proteins, often to segregate large protein complexes such as the MCM helicase at termination of replication. The authors now demonstrate that an important difference exists between the yeast and metazoan system. While Cdc48-Ufd1-Npl4 can target MCM7 with a relatively short ubiquitin chain (5-7 units), the metazoan p97 requires either much longer chains or one of a group of accessory proteins that have previously been connected to various p97 and Ufd1-Npl4 mediated processes.

Essential revision:

The reviewers agreed that the paper has several strengths, including the very nice in vitro system for analysis of the role of p97 in segregase activity on a multi-subunit complex. Some aspects of the biochemistry are strong, and the system itself – CMG disassembly – is embedded in rich biology, which is also a plus. However, the reviewer's feel that the paper falls short of fully demonstrating the proposed mechanism, and several weaknesses were noted (as provided in the detailed reviewer comments). In particular, given where the p97 field is currently, further information is needed to fully understand the actual biochemical mechanism of UBX-dependent enhancement of unfolding. Thus, we would ask for major revisions.

After discussion among the reviewers, the consensus is that for further consideration, the following points would need to be addressed. We realize that this represents a substantial revision to the paper, with significant experimental aspects needing to be addressed.

1. The description of the mechanism by which the UBX domain (together with the coiled-coil) can promote unfolding needs further elaboration. Further experiments could employ crosslink sensors of protein unfolding coupled with binding assays to understand where/how in the unfolding process the UBX effect is manifested. Without this, the reader is left wondering what is actually happening biochemically. Given the advanced stage of our understanding of p97 Ub recognition and unfolding, some additional mechanistic understanding is relevant in this case.

2. The reviewer's also commented on the treatment of the Ub chains in your analysis. Given that the primary conclusion is that Ub chain length is relevant, a more definitive description of the chains on the different forms of the CMG being examined. There are various aspects here, ranging from length to chain topology. Given that you have used Cdc34 to build chains on CMG, this is almost certainly exclusively K48-linked chains, so perhaps the topology aspect isn't as import as other aspects. Related to this is the question of whether the differential effects seen on unfolding with apparent chain length reflects chain length, weak binding, or a differential intrinsic activity of yeast versus human p97. One suggestion is to perform kinetic experiments in order to address this critical question.

3. There are significant concerns about the viability and survival assays in Figure 6. For the cell-based SLD5 mobilization assay, you would need to: (1) discuss how they quantified (preferably automated), (2) depict the data in a box blot for that type of measurement, and (3) show significance with an appropriate test. More importantly, the data as it stands only shows the requirement of the accessory protein, which has been shown in several other papers previously. Ultimately, to bring the story full circle, you would need to demonstrate that the structural components of the UBX protein (UBX domain and coiled-coil) do (or do not) rescue the various cellular phenotypes. Given that the actual function of the coiled coil and the UBX haven't been mapped to specific residues, then it seems that some understanding of these elements would be needed.

Reviewer #1 (Recommendations for the authors):

This paper describes a systematic biochemical analysis of UBX proteins in facilitating protein unfolding by the p97-UFD1-NPL4 (referred to here is the p97 complex). The p97 complex binds Ub and unfolds it to allow the ubiquitylated protein to be translocated into the p97 ATPase pore for unfolding. This paper demonstrates that UBX proteins are able to reduce the necessary ubiquitin chain length in order to support unfolding by p97. They explore this using ubiquitylated CMG helicase as a substrate. Removal of CMG helicase from replicated DNA is required for completion of DNA synthesis.

First the authors demonstrate that the p97 complex only only unfolds CMG with very long Ub chains. The then show that the high threshold for Ub is reduced when UBXN7, FAF1 or FAF2 are added. These proteins bind to both the p97 complex and Ub in substrates. This is then followed up in cells by demonstrating that removal of UBXN7 and FAF1 reduces CMG disassembly and is synthetic with reduced CMG ubiquitin ligase activity.

1. I found the section describing Figure 2 to be very unclear (especially lines 152-155). It sort of seems like 2 things are being changed at once and interpreted in a way that isn't very clearly described. First, the interpretation of Low versus High conditions and extent of ubiquitylation needs to be justified better. It appears to be based on prior studies but has the site(s) of ubiquitylation in the "Low" sample been demonstrated independently for the samples used here? This seems critical to validate, since if for some reason, the Low sample has substantial multi-mono versus mostly ubiquitylation on a single site (K29), the interpretation would be different.

The description of the results is also sufficiently terse as to not be clear as to what is really being measured and concluded. It is conceivable that the processivity of the CRL in this case (i.e. the actual chain length) varies with the primary ubiquitylation site. Thus, it may not be easy to compare the apparent efficiency of different sites as the length of chains on each site may be different. It would be nice if there was a way to chain amputate the UB chains and measure the average length of chains for the 2 populations of substrates. Proteasomes from yeast lacking Ubp6 should release Ub chains from the Mcm7 ubiquitylated species as "amputated" chains via Rpn11 activity, and so in the in vitro setting, it might be possible to determine the constellation of chain length using anti-Ub to blot reactions. An alternative would be to use Ub-AQUA proteomics or possibly the Clippase enzyme to measure the average chain length for each population. If there are several sites of ubiquitylation in WT MCM7, the chains themselves might not be significantly longer than on the mostly K29-modified sample, but the authors seem to want to conclude its chain length that is the relevant thing changing.

In terms of the interpretation of the results – it seems pretty clear that for the K29 only/Low sample, Cdc48 prefers the longer forms but for the WT/High sample, it looks like Cdc48 does not prefer the longest forms, as there is a slight bias towards longer forms for the protein left on the beads (if reproducible).

An additional issue for this analysis concerns whether the intrinsic or specific activity of p97 and Cdc48 for ATP binding, hydrolysis and unfolding are equivalent. This is especially important for interpreting the activity of p97 towards the "Low" sample. Do the authors have a way to measure the specific activity of their p97 and Cdc48 preps? The proteins themselves could be differentially prone to damage or inactivation during purification, thereby creating distinct specific activities for the two preps that could affect the interpretation.

In the end, I am not sure that the conclusion "These findings indicated that human p97-UFD1-NPL4 has a much higher ubiquitin threshold than yeast Cdc48-Ufd1-Npl4." Is that strong based on the data in Figure 2.

Also, in figure 2A and text, I am not sure the words "limited" and "efficient" are the best terms to use. Chain lengths may be comparable in both cases. Maybe something like "site-specific" for the K29-only version and "multi-site" for the other.

Given the inability of the p97 complex to unfold the CMG under "Low" ubiquitylation conditions, it might be worth asking whether increasing the amount of p97 complex in the reaction can overcome the barrier. This could potentially address issues related to reduced specific activity. It could also potentially address reduced Ub affinity for the p97 complex versus the Cdc48 complex.

