Discussion

51

Additionally, the degradation of FAT10 was analyzed in pre1-1 yeast strain that exhibit reduced protein degradation and accumulation of ubiquitin-protein conjugates under stress conditions (Heinemeyer et al., 1991). Cycloheximide chase experiments were performed by transforming HA-FAT10 in pre1-1 and its parental wild-type yeast strain.

We observed the stabilization of FAT10 in pre1-1 strain under heat stress conditions (grown at 38ºC) as compared to the wild-type strain (Figure 2.7d). These data strongly suggests that the degradation of FAT10 is proteasome dependent. Furthermore, the reconstitution experiments demonstrate that the VWA domain of hRpn10 is involved in this process.

2.3 DISCUSSION

Degradation of proteins by the 26S proteasome is regulated at several levels. One of them is the docking step to the RP of the 26S proteasome which, in the case of ubiquitin system, is mediated by direct binding of poly-ubiquitin chains to the canonical proteasome subunits Rpn10 or Rpn13 or by binding to soluble ubiquitin receptors of the UBL-UBA family which in turn bind to the Rpn1 or Rpn2 subunits of the RP (Finley, 2009). Interestingly, these two binding modes seem to mutually affect each other as the degradation of poly-ubiquitylated Sic1 at the isolated 26S proteasome via Rad23 was shown to be facilitated by binding of the VWA domain of Rpn10 to the proteasome (Verma et al., 2004). Similar evidence regulating the pace of proteasomal degradation has also been observed for the FAT10 system where NUB1L was shown to accelerate FAT10 degradation about eightfold in the cells (Hipp et al., 2004). This stimulating effect relied on the N-terminal UBL domain of NUB1L required for the proteasome binding and was independent of the three UBA domains of NUB1L which bind FAT10 (Schmidtke et al., 2006). FAT10 therefore resembles ubiquitin in that it can bind directly to the proteasome as well as become tethered to the proteasome via a UBL-UBA protein, i.e. NUB1L. To better understand how NUB1L influences the rate of FAT10 degradation, we set out to identify the subunit(s) of the RP which bind FAT10. From all RP subunits, Rpn10 was the only one that interacted with FAT10 in our yeast two hybrid analysis. Interestingly, none of the subunits, known to interact with ubiquitin, namely Rpn13, or Rpt5, showed an interaction with FAT10 in this analysis. We cannot rule out from these data that other RP subunits may also bind FAT10, as false negatives cannot be excluded during yeast-two

Proteasomal targeting of FAT10 by the VWA domain of hRpn10 and NUB1L  2.

hybrid screens, but at least Rpn13 or Rpt5 subunits did not significantly bind to FAT10 in pull down assays also (Figure 2.2b and data not shown).

The finding that FAT10 and NUB1L bind the same subunit raised the important question how NUB1L may nevertheless act as a facilitator of FAT10 mediated degradation. One possibility was that FAT10 and NUB1L bind to different domains of Rpn10. It came, therefore, as a major surprise to find both of them binding to the same domain. Even more surprising was the finding that these two ubiquitin-like proteins do not bind to the UIM domains of Rpn10, like ubiquitin, but instead bind the VWA domain of Rpn10, which has never been proposed to bind ubiquitin-like domains, thus, making the VWA domain a new UBL binding domain (Figure 2.2e). It will be interesting to investigate whether VWA domain binding is also an intrinsic property of other UBL domain protein(s). The VWA domain of various hRpn10 orthologs shares about 25% identity (Supplementary Figure 2.2a). VWA domains are characterized by a β-sheet sandwiched by multiple α-helices and they usually bind metal ions via a metal ion-dependent adhesion site (MIDAS). It was hypothesized that the VWA domain in intracellular proteins mediate protein-protein interactions which might be involved in the assembly or function of multi-protein complexes (Whittaker and Hynes, 2002). We have extensively tried to identify the residues within the VWA domain that are directly involved in FAT10- and NUB1L-binding by mutagenesis of the VWA domain. However, so far all attempts including several combined point mutations were unsuccessful in abolishing the binding whereas the deletion variants, on the other hand, could not be stably expressed (N. Rani, unpublished data). It was, therefore, quite comforting that functional data in the yeast supported the in vitro binding studies and the central role of the VWA domain in FAT10 degradation.

Here, we showed that the human Rpn10 subunit was able to reconstitute growth of rpn10 S. cerevisiae cells in canavanine sensitivity assay (Figure 2.3) which, at least to our knowledge, has never been shown before. Considering that there is merely 50%

identity between yeast and human Rpn10 and the fact that human Rpn10 has two UIMs while yeast protein contains only one, a functional transcomplementation of the yeast protein by the human counterpart was not necessarily expected. Rpn10 is required to keep the base and the lid of the RP firmly associated (Fu et al., 1998) and our

Proteasomal targeting of FAT10 by the VWA domain of hRpn10 and NUB1L  2.

53

transcomplementation data with human Rpn10 suggest that the human ortholog can serve this structural function also in the context of the yeast 26S proteasome. Moreover, human Rpn10 co-migrated, at least partially, with the 26S proteasome in yeast transfectants indicating that it was readily incorporated into the yeast proteasome (Figure 2.4).

