Transfer of Ho Endonuclease and Ufo1 to the Proteasome by the UbL-UbA Shuttle Protein, Ddi1, Analysed by Complex Formation In Vitro

The F-box protein, Ufo1, recruits Ho endonuclease to the SCFUfo1 complex for ubiquitylation. Both ubiquitylated Ho and Ufo1 are transferred by the UbL-UbA protein, Ddi1, to the 19S Regulatory Particle (RP) of the proteasome for degradation. The Ddi1-UbL domain binds Rpn1 of the 19S RP, the Ddi1-UbA domain binds ubiquitin chains on the degradation substrate. Here we used complex reconstitution in vitro to identify stages in the transfer of Ho and Ufo1 from the SCFUfo1 complex to the proteasome. We report SCFUfo1 complex at the proteasome formed in the presence of Ho. Subsequently Ddi1 is recruited to this complex by interaction between the Ddi1-UbL domain and Ufo1. The core of Ddi1 binds both Ufo1 and Rpn1; this interaction confers specificity of SCFUfo1 for Ddi1. The substrate-shield model predicts that Ho would protect Ufo1 from degradation and we find that Ddi1 binds Ho, Ufo1, and Rpn1 simultaneously forming a complex for transfer of Ho to the 19S RP. In contrast, in the absence of Ho, Rpn1 displaces Ufo1 from Ddi1 indicating a higher affinity of the Ddi1-UbL for the 19S RP. However, at high Rpn1 levels there is synergistic binding of Ufo1 to Ddi1 that is dependent on the Ddi1-UbA domain. Our interpretation is that in the absence of substrate, the Ddi1-UbL binds Rpn1 while the Ddi1-UbA binds ubiquitin chains on Ufo1. This would promote degradation of Ufo1 and disassembly of SCFUfo1 complexes.


Introduction
The Ubiquitin-proteasome system has a major role in regulation of cellular processes, in particular the cell cycle and many signaling pathways [1,2]. Proteins targeted for degradation are conjugated to ubiquitin (Ub) by a cascade of enzymes, an E1 Ub activatingand E2 Ub conjugating enzyme, and an E3 Ub ligase responsible for substrate identification [3]. In some instances an E4 Ub chain elongating activity is also involved [4]. Ub chains comprising at least four K48-linked Ub molecules are recognized by the 19S Regulatory particle (RP) of the proteasome, either by an endogenous 19S RP subunit [5][6][7], or by a member of the UbL-UbA protein family. UbL-UbA proteins bind specific 19S RP subunits through their Ub-like (UbL) domain and K48-Ub chains on the substrate through their Ub-associated (UbA) domain. The yeast family of UbL-UbA proteins comprises Rad23, Dsk2, and Ddi1, and each family member participates in the degradation of a range of substrates either by itself, or as a Rad23-Dsk2 pair (reviewed in [8]).
UbL-UbA proteins are often referred to as shuttle proteins based on their recruitment of the ubiquitylated substrate from the E2-E3 complex and transfer to the 19S RP. This is supported particularly by the interaction between Rad23 and Dsk2 with the chain elongating E4, Ufd2, that occurs in the framework of a complex between Ufd2 and the AAA-ATPase ring hexamer, Cdc48 [9]. However, many E3s bind the 19S RP directly: these include Ubr1 and Ufd4 [10], Hul5 [11], Ufo1 [12], SCF (Skp1-Cullin1-F-box protein) and APC (Anaphase Promoting complex) [13,14]. In the case of Ufd4, direct interaction between the E3 and the proteasome is essential for substrate degradation [15]. In some instances the UbL-UbA protein may be an essential stochiometric subunit of the E3 complex, as reported for KPC2 (Kip1 ubiquitylation-promoting complex 2) that regulates degradation of the p27 cell cycle inhibitor [16]. These reports raise the question whether other UbL-UbA proteins may also occur as intrinsic components of an E3-19S RP complex and if so whether it is possible to detect additional interactions between the core domain of the UbL-UbA protein and subunits of this complex. In the event of such interactions are they a prerequisite for interaction of the E3 complex with the 19S RP?
The SCF complex comprises a rigid cullin scaffold, in S. cerevisiae Cdc53, with the RING protein, Rbx1, attached to a C-terminal domain [17]. The RING domain serves as a landing pad for the Ub-charged E2, Cdc34 [18]. Substrate recruitment is executed by a series of different F-box proteins (FBPs), each of which binds a subset of targets many of which are recognized by phosphorylation [19][20][21]. FBPs have a F-box domain and a WD40-or LRR substrate-binding domain. The F-box domain binds the Skp1 adaptor that interacts with the N-terminal domain of Cdc53 [17,[22][23][24][25]. Exchange of FBPs within the SCF complex is achieved by auto-ubiquitylation of the FBP followed by degradation in the proteasome [26,27]. A number of FBPs of SCF complexes and the related BTB/3-box domain receptor proteins [28] have been shown to occur as homo-or heterodimers. These include homodimers of yeast Cdc4 and of human Fbw7 FBP [29] and the heterodimeric S. pombe Pop1-Pop2 FBPs [30,31]. FBP dimerization was shown to be required for dimerization of Cdc53 [32] and increasing experimental data support a model of a dimeric cullin-RING ligase complex. Indeed although monomeric FBPs bind their substrates and Skp1, substrate ubiquitylation was reported to require their dimerization [28,32,33].