2. The data in Figure 3 are said to be based on CMG complexes that had up to ~12 ubiquitin conjugated to the Mcm7 subunit. This figure is also subject to the same issue as above – can the authors distinguish between one long chain versus some amount of multi-site ubiquitylation and shorter chains in this case?

3. The deletion analysis in Figure 4 doesn't seem to example the extent to which the various mutants assemble with the p97 complex. One presumes a correlation, but this wasn't tested directly.

One line 216, it is stated: "These findings indicated that the coiled-coil domain of FAF1 has a previously unanticipated role in stimulating the unfoldase activity of p97-UFD1-NPL4." However, it is possible that without this coiled-coil, the UBX domain doesn't associate effectively with the p97 complex. This would be sort of a trivial result, in that there isn't a "direct" role of the copied-coil, but it is simply required for the UBX domain to interact. Can the coiled-coil be replaced by an alternative coiled-coil sequence or is there something special about this coiled-coil?

4. Although a bit hard to know for sure, it looks like the ubiquitylated forms of Mcm7 in the ES cell experiment (Figure 5) have much shorter chains than the experiments with yeast CMG – especially in 5b. Of course it is possible that this reflects the presence of DUB activity during the IP, which would tend to shorten the chains. But at face value, it looks like FAF1/UBXN7 can allow p97 to operate on very short chains (although one really cant distinguish this from multi-mono ubiquitylation).

This raises questions about the overall conclusion. If p97 inhibitor increases the appearance of MCM7 conjugates (albeit with shorter chains) in the FAF1/UBXN7 knockout substantially, this suggest that p97 can disassemble shorter chains in cells without an adaptor OR that there is an additional UBX protein in cells that can do the job. Of course if the ubiquitylation levels are a result of DUB activity, that could explain it but it seems like this should be ruled out.

Reviewer #2 (Recommendations for the authors):

1) The authors nail down the critical activity to the coiled-coil domain of the accessory factors FAF1/FAF2 (in addition to the UBX domain that binds p97). This is surprising and interesting because the UBA domains of FAF1 or FAF2 are not required. Frustratingly, it is not revealed what the function of the coiled-coil region is, which limits the mechanistic understanding. The authors speculate that the coiled coil region binds ubiquitin. This needs to be tested because it is a simple experiment. The comparison to Pup is inappropriate because Pup is not homologous to ubiquitin. Alternatively, the coiled coil region could modulate p97 or Ufd1-Npl4 binding activity. The question would be whether the coiled coil region binds p97 and/or Ufd1-Npl4. Technically, this can be tested.

2) The biochemical experiments show single time points. In contrast, the field has reached a point where kinetic data can be provided. That would allow to quantitatively define even smaller differences such as the deltaUBA or UIM effect in Figure 4C

3) The ES cell microscopy in Figure 5C needs to be quantified by an automated, unbiased pipeline.

4) Figure 6, cell viability and survival: The information gleaned from these experiments is questionable. Effects on colony formation are mostly mediated by MLN4924. The additional effect by the deletions is minor, and needs to be quantified and statistically validated, also to clarify whether it is additive or synthetic. The same applies to the viability assay. More importantly, it is impossible to connect the effects to replication termination. Firstly, there are many processes affected by the manipulations. Secondly, the MLN4924 treatment is so short that it will affect replication termination only in a small percentage of cells.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Multiple UBX proteins reduce the ubiquitin threshold of the mammalian p97-UFD1-NPL4 unfoldase" for further consideration by eLife. Your revised article has been evaluated by David Ron (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

There are just a couple of small items that need attention – noting the p-value for Figure 6D is weakly significant and then a couple of items mentioned by reviewer 2. The paper will then be ready to accept.

Reviewer #1 (Recommendations for the authors):

Overall, the authors have done extensive work to try to address the previous reviewers' comments. The authors have addressed all of the specific comments, and in general, the results support the conclusions. One issue is that in Figure 6D, the p-value is only 0.09, which is not strongly significant. This may have to do with the assay itself. The majority of experiments employ ubiquitin smears to monitor p97 activity as it relates to the length dependence on the biological processes.

Reviewer #2 (Recommendations for the authors):

The authors have addressed concerns regarding data quality and evaluation. Moreover, they added a lot of valuable data on how the Ufd1-Npl4 adapter determines the choice of Ub chain length, including the Npl4 groove mutations and the very neat and informative "species swap". The dissection of effects of Ub chain number, branching, and type is very nice. They also added some more informative biochemical characterization of the critical coiled-coil region of FAF1, and I agree with the authors that more detailed mechanistic dissection goes beyond this analysis. This is a very rich and careful study on a timely and important question in replication termination and p97 function.

Reviewer #3 (Recommendations for the authors):

The authors performed a large number of additional experiments adequately addressing all key points of the Essential Revision section of the decision letter. Moreover, they addressed all specific criticisms raised by this reviewer satisfactorily. The revised manuscript is substantially improved and contains important new data on mechanistic aspects of the observed 'ubiquitin threshold'. Acceptance for publication is strongly recommended.

https://doi.org/10.7554/eLife.76763.sa1

Author response

Essential revision:

The reviewers agreed that the paper has several strengths, including the very nice in vitro system for analysis of the role of p97 in segregase activity on a multi-subunit complex. Some aspects of the biochemistry are strong, and the system itself – CMG disassembly – is embedded in rich biology, which is also a plus. However, the reviewer's feel that the paper falls short of fully demonstrating the proposed mechanism, and several weaknesses were noted (as provided in the detailed reviewer comments). In particular, given where the p97 field is currently, further information is needed to fully understand the actual biochemical mechanism of UBX-dependent enhancement of unfolding. Thus, we would ask for major revisions.

After discussion among the reviewers, the consensus is that for further consideration, the following points would need to be addressed. We realize that this represents a substantial revision to the paper, with significant experimental aspects needing to be addressed.

1. The description of the mechanism by which the UBX domain (together with the coiled-coil) can promote unfolding needs further elaboration. Further experiments could employ crosslink sensors of protein unfolding coupled with binding assays to understand where/how in the unfolding process the UBX effect is manifested. Without this, the reader is left wondering what is actually happening biochemically. Given the advanced stage of our understanding of p97 Ub recognition and unfolding, some additional mechanistic understanding is relevant in this case.