The reconstitution of Rpn10 deficient budding yeast cells with human Rpn10 encouraged us to test further if FAT10 degradation could be reconstituted in yeast. This again is not self-evident as FAT10 is only found in mammals and not in lower eukaryotes (Groettrup et al., 2008). The reconstitution of FAT10 degradation in yeast seemed justified as Rpn10 has been shown to be essential in mouse cells (Hamazaki et al., 2007) and is required for normal degradation of ubiquitin conjugates in Drosophila (Lundgren et al., 2003). We could show that FAT10 is degraded in yeast in a Rpn10 dependent manner (Figure 2.5).

The fact that FAT10 degradation was supported by both, the yeast and human Rpn10, is consistent with the finding that not the UIMs (which differ in number between human and yeast) but the VWA domain of Rpn10 is required for FAT10 degradation. This system further allowed us to show that the VWA domain sufficed to robustly reconstitute FAT10 degradation (Figure 2.5e) which is in accordance with our binding studies (Figure 2.2e).

Interestingly, the VWA domain could largely but not completely restore wild type-like growth of Rpn10 deficient yeast on canavanine containing plates (Figure 2.3).

Consistently, mice bearing only a VWA domain survive longer than mice which completely lack mRpn10 (Hamazaki et al., 2007). These results could be interpreted in a way that the facilitator function of VWA (Verma et al., 2004) promotes the degradation of misfolded poly-ubiquitylated proteins via Rad23. The finding that NUB1L and FAT10 bind to the same domain of the same RP subunit encouraged us to re-investigate by pull-down experiments whether NUB1L might bind to either Rpn1 or Rpn13 in addition to Rpn10. Indeed, we observed that NUB1L but not FAT10 binds hRpn1 (Figure 2.2b, c) reminiscent of the UBL-UBA protein Rad23 in the ubiquitin pathway.

Earlier, the only subunit known to interact with NUB1L was hRpn10 and it was shown that the C-terminus of NUB1L containing the three UBA domains binds to hRpn10 (Tanji et al., 2005). Another study reported that the association of NUB1L with the 26S proteasome would occur via the N-terminal UBL domain (Schmidtke et al., 2006). We could show that UBL domain of NUB1L binds the hRpn1 subunit as well as hRpn10 of the 26S proteasome in accordance with the latter report (Figure 2.2c). Interestingly, we

Proteasomal targeting of FAT10 by the VWA domain of hRpn10 and NUB1L  2.

could also show that the C-terminal UBL domain of FAT10 binds hRpn10 (Figure 2.2d) which seems to be reasonably valid as the N-terminal UBL domain of FAT10 docks to NUB1L (Schmidtke et.al., 2006).

Considering that Rpn1 binds to the Rpn10 subunit and the ATPases Rpt1/S7 and Rpt2/S4 within the base of the 19S regulator (Fu et al., 1998), our findings may explain why NUB1L serves as a potent accelerator of FAT10 degradation. Since Rpn1 knock out yeast are not viable (Hampton et al., 1996), deletion studies on the function of Rpn1 could not be performed. However, the key findings of this study can be exploited to design models how NUB1L may accelerate the degradation of FAT10 (Figure 2.8). FAT10 might directly bind to the VWA domain of hRpn10 and become degraded at a slow rate (Figure 2.8a) whereas NUB1L could bind weakly to the VWA domain of hRpn10 as well as strongly to the hRpn1 subunit. The second possibility could be that FAT10 gets transferred to hRpn10 subsequent to its interaction with the UBA domains of NUB1L which remains bound to the hRpn1 subunit (Figure 2.8b). Alternatively, conformational change(s) could be induced in hRpn10 when NUB1L binds either hRpn1 (Figure 2.8c) or Rpn10 (Figure 2.8d), thus functioning as a facilitator for the degradation of FAT10 and its conjugates. Testing the validity of the different models will help to further characterize FAT10 mediated degradation at the 26S proteasome.

Proteasomal targeting of FAT10 by the VWA domain of hRpn10 and NUB1L  2.

55

Figure 2.8 Models for the degradation of FAT10. (a) Slow degradation of FAT10 and its conjugates (represented by ‘S’) by 26S proteasome when its C-terminal UBL domain interacts with the VWA domain of hRpn10 in the absence of NUB1L. The UBL domain of NUB1L docks onto the VWA domain of hRpn10 weakly as compared to its binding to hRpn1. (b) Model showing the accelerated degradation of FAT10 when NUB1L transfers FAT10 to hRpn10. The UBL and UBA domain(s) of NUB1L interact with the hRpn1 subunit and the N-terminal UBL domain of FAT10, respectively. (c) Model showing the facilitator function of NUB1L for the degradation of FAT10 and its conjugates by the 26S proteasome. FAT10 docks onto hRpn10 while its degradation is facilitated in the presence of NUB1L, bound to hRpn1, in its vicinity by inducing conformational changes in both of the subunits. (d) Facilitator function of NUB1L when both FAT10 and NUB1L bind to hRpn10. The model suggests a conformational change in hRpn10 when NUB1L binds to it leading to an enhanced degradation of FAT10.

Proteasomal targeting of FAT10 by the VWA domain of hRpn10 and NUB1L