Ddi1 is required for the final stages of proteasomal degradation of both Ho endonuclease [34] and of Ufo1, its cognate FBP [35]. Ubiquitylated Ho interacts with the UbA domain of Ddi1 via its ubiquitin chains and its transfer to the 19S RP requires the UbL domain of Ddi1 that interacts with the LRR domain of the 19S RP subunit, Rpn1 [36]. Ddi1 forms a homodimer mediated by residues in its core (residues 180-325) giving rise to an active aspartyl protease site [37,38]. In ddi1D mutants ubiquitylated Ho endonuclease accumulates in the cytoplasm and is not transferred to the proteasome for degradation [34]. Ufo1 and its fungal orthologs are unique FBPs as they have four copies of the Ub interacting motif (UIMs) at their C-terminus in addition to the Fbox and WD40 domains present in other FBPs [35,39,40]. The UIM is a simple a-helical ubiquitin binding domain [41] and the Ufo1-UIMs are separated by long linkers suggesting this is a flexible region. Turnover of Ufo1 is dependent on an interaction between its UIMs and the UbL domain of Ddi1 [35]. Furthermore the rpn1-D517A mutation that disrupts binding of Ddi1 to the proteasome stabilizes Ufo1 [36].
A protein fragment comprising the Ufo1-UIMs interacts with all three UbL-UbA proteins, Rad23, Dsk2, and Ddi1, however fulllength (FL) Ufo1 interacts only with Ddi1 suggesting that the core residues are important for specificity [35]. UIMs have been shown to interact with Ub-charged E2s to promote monoubiquitylation of a different domain of their host protein [42][43][44][45]. Deletion of UFO1 has no obvious phenotype under normal growth conditions, however, a genomic UFO1 allele deleted for the UIMs is dominant lethal. Ectopic high level expression of UFO1 without the UIMs leads to stabilization of the protein and to cell cycle arrest at the end of G 1 . Substrates of other FBPs accumulate suggesting that Ddi1 is required for disassembly of SCF Ufo1 complexes and recycling of the core complex subunits into alternative SCF complexes [35].
Here we used complex reconstitution in vitro to augment our in vivo data showing a role for Ddi1 in degradation of Ho and of its cognate FBP, Ufo1 [34,35,40,46]. In particular we aimed to identify stages in the handover of Ho from the SCF Ufo1 complex to the 19S RP and subsequent degradation of Ufo1. We delineate stages in the formation of SCF Ufo1 -Ho-Ddi1-19S RP complex. Domain analysis showing different modes of interaction of Ddi1 with Ufo1 and Rpn1 in the presence and absence of Ho support the ''Substrate shield'' model of protein degradation [47]. We present a model for sequential handover of Ho and Ufo1 to the proteasome.

Ufo1 Forms Dimers Initiated by the UIMs
Given the importance of FBP dimerization for substrate ubiquitylation [28,32,33] we examined whether Ufo1 forms a dimer. Furthermore we aimed to determine which domain(s) of Ufo1 could have a role in dimerization. We incubated yeast extract from cells that produced full-length (FL), GFP FL-Ufo1, or Ufo1 truncated for the C-terminal UIMs, GFP Ufo1Duims, with GSH beads bound to recombinant GST FL-Ufo1, GST Ufo1-WD40 domain, GST Ufo1-UIMs, or control GST ( Figure S1).
We observed a robust interaction of GFP FL-Ufo1 with GST FL-Ufo1 and with GST Ufo1-UIMs whereas the interaction between GFP FL-Ufo1 and GST Ufo1-WD40 domain was extremely weak. GFP Ufo1Duims did not interact with GST FL-Ufo1 or with GST Ufo1-UIMs. However, in contrast to GFP FL-Ufo1, truncated GFP Ufo1Duims interacted robustly with GST Ufo1-WD40 ( Figure 1A).
These results suggest both a positive and a negative role for the Ufo1-UIMs in Ufo1 dimerization. The positive role is indicated by the ability of FL-Ufo1 to dimerize with both FL-Ufo1 and with the isolated Ufo1-UIM fragment, whereas the negative role is indicated by the absence of dimerization between FL-Ufo1 and the Ufo1-WD40 domain fragment. This may indicate that the Ufo1-UIMs regulate access to the WD40 domain. To test directly whether the Ufo1-UIMs dimerize we incubated yeast extract with GFP Ufo1-UIMs with recombinant GST Ufo1-UIMs on beads. We observed a robust interaction that was not found with the control GST beads indicating that isolated Ufo1-UIMs fragments dimerize ( Figure 1B). The interaction between GFP Ufo1Duims and GSTUfo1-WD40 ( Figure 1A) suggests that the Ufo1-WD40 domain by itself can dimerize. Indeed when we expressed the Ufo1-WD40 domain in bacteria with two different epitope tags we observed that GST Ufo1-WD40 bound to HIS Ufo1-WD40 ( Figure 1C). Thus Ufo1 resembles other FBPs in forming dimers and both the unique Ufo1-UIMs and the Ufo1-WD40 domain participate in dimerization. Dimerization via the Ufo1-WD40 domains is supported by our previous finding of turnover of Ho in ufo1D mutants that produce plasmid-encoded Ufo1Duims [35].