To elaborate further the action of the UBX proteins, we first needed to clarify the mechanistic basis for the difference in ubiquitin threshold between human p97-UFD1-NPL4 and yeast Cdc48-Ufd1-Npl4. Taking advantage of the very high sequence conservation between human p97 and yeast Cdc48 (Figure 3—figure supplement 1A of the revised manuscript), we performed reconstituted unfoldase assays that tested the ability of human p97 to work with yeast Ufd1-Npl4, and yeast Cdc48 to function with human UFD1-NPL4. In this way, we tested whether the difference in ubiquitin threshold between yeast Cdc48-Ufd1-Npl4 (five or more ubiquitins) and human p97-UFD1-NPL4 (many ubiquitins) was due to differences between Sc Cdc48 & Hs p97, Sc Ufd1-Npl4 & Hs UFD1-NPL4, or both. Despite the evolutionary distance between humans and yeast, such ‘species swap’ experiments worked well and the new data in Figure 3A show that all combinations of Sc Cdc48 / Hs p97 with Sc Ufd1-Npl4 / Hs UFD1-NPL4 supported efficient disassembly of an extensively ubiquitylated CMG substrate. Crucially, however, human UFD1-NPL4 did not support the disassembly of CMG with 5-15 ubiquitins conjugated to the CMG-Mcm7 subunit, either in reactions with human p97 or yeast Cdc48 (Figure 3B, lanes 5-8 and Figure 4—figure supplement 3, lanes 3-4 and 7-8). In contrast, yeast Ufd1-Np4 allowed both yeast Cdc48 (Figure 3B lanes 3-4) and human p97 (Figure 3B lanes 9-10) to disassemble CMG with five or more ubiquitins conjugated to CMG-Mcm7. These data demonstrate that the different ubiquitin thresholds of human p97-UFD1-NPL4 and yeast Cdc48-Ufd1-Npl4 are determined by differences between human UFD1-NPL4 and yeast Ufd1-Npl4, rather than reflecting differences in the enzymatic activity of human p97 and yeast Cdc48.

Human UFD1-NPL4 and yeast Ufd1-Npl4 are less well conserved in primary sequence than human p97 and yeast Cdc48 (Figure 3—figure supplement 1A-C). Nevertheless, previous structural studies showed that human UFD1-NPL4 and yeast Ufd1-Npl4 interact in an analogous manner with human p97 and yeast Cdc48 (Bodnar et al., 2018; Pan et al., 2021). Moreover, both yeast Ufd1-Npl4 and human UFD1-NPL4 share conserved interaction sites with K48-linked polyubiquitin, via the top of the Sc Npl4 / Hs NPL4 ‘tower’ that sits on top of the N-terminal ring of p97 adjacent to the central pore, and the UT3-domain of Sc Ufd1 / Hs UFD1 (discussed on lines 220-227 of the revised manuscript). Elegant studies of yeast Cdc48-Ufd1-Npl4 from Tom Rapoport’s group showed that binding of yeast Ufd1-Npl4 to K48-linked polyubiquitin leads to capture of an unfolded ubiquitin intermediate via a groove on the outside of the Npl4 tower, which thereby directs unfolded ubiquitin into the Cdc48 central pore (Twomey et al., 2019). Correspondingly, mutations of conserved residues in the groove of yeast Npl4 block substrate unfolding (Twomey et al., 2019). We now show in the revised manuscript that the equivalent mutations in human NPL4 (Figure 3—figure supplement 2A-B, NPL4-AAE) block the disassembly of ubiquitylated CMG (Figure 3—figure supplement 2C lanes 7-8), without affecting the binding of NPL4 to UFD1 (Figure 3—figure supplement 2B) or p97 (Figure 3—figure supplement 2D). Together with the previous studies mentioned above and discussed in our manuscript, this indicates that human p97-UFD1-NPL4 also initiates substrate unfolding via an unfolded ubiquitin intermediate, just like yeast Cdc48-Ufd1-Npl4.

However, previous studies also point to functional diversification between human UFD1-NPL4 and yeast Ufd1-Npl4. In particular, human NPL4 has a ubiquitin-binding Zn finger at its C-terminus (NPL4-NZF) that is not found in yeast and represents the highest affinity binding site for ubiquitin in human UFD1-NPL4. Nevertheless, the NPL-NZF was previously shown to be dispensable in human cells for the ER-associated degradation of misfolded and ubiquitylated proteins (Ye et al., 2003). This suggested that the evolutionarily conserved low-affinity binding sites for K48-linked ubiquitin chains in UFD1-NPL4, at the top of the NPL4 tower and in the UT3 domain of UFD1 (Ji et al., 2021; Pan et al., 2021) (Park et al., 2005; Sato et al., 2019; Twomey et al., 2019; Ye et al., 2003), are sufficient to initiate the unfolding of ubiquitylated substrates of p97. Consistent with past work (Meyer et al., 2002), we show in the revised manuscript that deletion of the NPL4-NZF did not impair association with p97 or UFD1 (Figure 3—figure supplement 2B and D) but abrogated the detectable interaction of UFD1-NPL4 with K48-linked ubiquitin chains in vitro (Figure 3—figure supplement 2E). Importantly, we show that human p97-UFD1-NPL4-∆NZF still supports the disassembly of heavily ubiquitylated CMG helicase in reconstituted in vitro assays, albeit with reduced efficiency compared to wild type NPL4 (Figure 3—figure supplement 2C, compare lanes 1-6). Considered together with the data and previous observations discussed above, these findings indicate that the dynamic association of UFD1-NPL4 with very long K48-linked ubiquitin chains, likely via the conserved low affinity ubiquitin-binding sites in the NPL4 tower and UFD1-UT3 domain, is sufficient to initiate substrate unfolding, dependent upon the NPL4 groove. Additional ubiquitin binding by the NPL4 NZF domain is not an essential part of the mechanism of substrate unfolding, but likely increases the efficiency of substrate engagement by human p97-UFD1-NPL4, as does increased length of the K48-linked ubiquitin chains (see below for a detailed characterisation of ubiquitin chains in the revised manuscript).

The above insights into the mechanism of human p97-UFD1-NPL4, and the role of UFD1-NPL4 in setting the ubiquitin threshold, provide a framework for understanding how UBX proteins stimulate the ability of human p97-UFD1-NPL4 to process substrates with shorter K48-linked ubiquitin chains. CMG disassembly by human p97-UFD1-NPL4 in the presence of UBX proteins required five or more ubiquitins on the CMG-Mcm7 subunit (Figure 4E lanes 7-8), thereby matching the minimal ubiquitin threshold of yeast Cdc48-Ufd1-Npl4 (Figure 4E lanes 3-4) that is thought to reflect the presence of multiple conserved ubiquitin-binding domains within yeast Ufd1-Npl4 (Deegan et al., 2020; Park et al., 2005; Sato et al., 2019; Twomey et al., 2019). This suggested that the UBX proteins function by stimulating productive interactions between the multiple ubiquitin-binding modules of human p97-UFD1-NPL4 and ubiquitylated substrate proteins. The initial engagement of UFD1-NPL4 with ubiquitylated substrates in the presence of UBX proteins should then lead to the trapping of an unfolded ubiquitin intermediate on the NPL4 groove. Consistent with this view, we now show in the revised manuscript that CMG disassembly in the presentence of UBXN7, FAF1 and FAF2 is dependent upon the NPL4 groove (Figure 4—figure supplement 2B; compare lanes 5-6 and 13-14).

We also show in the revised manuscript that UBX proteins suppressed the partial defect in disassembling heavily ubiquitylated CMG that was observed in the absence of the ubiquitin-binding Zinc finger at the carboxyl terminus of NPL4 (Figure 4—figure supplement 2, compare release of Mcm6 and Sld5 into supernatant in lanes 5-10). Moreover, UBX proteins restored the ability of yeast Cdc48 to disassemble CMG complexes in the presence of human UFD1-NPL4, when the K48-linked ubiquitin chains on CMG-Mcm7 were otherwise too short to allow disassembly (Figure 4—figure supplement 3, lanes 1-6). These findings indicated that the UBX proteins augment the ability of human UFD1-NPL4 to initiate the unfolding of ubiquitylated substrates by p97.

The UBXN7 protein is predominantly nuclear in mammalian cells and our manuscript showed that UBXN7 stimulates the disassembly of ubiquitylated CMG by p97-UFD1-NPL4, both in reconstituted assays with purified proteins (Figure 4D) and in mouse ES cells (Figure 6; Figure 6—figure supplement 4). The reconstituted assays indicate that the ability of UBXN7 to stimulate p97-UFD1-NPL4 requires both the UBX domain that binds p97, and the UBA domain that binds ubiquitin (Figure 5A-C; Figure 5—figure supplement 1A-B show that the UBXN7-∆UBA still interacts with p97-UFD1-NPL4). Together with the data discussed above, these findings indicate that UBXN7 functions by binding to p97-UFD1-NPL4 and stabilising productive interactions between UFD1-NPL4 and K48-linked ubiquitin chains, which then leads firstly to trapping of an unfolded ubiquitin intermediate on the NPL4 groove and then to translocation of the substrate polypeptide through the p97 central channel.

The situation with FAF1 (and FAF2) is more complex, since stimulation of p97-UFD1-NPL4 is dependent upon a coiled coil domain adjacent to the UBX domain, in addition to the UBX domain itself. We show in the revised manuscript that the FAF1 coiled coil is not required for association of the FAF1-UBX with p97-UFD1-NPL4 (Figure 5—figure supplement 1E-F). Nor does the coiled coil of FAF1 bind detectably to UFD1-NPL4 in the absence of p97 (Figure 5—figure supplement 1E-F). Furthermore, human p97 associates with UFD1-NPL4 in the absence of UBX proteins (Figure 3—figure supplement 2D), as shown previously (Meyer et al., EMBO J, 2002). We could not detect stable binding of the FAF1 coiled-coiled to K48-linked ubiquitin chains in vitro (Figure 5—figure supplement 3, FAF1-∆268 and FAF1-∆480). However, the same is true for UFD1-NPL4-∆NZF (Figure 3—figure supplement 2), which nevertheless contains conserved low affinity ubiquitin-binding modules, which we show are sufficient to support the disassembly of ubiquitylated CMG in reconstituted assays (Figure 3—figure supplement 2C). Therefore, we cannot exclude that the FAF1 coiled coil also functions by binding to both p97-UFD1-NPL4 and ubiquitin, with ubiquitin binding being low affinity yet functionally relevant. In our opinion, further mechanistic insight into the action of the coiled coil domain will require structural studies of early intermediates in the unfolding reaction, which are considerably beyond the scope of the present study.

In summary the manuscript provides the following new mechanistic insights for mammalian p97-UFD1-NPL4:

– Substrate unfolding by human p97-UFD1-NPL4 is dependent on the NPL4 groove and thus is likely to initiate via an unfolded ubiquitin intermediate, as for yeast Cdc48-Ufd1-Npl4.

– Human UFD1-NPL4 sets a high ubiquitin threshold for substrate processing by human p97, likely reflecting dynamic association of the conserved ubiquitin-binding modules in the NPL4 tower and UFD1 UT3-domain with K48-linked ubiquitin chains.

– A set of human UBX proteins reduce the ubiquitin threshold of human p97-UFD1-NPL4 to the same level as yeast Cdc48-Ufd1-Npl4. The same UBX proteins also stimulate yeast Cdc48 in combination with human UFD1-NPL4. Stimulation by the UBX proteins is dependent on the NPL4 groove that in yeast is known to bind unfolded ubiquitin during initiation of unfolding. Overall, these findings indicate that the human UBX proteins stabilise productive interactions between the multiple conserved ubiquitin binding sites in Hs UFD1-NPL4 and K48-linked polyubiquitin, thereby promoting the initiation of substrate unfolding via an unfolded ubiquitin intermediate.

– Consistent with the above model, stimulation of p97-UFD1-NPL4 by UBXN7 requires both the p97-binding UBX domain and the ubiquitin-binding UBA domain. These findings provide proof of principle that a UBX protein can stimulate p97-UFD1-NPL4, by binding to the unfoldase and supplying additional ubiquitin binding, which then facilitates the initiation of ubiquitin unfolding by UFD1-NPL4, dependent on the NPL4 groove.

2. The reviewer's also commented on the treatment of the Ub chains in your analysis. Given that the primary conclusion is that Ub chain length is relevant, a more definitive description of the chains on the different forms of the CMG being examined. There are various aspects here, ranging from length to chain topology. Given that you have used Cdc34 to build chains on CMG, this is almost certainly exclusively K48-linked chains, so perhaps the topology aspect isn't as import as other aspects.

The revised manuscript contains new data that demonstrate clearly how CMG-Mcm7 is ubiquitylated under the various conditions employed in our study:

– Figure 1—figure supplement 1A indicates that 2-3 ubiquitin chains are conjugated to the CMG-Mcm7 subunit in the presence of high concentrations of E2 (25 nM) and E3 (10 nM), since 2-3 ubiquitins are conjugated to CMG-Mcm7 in reactions performed with lysine-free ubiquitin (Figure 1—figure supplement 1A, K0 ubiquitin). Under these very efficient conditions, much of the chain-linkage is via K48 of ubiquitin but other linkages are also involved (Figure 1—figure supplement 1A, reactions with K48R ubiquitin).

– Figure 1—figure supplement 1B indicates that a single ubiquitin chain of up to ~15 ubiquitins is conjugated to the CMG-Mcm7 subunit in the presence of low concentrations of E2 (2.5 nM) and E3 (1 nM), since a single ubiquitin is conjugated to CMG-Mcm7 in reactions performed with lysine-free ubiquitin (Figure 1—figure supplement 1B, K0 ubiquitin). Under these less efficient conditions, essentially all of the chains are linked via K48 of ubiquitin (Figure 1—figure supplement 1B, compare reactions with K0 ubiquitin and K48R ubiquitin). Doubling the concentration of E3 under such conditions produces some longer chains on CMG-Mcm7 (Figure 1—figure supplement 1D, 2.5 nM E2 and 2 nM E3) and the linkages are still almost entirely via K48 of ubiquitin (Figure 1—figure supplement 1D, compare reactions with K0 ubiquitin and K48R ubiquitin).