SCF Ufo1 Complexes Interact with the 19S RP in vitro Only in the Presence of Substrate
Despite its nuclear role Ho must exit the nucleus to be degraded [46] and in ddi1D mutants stabilized Ho accumulates in the cytoplasm as an ubiquitylated conjugate [34]. SCF Ufo1 complexes that have bound Ho may associate with the 19S RP as reported for SCF Cdc4 -Sic1 complexes [14], or alternatively Ddi1 could shuttle ubiquitylated Ho from a SCF Ufo1 -Ho complex to the proteasome. We therefore reconstituted SCF Ufo1 complexes in vitro in the presence or absence of Ho. Recombinant GST FL-Ufo1 and the GST Ufo1-WD40 domain proteins on GSH beads were incubated with yeast extract from cells that produced myc Cdc53 and with the 19S RP complex tagged with Rpn11 GFP . The experiment was performed both in the presence and the absence of GFP Ho endonuclease. Experimental conditions are such that the 19S RP complex with a single tagged subunit remains intact in the yeast extract [5,13,14,36,51]. Both FL-Ufo1 and the Ufo1-WD40 domain on beads supported the formation of SCF Ufo1 -Ho-19S RP complexes and interacted with yeast myc Cdc53, GFP Ho, and with the tagged 19S RP complex. In addition endogenous Ddi1 was present as a major component of the GST FL-Ufo1 and the GST Ufo1-WD40 domain bead fractions of complexes formed in the presence of Ho. In the absence of Ho, we found an interaction of GST Ufo1 with myc Cdc53, but there was no interaction with Rpn11 GFP . Ddi1 could still be detected in the GST FL-Ufo1 and the GST Ufo1-WD40 domain bead fractions, although in a considerably diminished amount ( Figure 2A). A similar result was observed using tagged Rpn1 GFP ( Figure S2). Rpn12 was present in the bead fraction indicating that 19S RP complexes and not just the tagged subunit were interacting with SCF Ufo1 ( Figure S3).
Ddi1 is involved in the final stages of transfer of Ho and of Ufo1 to the 19S RP and could be recruited to the SCF Ufo1 -Ho-19S RP complex after its assembly. We therefore repeated the above experiment using extracts of transformed ddi1D mutants. As in w.t. cells, Ho was crucial for formation of complex between SCF Ufo1 and the19S RP, however, there was no requirement for Ddi1 for formation of the SCF Ufo1 -Ho-19S RP complex ( Figure 2B). These results taken together and supported by our in vivo data that show that both Ho and Ufo1 accumulate as ubiquitylated conjugates in ddi1D mutants suggest that in vivo Ddi1 is recruited to the SCF Ufo1 -Ho-19S RP complex after its assembly. SCF Ufo1 -Ho-Ddi1-19S RP Complexes can be Reconstituted in vitro with Immobilized GST Ddi1 or GST Rpn1 Reconstitution of SCF Ufo1 -Ho-19S RP complexes in vitro in the above experiments was achieved with GST FL-Ufo1 or the GST Ufo1-WD40 domain on beads. To determine whether complex reconstitution is also possible with immobilized GST Ddi1 or GST Rpn1, the 19S RP subunit bound by Ddi1 [36,52], we incubated GST Ddi1 or control GST on GSH beads with yeast Figure 1. Ufo1 forms a homodimer via its UIMs. A. GST FL-Ufo1, GST Ufo1-WD40 domain, GST Ufo1-UIMs or GST beads were incubated with yeast extract from cells expressing full-length pGAL-GFP-UFO1 or pGAL-GFP-UFO1Duims. The bead fraction was analysed by Western blotting with anti-GFP and anti-GST antibodies. T is 10% of yeast extract with which the beads were incubated. *denotes contaminant band. B. Recombinant GST Ufo1-UIMs or control GST beads were incubated with yeast extract with GFP Ufo1-UIMs and analysed as above. T is 10% of yeast extract as above. C. Recombinant GST Ufo1 WD40 domain protein or control GST on GSH beads were incubated with bacterial lysate from cells that expressed HIS Ufo1-WD40 and the bead fraction was analysed by Western blotting initially with anti-HIS and then with anti-GST antibodies. T is 10% of yeast extract as above. The brackets around HIS Ufo1-WD40 in the anti-GST Western blot indicate that these bands were observed after incubation with anti-HIS antibodies as shown in the upper part of the blot. doi:10.1371/journal.pone.0039210.g001 extract from cells that produced myc Cdc53 and GFP Ufo1. We observed a robust interaction of both proteins with GST Ddi1 that was not observed with the GST control beads ( Figure 3A). Similarly GST Rpn1 on beads could reconstitute SCF Ufo1 -GF-P Ho-GST Rpn1 complexes that included endogenous Ddi1 present in the yeast extract ( Figure 3B). No complexes were formed with the control GST beads. Thus it is possible to reconstitute complexes in vitro irrespective of which component is immobilized. Recombinant HIS Ufo1 interacted extremely weakly with irrelevant control GST Rpn10 beads, however we did observe an interaction of GFP Ho and of myc Cdc53 with this 19S RP subunit.