Human p97-UFD1-NPL4 disassembles extensively ubiquitylated CMG (25 nM E2 and 10 nM E3). In the revised manuscript we show that the action of p97-UFD1-NPL4 under such conditions is not dependent upon the presence of 2-3 ubiquitin chains on CMG-Mcm7, using a new allele with lysine mutations in Mcm7 that limit ubiquitylation to a single chain on Mcm7 (Figure 1—figure supplement 1C, CMG-Mcm7-KR, compare wt ubiquitin and K0 ubiquitin). The very long chain on CMG-Mcm7 under such conditions includes much K48-linkage of ubiquitin but with other linkages too (Figure 1—figure supplement 1C, compare wt ubiquitin and K48R ubiquitin). Disassembly of CMG-Mcm7-KR with a single long ubiquitin chain is equally efficient as when 2-3 ubiquitin chains are conjugated to wild type CMG-Mcm7 (Figure 2—figure supplement 3A). Moreover, human p97-UFD1-NPL4 and yeast Cdc48-Ufd1-Npl4 show a similar ability to disassemble CMG complexes ubiquitylated with ‘K48-only ubiquitin’ that can only form K48-linked chains, either under conditions with 2-3 chains per CMG-Mcm7 subunit (Figure 2—figure supplement 3B, lanes 1-6, wild type Mcm7) or under conditions with just one long K48-linked chain conjugated to CMG-Mcm7-KR (Figure 2—figure supplement 3B, lanes 7-12, CMG-Mcm7-KR).

Overall, the above data indicate that human p97-UFD1-NPL4 preferentially disassembles CMG complexes that have long K48-linked ubiquitin chains conjugated to CMG-Mcm7, without any requirement for multiple chains or other chain linkages.

Related to this is the question of whether the differential effects seen on unfolding with apparent chain length reflects chain length, weak binding, or a differential intrinsic activity of yeast versus human p97.

Please see the reply to ‘essential point 1’ above, where we discuss our new ‘species swap’ experiments that show that the high ubiquitin threshold of human p97-UFD1-NPL4 is due to the properties of UFD1-NPL4, rather than due to differential intrinsic activities of yeast Cdc48 versus human p97 (Figure 3).

We also show that deletion of the ubiquitin-binding Zn finger of NPL4 has only a mild effect on CMG disassembly (Figure 3—figure supplement 2A-C), though in its absence the interaction of UFD1-NPL4 with K48-linked ubiquitin chains is highly dynamic (Figure 3—figure supplement 2E). These findings indicate that the high ubiquitin threshold of human p97-UFD1-NPL4 reflects the dynamic interaction of the conserved ubiquitin-binding modules of UFD1-NPL4 (within the NPL4 tower and the UFD1 UT3 domain) with K48-linked ubiquitin, with long chains compensating for such dynamic interactions.

One suggestion is to perform kinetic experiments in order to address this critical question.

The revised manuscript presents a new kinetic experiment in Figure 2—figure supplement 1. Under conditions where the majority of extensively ubiquitylated CMG complexes are disassembled within ~10 minutes, very few CMG complexes with up to ~10 ubiquitins are disassembled even by 30 minutes.

Moreover, Figure 2—figure supplement 2 presents a titration of p97-UFD1-NPL4. Whereas the majority of extensively ubiquitylated CMG complexes were disassembled in the presence of 15 nM p97-UFD1-NPL4, very little disassembly of CMG complexes with up to ~12 ubiquitins is observed even in the presence of 150 nM p97-UFD1-NPL4.

3. There are significant concerns about the viability and survival assays in Figure 6. For the cell-based SLD5 mobilization assay, you would need to: (1) discuss how they quantified (preferably automated), (2) depict the data in a box blot for that type of measurement, and (3) show significance with an appropriate test. More importantly, the data as it stands only shows the requirement of the accessory protein, which has been shown in several other papers previously. Ultimately, to bring the story full circle, you would need to demonstrate that the structural components of the UBX protein (UBX domain and coiled-coil) do (or do not) rescue the various cellular phenotypes. Given that the actual function of the coiled coil and the UBX haven't been mapped to specific residues, then it seems that some understanding of these elements would be needed.

Full details of data quantification for microscopy experiments are now provided in Materials and methods (lines 1061-1079). We also include a new section of Materials and methods entitled Statistics and Reproducibility.

In the original version of the manuscript, Figure 5D contained a histogram that summarised the GFP-SLD5 data representing the mitotic pathway for CMG helicase disassembly. In the revised version of the manuscript, we have replaced the histogram with a scatter plot (Figure 6D in the revised version) that presents the raw data together with the mean values from three independent experiments, the average of the three means, and a statistical analysis of the data via a Kruskal-Wallis test followed by Dunn’s test. The associated p values are now shown in Figure 6D.

We also present new data in Figure 6—figure supplement 4, whereby the defects of UBXN7∆ FAF1∆ cells are rescued by expression of full-length UBXN7, full length FAF1, FAF1-∆UBX or FAF1-∆CoiledCoil (FAF1-∆CC). By monitoring the accumulation of ubiquitylated CMG-MCM7 in UBXN7∆ FAF1∆ cells (Figure 6—figure supplement 4D), we show that wild type FAF1 rescues but FAF1-∆UBX and FAF1-∆CoiledCoil do not. Similarly, we monitored the mitotic CMG disassembly defect of UBXN7∆ FAF1∆ cells by spinning disk confocal microscopy (Figure 6—figure supplement 4E) and showed that expression of wild type FAF1 produced a statistically significant rescue (as did expression of wild type UBXN7), but expression of FAF1-∆UBX or FAF1-∆CoiledCoil did not.

Reviewer #1 (Recommendations for the authors):

This paper describes a systematic biochemical analysis of UBX proteins in facilitating protein unfolding by the p97-UFD1-NPL4 (referred to here is the p97 complex). The p97 complex binds Ub and unfolds it to allow the ubiquitylated protein to be translocated into the p97 ATPase pore for unfolding. This paper demonstrates that UBX proteins are able to reduce the necessary ubiquitin chain length in order to support unfolding by p97. They explore this using ubiquitylated CMG helicase as a substrate. Removal of CMG helicase from replicated DNA is required for completion of DNA synthesis.

First the authors demonstrate that the p97 complex only only unfolds CMG with very long Ub chains. The then show that the high threshold for Ub is reduced when UBXN7, FAF1 or FAF2 are added. These proteins bind to both the p97 complex and Ub in substrates. This is then followed up in cells by demonstrating that removal of UBXN7 and FAF1 reduces CMG disassembly and is synthetic with reduced CMG ubiquitin ligase activity.