The Core of Ddi1 Binds Cdc53, the Ufo1-WD40 Domain, and Rpn1 The Ufo1-UIMs fragment in isolation interacts with all three UbL-UbA proteins, Rad23, Dsk2, and Ddi1, however, FL-Ufo1 discrim-inates between them [35]. This suggests that the initial interaction between Ufo1 and Ddi1 occurs via interaction of its UIMs with the Ddi1-UbL domain and that specificity of UbL-UbA protein may be conferred by further interactions between Ufo1 and the core of Ddi1. We subcloned HIS DDDdi1 without the UbL and UbA domains comprising residues 180-325. Indeed both the GST Ufo1-WD40 domain and GST Cdc53 bound core HIS DDDdi1 ( Figure 3C). Ddi1 binds the LRR domain of the Rpn1 subunit of the 19S RP [36,52] via its UbL domain and here we found that the core HIS DDDdi1 fragment bound GST Rpn1 robustly but showed only extremely weak binding to control GST Rpn10 ( Figure 3D). Thus after the initial interaction between the Ufo1-UIMs and the Ddi1-UbL these additional interactions with the Ddi1 core could secure Ddi1 within the SCF Ufo1 -Ho-Ddi1-19S RP complex. They could also allow flexibility to the Ddi1-UbL allowing it to switch to binding Rpn1 for substrate or FBP transfer. A. GST Ufo1, GST Ufo1-WD40 domain, or control GST beads were incubated with yeast extract from cells with tagged genomic RPN11-GFP that were transformed with pGAL-MYC-CDC53 either with pGAL-GFP-HO or alone. The bead fraction was analysed by Western blotting with anti-GFP antibodies to detect Rpn11 GFP and GFP Ho, with anti-myc antibodies to detect myc Cdc53, and with anti-Ddi1 and anti-GST antibodies. T is 10% of total yeast extract with which the beads were incubated (Lanes 1 and 2). Lane 3: GST Ufo1 beads incubated with myc Cdc53, Rpn11 GFP and GFP Ho; Lane 4: GST Ufo1 WD40 domain incubated with myc Cdc53, Rpn11 GFP and GFP Ho; Lane 5: control GST beads incubated with these yeast extracts; Lane 6: GST Ufo1 beads incubated with myc Cdc53 and Rpn11 GFP ; Lane 7: GST Ufo1 WD40 domain incubated with myc Cdc53 and Rpn11 GFP ; Lane 8: control GST beads incubated with these yeast extracts. B. GST Ufo1, GST Ufo1-WD40, or control GST beads were incubated with yeast extract from ddi1D mutant cells that expressed pGAL-MYC-CDC53, pGFP-RPN11, with or without pGAL-GFP-HO. The bead fractions were analysed by Western blotting with anti-myc, anti-GFP, anti-Ddi1, and anti-GST antibodies as in A. T is 10% of total yeast extract with which the beads were incubated (Lanes 1-3). Lane 4: GST Ufo1 beads incubated with myc Cdc53, Rpn11 GFP and GFP Ho; Lane 5: GST Ufo1 WD40 domain incubated with myc Cdc53, Rpn11 GFP and GFP Ho; Lane 6: control GST beads incubated with these yeast extracts; Lane 7: GST Ufo1 beads incubated with myc Cdc53 and Rpn11 GFP ; Lane 8: GST Ufo1 WD40 domain incubated with myc Cdc53 and Rpn11 GFP ; Lane 9: control GST beads incubated with these yeast extracts. doi:10.1371/journal.pone.0039210.g002 . Immobilized Ddi1 and Rpn1 reconstitute SCF Ufo1 complexes in vitro. A. GST Ddi1 or control GST on GSH beads were incubated with yeast extract from cells that produced myc Cdc53 and GFP Ufo1. Analysis was by Western blotting with anti-myc, anti-GFP, and anti-GST antibodies. T represents 10% of the yeast extract with which the beads were incubated. B. GST Rpn1, GST Rpn10, or GST beads were incubated with yeast extract from cells that produced myc Cdc53 and GFP Ho mixed with bacterial lysate with recombinant HIS Ufo1. The bead fraction was analysed by Western blotting with anti-myc, anti-HIS, anti-GFP, anti-Ddi1, and anti-GST antibodies. T represents 10% of the yeast extract incubated with the beads. C. GST Ufo1 WD40 domain, GST Cdc53, or control GST on GSH beads were incubated with bacterial lysate from cells that produced recombinant HIS DDDdi1. The bead fraction was analysed by Western blotting with anti-HIS and with anti-GST antibodies as indicated. T is 10% of the HIS DDDdi1 bacterial lysate incubated with the beads. D. The HIS DDDdi1 bacterial lysate was incubated with GST Rpn1, GST Rpn10, or GST beads and analysed as above. doi:10.1371/journal.pone.0039210.g003 SCF Ufo1 -Ddi1-19S RP Complex Subunits Immunoprecipitate Together in the Presence of Ho The above experiments demonstrate that in the presence of Ho a SCF Ufo1 -Ho-Ddi1-19S RP complex is formed in vitro. To verify that this is indeed a complex we prepared a reaction mix comprising yeast extract with myc Cdc53, with or without GFP Ho, and bacterial lysate with GST Ufo1 and HIS Rpn1, and immunoprecipitated each tagged protein separately. In the presence of Ho, immunoprecipitation of myc Cdc53, of GFP Ho, of GST Ufo1 or of HIS Rpn1 led to reciprocal coimmunoprecipitation of the other three proteins and of Ddi1 present in the yeast extract. In the absence of Ho, immunoprecipitation of myc Cdc53, GST Ufo1 or HIS Rpn1 led to coimmunoprecipitation of endogenous Ddi1 from the yeast extract, but not of any of the other proteins of the complex formed in the presence of substrate. This result indicates that in the presence of Ho a bona fide complex is formed between SCF Ufo1 -Ho-Ddi1 and Rpn1. This complex does not form in the absence of Ho (Figure 4).

Ufo1 and Rpn1 Bind Ddi1 in Both a Competitive and a Synergistic Manner
(a) Competitive interaction: GST Rpn1 abrogates binding of GFP Ufo1 to HIS Ddi1. The Ddi1-UbL domain binds both the Ufo1-UIMs and Rpn1 [35,52], however, interaction between Ddi1 and Rpn1 is essential for turnover of Ufo1 [36]. Both Ufo1 and Rpn1 bind the core of Ddi1 ( Figure 3C and 3D) and this interaction may facilitate the switch of the Ddi1-UbL domain from the Ufo1-UIMs to Rpn1 for transfer of Ho or Ufo1 to the 19S RP. We therefore examined whether there is competition between Ufo1 and Rpn1 for interaction with Ddi1. Each protein incubated separately with Ddi1 beads was present in the HIS Ddi1 bead fraction ( Figure 5A, Lanes 4-6). However, Rpn1 displaced Ufo1 from Ddi1 when both GST Ufo1 and GST Rpn1 were incubated together with the HIS Ddi1 beads (Lane 7). In contrast addition of yeast extract with ubiquitylated GFP Ho to the reaction mix with either GST Ufo1 or GST Rpn1 did not affect the binding of either protein to HIS Ddi1 (Lanes 8 and 9). Furthermore, Ho in the reaction mix comprising Ufo1, Rpn1, and Ddi1, abrogated the competition between Ufo1 and Rpn1 and all three proteins bound the HIS Ddi1 beads (Lane 10) and Figure 2. Thus Ho protects Ufo1 from displacement from Ddi1 by Rpn1. In this complex the Ddi1-UbL would bind Rpn1, Ufo1 would be bound via its WD40 domain to Ho and to the Ddi1 core, and further interactions would occur between the Ddi1-UbA and the Ub chains on Ho. This is the complex we predict to underlie transfer of ubiquitylated Ho to the 19S RP ( Figure 6).
(b) Synergistic interaction: GST Rpn1 and GFP Ufo1 bind HIS Ddi1 in a tertiary complex that requires the Ddi1 UbA domain and does not involve the Ddi1 UbL domain. The competitive interaction between Ufo1 and Rpn1 may occur during handover of the FBP to the 19S RP after degradation of Ho. To explore this hypothesis we examined whether exclusion of GST Ufo1 from binding to HIS Ddi1 by GST Rpn1 is concentration dependent. We calibrated the system by determining an amount for each lysate/extract that would give detectable binding of protein to the Ddi1 beads (x1, Figure 5B, Lanes 3 and 4). Then keeping the amount of GFP Ufo1 extract constant in a fixed reaction volume we increased the amount of GST Rpn1 lysate two-and threefold. In this experiment we used ubiquitylated GFP Ufo1 produced in yeast [35]. GST Rpn1 at x1 and x2 in the reaction mix gave a similar amount bound to the Ddi1 beads. Both these GST Rpn1 concentrations abrogated binding of GFP Ufo1 to Ddi1 ( Figure 5B, Lanes 5 and 6 and as observed in Figure 5A, Lane 7). However, x3 the amount of GST Rpn1 lysate induced synergistic binding of GST Rpn1 and GFP Ufo1 to the HIS Ddi1 beads. A similar although considerably weaker signal was obtained when core HIS DDDdi1 beads were used. In contrast binding of GST Rpn10 to the HIS Ddi1 beads was not affected by GST Ufo1 nor was any synergistic effect observed between them in binding to Ddi1 ( Figure 5C). In contrast to Ddi1 [36] there is no direct binding between Ufo1 and Rpn1 ( Figure 5D).