1. I found the section describing Figure 2 to be very unclear (especially lines 152-155). It sort of seems like 2 things are being changed at once and interpreted in a way that isn't very clearly described. First, the interpretation of Low versus High conditions and extent of ubiquitylation needs to be justified better. It appears to be based on prior studies but has the site(s) of ubiquitylation in the "Low" sample been demonstrated independently for the samples used here? This seems critical to validate, since if for some reason, the Low sample has substantial multi-mono versus mostly ubiquitylation on a single site (K29), the interpretation would be different.

We take the reviewer’s point and in the revised manuscript we characterise carefully the ubiquitin chains on CMG-Mcm7 under the various conditions used (data discussed above in relation to ‘essential revision point 2’). The new data are presented in Figure 1—figure supplement 1, confirming that ‘low ubiquitylation conditions’ involve the conjugation on CMG-Mcm7 of a single K48-linked ubiquitin chain.

The description of the results is also sufficiently terse as to not be clear as to what is really being measured and concluded. It is conceivable that the processivity of the CRL in this case (i.e. the actual chain length) varies with the primary ubiquitylation site. Thus, it may not be easy to compare the apparent efficiency of different sites as the length of chains on each site may be different. It would be nice if there was a way to chain amputate the UB chains and measure the average length of chains for the 2 populations of substrates. Proteasomes from yeast lacking Ubp6 should release Ub chains from the Mcm7 ubiquitylated species as "amputated" chains via Rpn11 activity, and so in the in vitro setting, it might be possible to determine the constellation of chain length using anti-Ub to blot reactions. An alternative would be to use Ub-AQUA proteomics or possibly the Clippase enzyme to measure the average chain length for each population. If there are several sites of ubiquitylation in WT MCM7, the chains themselves might not be significantly longer than on the mostly K29-modified sample, but the authors seem to want to conclude its chain length that is the relevant thing changing.

See above – in the revised version we characterise carefully the ubiquitylation of CMG-Mcm7 under each condition, showing that ‘low ubiquitylation’ conditions (2.5 nM E2 and 1 nM E3) involve a single K48-linked chain on CMG-Mcm7, whilst ‘high ubiquitylation’ conditions (25 nM E2 and 10 nM E3) produce 2-3 chains per CMG-Mcm7, largely K48-linked but with other linkages too (Figure 1—figure supplement 1). Importantly, we show in Figure 2—figure supplement 3 that multiple chains on CMG-Mcm7 are not required for efficient disassembly by p97-UFD1-NPL4 (by mutating Mcm7 lysines so that only one ubiquitin chain is formed, even under highly efficient ubiquitylation conditions). We also show that yeast Cdc48-Ufd1-Npl4 and human p97-UFD1-NPL4 are both able to disassemble CMG that has been ubiquitylated with ‘K48 only ubiquitin’ (with other lysines mutated to arginine), indicating that the ability of human p97-UFD1-NPL4 to disassemble extensively ubiquitylated CMG is not due to a requirement for other chain linkages that are produced under such conditions (K48-linked chains are processed by both human and yeast unfoldases).

In terms of the interpretation of the results – it seems pretty clear that for the K29 only/Low sample, Cdc48 prefers the longer forms but for the WT/High sample, it looks like Cdc48 does not prefer the longest forms, as there is a slight bias towards longer forms for the protein left on the beads (if reproducible).

We take the reviewer’s point but think that this detail likely reflects the fact that extremely long ubiquitin chains tend to stick to beads non-specifically due to reduced solubility.

An additional issue for this analysis concerns whether the intrinsic or specific activity of p97 and Cdc48 for ATP binding, hydrolysis and unfolding are equivalent. This is especially important for interpreting the activity of p97 towards the "Low" sample. Do the authors have a way to measure the specific activity of their p97 and Cdc48 preps? The proteins themselves could be differentially prone to damage or inactivation during purification, thereby creating distinct specific activities for the two preps that could affect the interpretation.

In Figure 3 of the revised manuscript we perform ‘species swap experiments’ that indicate that the high ubiquitin threshold of human p97-UFD1-NPL4 reflects the properties of human UFD1-NPL4, rather than the properties of human p97. We show that yeast Cdc48 functions with human UFD1-NPL4 but the hybrid unfoldase now behaves like human p97-UFD1-NPL4. Similarly, yeast Ufd1-Npl4 works with human p97 and the hybrid unfoldase behaves more like yeast Cdc48-Ufd1-Npl4.

In the end, I am not sure that the conclusion "These findings indicated that human p97-UFD1-NPL4 has a much higher ubiquitin threshold than yeast Cdc48-Ufd1-Npl4." Is that strong based on the data in Figure 2.

We take the reviewer’s point but now provide extensive new data in the revised manuscript to establish the point, as discussed above.

Also, in figure 2A and text, I am not sure the words "limited" and "efficient" are the best terms to use. Chain lengths may be comparable in both cases. Maybe something like "site-specific" for the K29-only version and "multi-site" for the other.

Once again, we take the reviewer’s point and in the revised manuscript we have removed ‘limited’ and ‘efficient’ and simply described the concentrations of E2 and E3. Importantly, we show clearly in the revised manuscript that ‘low ubiquitylation conditions’ (2.5 nM E2 and 1 nM E3) lead to a single ubiquitin chain on CMG-Mcm7, whereas ‘high ubiquitylation conditions’ (25 nM E2 and 10 nM E3) correspond to 2-3 ubiquitin chains on CMG-Mcm7 (Figure 1—figure supplement 1). We show that p97-UFD1-NPL4 (without UBX proteins) disassembles the latter but not the former (Figure 2 and Figure 2—figure supplement 1). However, this does not require either multiple chains or linkages other than K48-linked ubiquitin chains (Figure 2—figure supplement 3). So the simplest interpretation is that extensively ubiquitylated CMG-Mcm7 favours productive binding to p97-UFD1-NPL4.

Given the inability of the p97 complex to unfold the CMG under "Low" ubiquitylation conditions, it might be worth asking whether increasing the amount of p97 complex in the reaction can overcome the barrier. This could potentially address issues related to reduced specific activity. It could also potentially address reduced Ub affinity for the p97 complex versus the Cdc48 complex.

We thank the reviewer for this suggestion – Figure 2—figure supplement 2 now shows that increasing the concentration of p97-UFD1-NPL, to a concentration 10X higher than needed to disassemble extensively ubiquitylated CMG, does not lead to efficient disassembly of CMG with a single chain of up to ~15 ubiquitins.

2. The data in Figure 3 are said to be based on CMG complexes that had up to ~12 ubiquitin conjugated to the Mcm7 subunit. This figure is also subject to the same issue as above – can the authors distinguish between one long chain versus some amount of multi-site ubiquitylation and shorter chains in this case?