The competition between Ufo1 and Rpn1 for binding Ddi1 may involve the Ddi1-UbL which binds both proteins (above). To address this question we repeated the synergistic binding experiment described in Figure 5B but this time in addition to GST FL-Ddi1 beads we used Ddi1 that lacked either the UbL or UbA domain: GST Ddi1DUbL, and GST Ddi1DUbA, respectively ( Figure 5E, Lanes 1-3). Ddi1DUbL exhibited severely reduced binding to Rpn1 and did not bind Ufo1 when each protein was incubated separately with the beads. In contrast, Ddi1DUbL bound both Rpn1 and Ufo1 synergistically when both were present in the reaction mix. This suggests a role for the Ddi1-UbA in the synergistic binding of Rpn1 and Ufo1 to Ddi1. Surprisingly although Rpn1 binds the Ddi1-UbL, when we incubated HIS Rpn1 with GST Ddi1DUbA beads it interacted less strongly than with GST FL-Ddi1 beads ( Figure 5E, compare Lane 1 with Lane 4). GFP Ufo1 bound GST Ddi1DUbA beads and there was an extremely  PLoS ONE | www.plosone.org weak synergistic binding of Rpn1 and Ufo1 to GST Ddi1DUbA beads when both were present in the reaction mix. Our previous in vivo experiments indicated that Ufo1 and Ddi1 interact via the Ufo1-UIMs and the Ddi1-UbL [35]. We therefore substituted GFP Ufo1Duims for GFP FL-Ufo1. Indeed GFP Ufo1Duims did not interact with GST FL-Ddi1, GST Ddi1DUbL or GST Ddi1DUbA beads both in the presence or the absence of Rpn1 ( Figure 5F).

Discussion
Complex reconstitution in vitro indicated that SCF Ufo1 complexes that contain their substrate, Ho, are associated with the 19S RP. These complexes can assemble in the absence of Ddi1, however, in experiments with extracts from w.t. cells Ddi1 is found in association with the SCF Ufo1 -Ho-19S RP complex. Our interpretation is that Ddi1 is recruited to preformed SCF Ufo1 -Ho-19S RP complex. Based on our previous experiments in vivo we propose that Ddi1 enters the SCF Ufo1 -Ho-19S RP complex via initial interaction between the Ufo1-UIMs and the Ddi1-UbL ( [35] and Figure 5F). Subsequent interaction between the Ufo1-WD40 and the core of Ddi1 detected here could explain the specificity of the interaction of SCF Ufo1 for Ddi1 [35]. The recruitment of Ddi1 after formation of the SCF Ufo1 -Ho-19S RP complex supports our in vivo results that suggested Ddi1 is required for disassembly of SCF Ufo1 complexes after substrate degradation. This hypothesis is based on accumulation of ubiquitylated Ho in the cytoplasm of ddi1D mutants [34], stabilization of full-length Ufo1 in ddi1D mutants, cell cycle arrest at the G 1 -S interphase by overexpression of UFO1Duims in wild type cells or of full-length UFO1 in ddi1D mutants, and by the accumulation of Cln2, a substrate of the FBP, Grr1 [21], in cells with a high level of Ufo1Duims [35].
The Ufo1-UIMs promote dimerization of Ufo1 and are crucial for all interactions of Ufo1 with Ddi1. They may fulfill two roles in dimerization: one is physical interaction between the UIMs of two Ufo1 molecules to initiate dimerization. The other is regulation of access to the Ufo1-WD40 domain as full-length Ufo1 did not dimerize with an Ufo1-WD40 domain fragment. Thus dimerization may start at the C-terminal UIMs and proceed to include the Ufo1-WD40 domains. We previously reported that SCF complexes from cells that produced Ufo1Duims are capable of degrading Ho [35]. Given that dimerization of FBPs has been shown to be a prerequisite for substrate ubiquitylation in some instances [28,32,33], our current results support an interpretation that in the absence of its UIMs the Ufo1-WD40 domains of each monomer are able to interact with one another in vivo. The Ufo1-WD40 domain alone is sufficient for formation of complexes that include GFP Ho, the 19S RP, and Ddi1 and indeed in our yeast two-hybrid experiments we reported an interaction between Cdc53 and the Ufo1-WD40 domain [35]. This is unusual as the solved SCF structures do not display interaction between the cullin and the WD40 domain of the FBP [23,25] or with the related BTB/3-box domain receptor protein [28]. The Ufo1 WD40 sequence has a rather degenerate b-propeller sequence and a full analysis of this unusual interaction awaits solution of the 3D structure of Ufo1. A dimerization sequence has been identified in blotting with anti-GST, anti-GFP, anti-HIS and anti-ubiquitin antibodies. Lanes 1-3 (T) show 10% of the lysate/extract for bead incubation. B. Yeast extract with GFP Ufo1 was incubated with HIS Ddi1 and HIS DDDdi1 beads in the absence or the presence of increasing amounts of HIS Rpn1. The bead fractions were analysed by Western blotting with anti-GFP, anti-GST antibodies and anti-HIS. Lanes 1 and 2 (T) show 10% of the lysate/extract with which the beads were incubated. C. Bacterial lysate with GST Ufo1 was mixed with increasing amounts of lysate with GST Rpn1 or GST Rpn10 and incubated with nickel beads with HIS Ddi1. The Western blots were analysed with anti-GST and anti-HIS antibodies. D. Bacterial lysate with HIS Rpn1 was incubated with GST Ufo1, the GST Ufo1-WD40 domain, the GST Ufo1-UIMs or control GST beads. The Western blots were analysed with anti-HIS and anti-GST antibodies. T indicates 10% of the lysate with which the beads were incubated. E. Recombinant HIS Rpn1 made in bacteria and GFP Ufo1 from yeast extract were incubated alone or together with GSH beads bound to GST Ddi1, GST Ddi1DUbL, or GST Ddi1DUbA produced in bacteria. The bead fractions were analysed by Western blotting with anti-HIS and anti-GFP antibodies to show proteins that bound the GSH beads. The latter were detected with anti-GST antibodies. F. As above except that GFP Ufo1Duims was used instead of FL-Ufo1. T denotes 10% of the yeast extract incubated with the beads. doi:10.1371/journal.pone.0039210.g005 Figure 6. Model for sequential interactions of Ho, Ufo1, and Rpn1 with Ddi1. Panel 1. Active SCF Ufo1 -Ho complexed with the 19S RP recruits Ddi1 by interaction of the Ufo1-UIMs with the Ddi1-UbL domain ( [35] and Figures 2, 4 and 5F). Subsequently the core of Ddi1 binds the Ufo1-WD40 domain and Rpn1 ( Figure 3C and D). Both Ufo1 (Figure 1) and Ddi1 [38] form dimers but are drawn here as monomers for clarity. Panel 2. The Ddi1-UbA domain interacts with ubiquitin chains on Ho and the Ddi1-UbL binds Rpn1 for transfer of ubiquitylated Ho to the 19S RP [34]. At this stage Ho, Ufo1, and Rpn1 bind Ddi1 simultaneously ( Figure 5A). Panel 3. After degradation of Ho, Ufo1 can no longer bind Ddi1 in the presence of Rpn1 (competitive interaction, Figure 5B and C). However, at high levels of Rpn1 there is synergistic binding that is supported to a small extent by the Ddi1 core ( Figure 5B) and is totally dependent on the Ddi1-UbA domain ( Figure E). Based on the higher affinity of the Ddi1-UbL for Rpn1 seen in the competitive interaction we propose that at this stage the Ddi1-UbL binds Rpn1 and the Ddi1-UbA binds ubiquitin chains on Ufo1. This would lead to degradation of Ufo1 [36]. doi:10.1371/journal.pone.0039210.g006 the N-terminal region of certain FBPs [26,27] and it is conceivable that there is one in Ufo1 too that could serve for dimerization in the absence of the Ufo1 UIMs.
The Ddi1-UbL -Ufo1-UIMs interaction is essential for recruitment of Ddi1 to the SCF Ufo1 -Ho-19S RP complex ( [35] and Figure 5F). However, degradation of the ubiquitylated substrate requires transfer of the Ddi1-UbL from its interaction with the Ufo1-UIMs to Rpn1 [36]. In the presence of Ho a complex is formed that includes Ufo1, Rpn1, and Ddi1. Binding of Ho to Ddi1 is mediated by interaction of its ubiquitin chains with the Ddi1-UbA domain [34]. We propose that interaction of the Ddi1-UbA with a critical amount of Ub chains on Ho could lead to switching of the Ddi1-UbL domain from the Ufo1-UIMs to Rpn1 for transfer of Ho to the 19S RP. Transfer of the Ddi1-UbL without disruption of the complex between these proteins would be supported further by concurrent binding of the Ddi1 core to both Ufo1 and Rpn1 and by interactions of Ho with both the Ufo1-WD40 domain [40] and with the Ddi1-UbA domain via its Ub chains ( [34] and Figure 6).
The ''substrate shield'' model proposes that the substrate protects the FBP from degradation [47]. In the reaction lacking Ho (comparable to an in vivo situation after substrate degradation but prior to SCF Ufo1 complex disassembly) we observed two different modes of interaction of Ufo1 and Rpn1 with Ddi1: (a) competitive -Rpn1 excludes Ufo1 from binding to Ddi1; (b) synergistic -high levels of Rpn1 formed a tertiary complex between Ufo1, Ddi1, and Rpn1. The competitive interaction indicates that the Ddi1-UbL has higher affinity for Rpn1 than for the Ufo1-UIMs. The dependence of synergistic binding of Ufo1 and Rpn1 on the Ddi1-UbA domain suggests that Ub chains on Ufo1 are involved. The higher ratio of Rpn1 to Ufo1 in the in vitro Ddi1 synergistic binding experiment could parallel molecular crowding within the SCF Ufo1 -19S RP complex. Thus in the absence of Ho our data support a complex in which the Ddi1-UbL is bound to Rpn1 while the Ddi1-UbA domain binds Ub chains on Ufo1. This model for sequential transfer of Ho and of Ufo1 to the 19S RP is presented in Figure 6.