The data now in Figure 1—figure supplement 1B+D show that the conditions used in Figure 3 involve predominantly one chain on CMG-Mcm7 (sometimes 2), almost exclusively K48-linked.

3. The deletion analysis in Figure 4 doesn't seem to example the extent to which the various mutants assemble with the p97 complex. One presumes a correlation, but this wasn't tested directly.

Figure 5—figure supplement 1 shows that the truncations of FAF1 and UBXN7 behave as expected (deletion of UBX blocks interaction with p97-UFD1-NPL4 but other truncated alleles can still bind the unfoldase).

One line 216, it is stated: "These findings indicated that the coiled-coil domain of FAF1 has a previously unanticipated role in stimulating the unfoldase activity of p97-UFD1-NPL4." However, it is possible that without this coiled-coil, the UBX domain doesn't associate effectively with the p97 complex. This would be sort of a trivial result, in that there isn't a "direct" role of the copied-coil, but it is simply required for the UBX domain to interact. Can the coiled-coil be replaced by an alternative coiled-coil sequence or is there something special about this coiled-coil?

Figure 5—figure supplement 1E-F shows that FAF1 UBX can interact with p97 without any requirement for the coiled coil, as predicted by past work (e.g. Kim et al., 2011, Proteins, 79, 2583-2587, which reported a crystal structure of the FAF1 UBX with the p97 N-terminus)

4. Although a bit hard to know for sure, it looks like the ubiquitylated forms of Mcm7 in the ES cell experiment (Figure 5) have much shorter chains than the experiments with yeast CMG – especially in 5b. Of course it is possible that this reflects the presence of DUB activity during the IP, which would tend to shorten the chains. But at face value, it looks like FAF1/UBXN7 can allow p97 to operate on very short chains (although one really cant distinguish this from multi-mono ubiquitylation).

This raises questions about the overall conclusion. If p97 inhibitor increases the appearance of MCM7 conjugates (albeit with shorter chains) in the FAF1/UBXN7 knockout substantially, this suggest that p97 can disassemble shorter chains in cells without an adaptor OR that there is an additional UBX protein in cells that can do the job. Of course if the ubiquitylation levels are a result of DUB activity, that could explain it but it seems like this should be ruled out.

The disassembly of ubiquitylated CMG is blocked when cells are treated with p97 inhibitor, but it is likely that p97 inhibition also reduces the pool of free ubiquitin, leading indirectly to shorter ubiquitin chains on CMG-MCM7 (Figure 6B of the revised manuscript, discussed on lines 444-445). Similarly, previous studies with yeast (Maric et al., 2014), C. elegans (Sonneville et al., 2017) and mouse ES cells (Villa et al., 2021) have shown that p97 inhibition leads not only to accumulation of short ubiquitin chains on CMG-MCM7, but also to the accumulation of a significant fraction of CMG with non-ubiquitylated CMG-MCM7 (despite addition of DUB inhibitors such as propargylated ubiquitin to cell pellets and extracts), again most likely reflecting ubiquitin depletion upon inactivation of p97-UFD1-NPL4.

Reviewer #2 (Recommendations for the authors):

1) The authors nail down the critical activity to the coiled-coil domain of the accessory factors FAF1/FAF2 (in addition to the UBX domain that binds p97). This is surprising and interesting because the UBA domains of FAF1 or FAF2 are not required. Frustratingly, it is not revealed what the function of the coiled-coil region is, which limits the mechanistic understanding. The authors speculate that the coiled coil region binds ubiquitin. This needs to be tested because it is a simple experiment. The comparison to Pup is inappropriate because Pup is not homologous to ubiquitin. Alternatively, the coiled coil region could modulate p97 or Ufd1-Npl4 binding activity. The question would be whether the coiled coil region binds p97 and/or Ufd1-Npl4. Technically, this can be tested.

Figure 5—figure supplement 1E-F shows that the coiled coil of FAF1 is not needed for interaction of the UBX with p97-UFD1-NPL4.

Figure 5—figure supplement 1C-D (FAF1-∆480) shows that the FAF1 coiled coil does not bind detectably to UFD1-NPL4 in the absence of p97.

We could not detect stable binding of the FAF1 coiled-coiled to K48-linked ubiquitin chains in vitro (Figure 5—figure supplement 3). However, the same is true for UFD1-NPL4-∆NZF (Figure 3—figure supplement 2), which nevertheless contains previously characterised low affinity ubiquitin-binding modules that we show are sufficient to support the disassembly of ubiquitylated CMG in reconstituted assays (Figure 3—figure supplement 2C, p97-UFD1-NPL4-∆NZF). Therefore, we cannot exclude that the FAF1 coiled coil also functions by binding to both p97-UFD1-NPL4 and ubiquitin, with ubiquitin binding being low affinity yet functionally relevant.

In our opinion, further mechanistic insight into the action of the coiled coil domain will require structural studies of early intermediates in the unfolding reaction, which are considerably beyond the scope of the present study.

2) The biochemical experiments show single time points. In contrast, the field has reached a point where kinetic data can be provided. That would allow to quantitatively define even smaller differences such as the deltaUBA or UIM effect in Figure 4C

Figure 2—figure supplement 1 presents a time course for the key experiment that illustrates the high ubiquitin threshold of human p97-UFD1-NPL4. Although we agree with the reviewer that further timecourses might have been applicable elsewhere in our study, the revision involves five months work leading to 11 new figures (plus new data in other figures). This meant that we had to prioritise our time and additional timecourse experiments were simply not possible.

3) The ES cell microscopy in Figure 5C needs to be quantified by an automated, unbiased pipeline.

We take the reviewer’s point but automated quantification is not easily applicable to such experiments, given the complex growth of mouse ES cells in three dimensions, which necessitates manual identification of cells to be counted. Moreover, we had to examine movie data manually to identify cells that entered mitosis and stayed in focus to allow quantification. The methods are now explained on lines 1061-1079.

4) Figure 6, cell viability and survival: The information gleaned from these experiments is questionable. Effects on colony formation are mostly mediated by MLN4924. The additional effect by the deletions is minor, and needs to be quantified and statistically validated, also to clarify whether it is additive or synthetic. The same applies to the viability assay. More importantly, it is impossible to connect the effects to replication termination. Firstly, there are many processes affected by the manipulations. Secondly, the MLN4924 treatment is so short that it will affect replication termination only in a small percentage of cells.

We have repeated the experiments formerly in Figure 6 (now presented in Figure 7) and the revised manuscript now presents quantification and statistical validation.