Growth media and yeast transformation by LiOAc are as in [50].
Immunoprecipitation and immunoblotting were performed as described in [5,40]. Briefly, proteins were induced from the GAL promoter by overnight growth in minimal medium with 2% galactose. Next morning the culture was diluted 1:3 and grown for a further 1.5 hours. 50 mls of logarithmic culture served as the source of a 300 ml extract with 80 mg/ml protein. 200 ml were taken for immunoprecipitation (IP) with the appropriate antibody and the immunoprecipitate was run in a single lane for Western blotting (WB).

Antibodies
Mouse anti-GFP (Roche Applied Science), mouse 9E10 antimyc (Enzo), and mouse anti-HIS (Sigma) antibodies were used at a dilution of 1:250 for IP and at 1:1,000 for WB; mouse anti-GST (Santa Cruz Biotechnology) antibodies were diluted 1:1,000 for IP and 1:2,000 for WB, rabbit anti-Ddi1 (gift from Jeffrey Gerst) was used at 1:5,000 for WB. Goat anti-mouse and anti-rabbit antisera, used at 1:1,000 were from Santa Cruz Biotechnology. Protein Asepharose was purchased from Amersham and used at 50%; 30 ml were added to each sample.
TCA precipitation proteins were precipitated from 300 ml cell extract by adding TCA to 10% with 10 minutes incubation on ice. The pellet was centrifuged at 12,000 g for 10 minutes and five volumes of cold acetone were added. The protein pellets were harvested and dried. For WB analysis the pellets were dissolved in 30 ml of sample buffer and 5 ml of each fraction was separated by SDS-PAGE.

Expression of GST and HIS Fusion Proteins in Rosetta Bacteria
Bacteria were transformed by electroporation and the colonies selected on LB-agar plates with ampicillin and kanamycin, each at 100 mg/ml, and chloramphenicol at 34 mg/ml. A single colony was grown in 1 liter of LB (with ampicillin and chloramphenicol) to an OD 600 of 0.6-0.8 (3-5 hours) with vigorous agitation at 37uC. IPTG was added to 0.4 mM to induce expression and the culture was incubated overnight at 20uC. The cells were harvested by centrifugation at 4uC for 10 min at 6,000 rpm. The cell pellet was washed with 20 ml of ice-cold PBS and resuspended in 3 ml yeast extract buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.1% NP40, 1:25 Protease Inhibitor cocktail (Roche)). The cell suspension was disrupted with an ultra-sound sonicator on ice using 6 cycles each of 10 seconds and clarified by centrifugation for 10 min at 4,000 rpm at 4uC. The supernatants with the GST-fusion proteins were incubated with Glutathionesepharose 4B (GSH) beads (Amersham Biosciences) prewashed in yeast extract buffer with 1% Triton-X100; HIS-fusion proteins were incubated with washed Ni-sepharose (Clontech) for 1.5 hr at 4uC. The bead fractions were washed 5 times in extract buffer with 2.5% Triton-X100. The GST-and HIS-fusion proteins on beads were stored at -20uC after addition of glycerol to 5%.

GST in vitro Binding Assay
Yeast cells were grown overnight to late log phase (OD 600 = 0.8) in 2% galactose medium for the GAL-regulated constructs, or in YePD. The cells were harvested by centrifugation at room temperature for 5 minutes at 4,000 rpm, washed in 50 ml TE and resuspended in 600 ml extract buffer. 0.5-0.6 mg of glass beads were added and the cells were broken by vigorous vortexing for 25 minutes at 4uC. The extract was clarified by centrifugation at 12,000 g for 20 minutes at 4uC and protein concentration was measured with the Bio-Rad protein reagent. 5-10 mg of protein extract were taken for each GST pull-down in a total volume of 350-400 ml extract buffer. 30-50 ml of 50% Glutathione Sepharose 4B beads coupled to GST fusion protein were added to each sample and incubated at 4uC for 1-2 hours with very mild shaking. The samples were washed 6 times with extract buffer with 2.5% Triton X100. The pellet was resuspended in 30-50 ml sample buffer x2, boiled for 5 minutes and centrifuged for 3 minutes at high speed to remove insoluble material. The supernatant was separated on a 12% polyacrylamide SDS gel with protein size standards followed by WB analysis.

HIS-tagged Protein in vitro Binding Assay
As above, but with 30-50 ml of 50% Ni Sepharose beads coupled to the HIS fusion protein added to each sample and incubated at 4uC for 1-2 hours with very mild shaking. The samples were washed 6 times with extract buffer with 2.5% Triton X100 and 100 mM Imidazole. The pellet was resuspended in 30-50 ml sample buffer x2, boiled for 5 minutes and centrifuged for 3 minutes at high speed to remove insoluble material. The supernatant was separated on a 12% polyacrylamide SDS gel with protein size standards followed by WB analysis.