The reviewer questioned the connection of this experiment to DNA replication termination, but this experiment was not meant to be about DNA replication. p97-UFD1-NPL4 affect many areas of cell biology, and our data predict that UBX proteins such as UBXN7 and FAF1 (and others) will be important in many other processes (not just replication). The point of the experiment was to explore such broad significance of FAF1 and UBXN7, taking these two factors as representatives of UBX proteins that reduce the ubiquitin threshold of p97-UFD1-NPL4. Our data predicted that FAF1 and UBXN7 should become particularly important in diverse areas of cell biology, whenever ubiquitin chains on substrates are short. We tested this experimentally by impairing the activity of cullin ubiquitin ligases (as a major class of E3 enzymes in cells) by treatment with the neddylation inhibitor MLN4924. The data now in Figure 7B illustrate that cells lacking UBXN7 and FAF1 are more sensitive to MLN4924 than control cells, whilst Figure 7C shows that expression of FAF1 wild type (but not FAF1-∆UBX or FAF1-∆CoiledCoil) rescues this defect. Figure 7D shows that expression of UBXN7 rescues the MLN4924 sensitivity of UBXN7∆ FAF1∆ cells to the level of cells that just lack FAF1 (comparison with Figure 7B, FAF1∆).

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

Overall, the authors have done extensive work to try to address the previous reviewers' comments. The authors have addressed all of the specific comments, and in general, the results support the conclusions. One issue is that in Figure 6D, the p-value is only 0.09, which is not strongly significant. This may have to do with the assay itself. The majority of experiments employ ubiquitin smears to monitor p97 activity as it relates to the length dependence on the biological processes.

We now note in the legend to Figure 6D (lines 1592-1595) the weak statistical significance of the data and we note that this contrasts with Figure 6—figure supplement 4E, where the difference between UBXN7∆ FAF1∆ and cells lacking either UBXN7 or FAF1 is statistically significant.

https://doi.org/10.7554/eLife.76763.sa2

Article and author information

Author details

  1. Ryo Fujisawa

    The MRC Protein Phosphorylation and Ubiquitylation Unit, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1985-1668
  2. Cristian Polo Rivera

    The MRC Protein Phosphorylation and Ubiquitylation Unit, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1433-4111
  3. Karim PM Labib

    The MRC Protein Phosphorylation and Ubiquitylation Unit, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Conceptualization, Supervision, Funding acquisition, Visualization, Methodology, Writing – original draft, Project administration
    For correspondence
    kpmlabib@dundee.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8861-379X

Funding

Medical Research Council (MC_UU_12016/13)

  • Karim PM Labib

Cancer Research UK (C578/A24558)

  • Karim PM Labib
  • Ryo Fujisawa

Japan Society for the Promotion of Science (202160572)

  • Ryo Fujisawa

Cancer Research UK (C578/A25671)

  • Karim PM Labib
  • Cristian Polo Rivera

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful for the support of the Medical Research Council (core grant MC_UU_12016/13) and Cancer Research UK (Programme Grant C578/A24558 to KL and PhD studentship C578/A25671 to CPR). RF is supported by a JSPS Overseas Research Fellowship from the Japan Society for the Promotion of Science (202160572). We thank Tom Deegan for help with reconstituted ubiquitylation reactions with budding yeast proteins, Yisui Xia for help with equivalent reactions involving C. elegans factors, Lee Armstrong and Yogesh Kulathu for recombinant K48-linked ubiquitin chains and Fabrizio Villa for assistance with mouse ES cell work. We are also grateful to Axel Knebel and Clare Johnson for purified Uba1 and Ubiquitin and MRC PPU Reagents and Services (https://mrcppureagents.dundee.ac.uk) for antibody production.

Senior Editor

  1. David Ron, University of Cambridge, United Kingdom

Reviewing Editor

  1. Wade Harper, Harvard Medical School, United States

Publication history

  1. Received: January 4, 2022
  2. Preprint posted: January 14, 2022 (view preprint)
  3. Accepted: August 2, 2022
  4. Accepted Manuscript published: August 3, 2022 (version 1)
  5. Version of Record published: August 15, 2022 (version 2)

Copyright

© 2022, Fujisawa et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,136
    Page views
  • 405
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Ryo Fujisawa
  2. Cristian Polo Rivera
  3. Karim PM Labib
(2022)
Multiple UBX proteins reduce the ubiquitin threshold of the mammalian p97-UFD1-NPL4 unfoldase
eLife 11:e76763.
https://doi.org/10.7554/eLife.76763

Further reading

    1. Biochemistry and Chemical Biology
    Holly Y Chen, Manju Swaroop ... Anand Swaroop
    Research Article

    Ciliopathies manifest from sensory abnormalities to syndromic disorders with multi-organ pathologies, with retinal degeneration a highly penetrant phenotype. Photoreceptor cell death is a major cause of incurable blindness in retinal ciliopathies. To identify drug candidates to maintain photoreceptor survival, we performed an unbiased, high-throughput screening of over 6,000 bioactive small molecules using retinal organoids differentiated from induced pluripotent stem cells (iPSC) of rd16 mouse, which is a model of Leber congenital amaurosis (LCA) type 10 caused by mutations in the cilia-centrosomal gene CEP290. We identified five non-toxic positive hits, including the lead molecule reserpine, which maintained photoreceptor development and survival in rd16 organoids. Reserpine also improved photoreceptors in retinal organoids derived from induced pluripotent stem cells of LCA10 patients and in rd16 mouse retina in vivo. Reserpine-treated patient organoids revealed modulation of signaling pathways related to cell survival/death, metabolism, and proteostasis. Further investigation uncovered dysregulation of autophagy associated with compromised primary cilium biogenesis in patient organoids and rd16 mouse retina. Reserpine partially restored the balance between autophagy and the ubiquitin-proteasome system at least in part by increasing the cargo adaptor p62, resulting in improved primary cilium assembly. Our study identifies effective drug candidates in preclinical studies of CEP290 retinal ciliopathies through cross-species drug discovery using iPSC-derived organoids, highlights the impact of proteostasis in the pathogenesis of ciliopathies, and provides new insights for treatments of retinal neurodegeneration.

    1. Biochemistry and Chemical Biology
    Lisa Goebel, Tonia Kirschner ... Daniel Rauh
    Short Report

    Mutations within Ras proteins represent major drivers in human cancer. In this study, we report the structure-based design, synthesis, as well as biochemical and cellular evaluation of nucleotide-based covalent inhibitors for KRasG13C, an important oncogenic mutant of Ras that has not been successfully addressed in the past. Mass spectrometry experiments and kinetic studies reveal promising molecular properties of these covalent inhibitors, and X-ray crystallographic analysis has yielded the first reported crystal structures of KRasG13C covalently locked with these GDP analogues. Importantly, KRasG13C covalently modified with these inhibitors can no longer undergo SOS-catalysed nucleotide exchange. As a final proof-of-concept, we show that in contrast to KRasG13C, the covalently locked protein is unable to induce oncogenic signalling in cells, further highlighting the possibility of using nucleotide-based inhibitors with covalent warheads in KRasG13C-driven cancer.