Real-time detection of PRT1-mediated ubiquitination via fluorescently labeled substrate probes

The N-end rule pathway has emerged as a major system for regulating protein functions by controlling their turn-over in medical, animal and plant sciences as well as agriculture. Although novel functions and enzymes of the pathway were discovered, ubiquitination mechanism and substrate specificity of N-end rule pathway E3 Ubiquitin ligases remained elusive. Taking the first discovered bona fide plant N-end rule E3 ligase PROTEOLYSIS1 (PRT1) as a model, we use a novel tool to molecularly characterize polyubiquitination live, in real-time. We gained mechanistic insights in PRT1 substrate preference and activation by monitoring live ubiquitination by using a fluorescent chemical probe coupled to artificial substrate reporters. Ubiquitination was measured by rapid in-gel fluorescence scanning as well as in real time by fluorescence polarization. Enzymatic activity, substrate specificity, mechanisms and reaction optimization of PRT1-mediated ubiquitination were investigated ad hoc in short time and with significantly reduced reagent consumption. We demonstrated for the first time that PRT1 is indeed an E3 ligase, which was hypothesized for over two decades. These results demonstrate that PRT1 has the potential to be involved in polyubiquitination of various substrates and therefore pave the way to understanding recently discovered phenotypes of prt1 mutants.

. In plants, only two of these, namely PRT1 and PRT6, 114 are associated with the N-end rule and assumed to function as N-recognins (Fig. 1b). This 115 is in contrast to the high number of proteolytically processed proteins which carry in their 116 mature form N-terminal amino acids that could potentially enter the enzymatic N-end rule 117 pathway cascade (Venne et al., 2015). In the light of more than 800 putative proteases in 118 the model plant Arabidopsis thaliana, it is likely that the N-end rule pathway plays an im-119 portant role for protein half-lives in a proteome-wide manner. Examples are found in the 120 METACASPASE9 degradome, i.e. that part of the proteome which is associated with degra-

123
PRT1, compared to the Saccharomyces cerevisiae N-recognin Ubr1 (225 kDa), is a 124 relatively small protein (46 kDa) and totally unrelated to any known eukaryotic N-125 recognins but with functional similarities to prokaryotic homologs (Fig. 1b)    Here, we report a novel advanced approach to molecularly characterize E3 ligases,

271
The reaction mixture was filtered, the solvent was removed under reduced pressure and 272 the remaining residue was dissolved in distilled water (300 mL).  strate protein moiety (Fig. 2a, Supporting Information Fig.S 1a). Cleavage by TEV gave here for the first time E3 ligase activity of PRT1 depending on E1, E2 and ATP (Fig. 2c). proach. We identified distinct subspecies via in-gel detection (Fig. 2c). A classical end-time 422 point assay where the reaction was stopped at different reaction time points followed by SDS-PAGE and in-gel fluorescence detection revealed the kinetics of PRT1 activity using F-424 eK-Flv as substrate (Fig. 2d). 425 However, a real-time monitoring of the kinetic profile of the enzymatic reaction is 426 only possible via FP in live detection measurements. The kinetic profile is best-fitted with 427 an S-shaped curve and a growth curve model of logistic type (Richards' equation) rather 428 than exponentially as expected for simple kinetics (Fig. 2e). 429 It was previously suggested that PRT1 binds to N-degrons carrying bulky destabiliz-  (Fig.2f, Supporting Information Fig. S1b,c). While the substrates carrying G-, R-,

435
While the eK-based substrate showed the kinetic curve discussed above, the control F-eΔK-

436
Flv substrate with mutated lysines (expected site of ubiquitination, Lys15 and Lys17, both 437 replaced by Arg) presented a faster initial rate of ubiquitination but levels of only half of 438 the final FP value (Fig. 2f). This is in good agreement with the in-gel fluorescence detection 439 where lower degrees of ubiquitination of F-eΔK-Flv, reduced mono-and di-ubiquitination -440 but still clear polyubiquitination -were observed (Supporting Information Fig. S1c). 441 Another remarkable observation of the ubiquitination pattern in the in-gel fluores-442 cence image (using three different independent substrate protein purifications of F-eK-Flv-443 NBD) was that the tri-ubiquitinated form presents three distinct subspecies which eventu-444 ally lead to a multitude of other species at higher level (Supporting Information Fig. S1b). 445 There was only one species of tri-ubiquitinated F-eΔK-Flv-NBD generated, where two optimum to be clearly above pH 7 but below pH 9 as indicated by the occurrence of 457 polyubiquitinated species of the fluorescent substrate probe F-eK-Flv-NBD (Fig. 3a). How-458 ever, using our real-time FP protocol, we additionally acquired the kinetic profile of the 459 PRT1-mediated ubiquitination process (Fig. 3b) and the maximum reached polarization 460 values of this reaction (Fig. 3c). These correlated with the amount of polyubiqutinated spe-461 cies detected in the SDS-PAGE gel-based end-time experiment (Fig. 3a) and the highest 462 initial rate (Fig. 3c) whereas the latter appears to be different from the reaction optimum 463 according to the detected max. FP. We also had previously observed, that F-eΔK-Flv ubiqui-464 tination presented a faster initial rate but only half of the final FP (Fig. 2f) and lower de-465 grees of final ubiquitination (Supporting Information Fig. S1c). Both bell-shaped forms of 466 the pH dependence for the highest initial reaction rate (pH 8.0) and the maximum sub-467 strate polyubiquitination rate (pH 7.5) indicated two competing processes that generate a 468 local maximum (Fig. 3c). 469 A strong decrease of the ubiquitination rate mediated by PRT1 was observed at 470 higher concentrations of the E2-conjugating enzyme UBC8 (>2 µM) both via in-gel fluores-471 cence (Fig. 3d) and FP ( Fig. 3e-g) (Fig. 3e). Moreover, the distri-475 bution pattern of the ubiquitinated substrate species at the end of the reaction (Fig. 3f) and 476 the kinetic profiles of ubiquitination (Fig. 3g) are different, depending on the used E2 con-  (Fig. 2f, Supporting Information Fig. S1b,c). Third, our test system allowed to de-499 scribe E3 ligase function and target specificity by using variants of labeled substrates. Using FP coupled to immunoblot analysis, we were able to confirm that PRT1 is an 522 active E3 ligase acting in concert with E2-conjugating enzyme UBC8. In a buffer system  (Fig. 2e).

545
The FP-based assay revealed that the kinetic profile of the ubiquitination was de-546 pendent on the position and availability of lysines as Ub acceptor sites as suggested to be 547 characteristic of N-degrons (Bachmair & Varshavsky, 1989). By lowering the overall num-548 ber of available lysines in the F-eΔK-Flv-NBD substrate (two Lys less than in X-eK-Flv con-549 structs with 11 Lys in total) the overall ubiquitination was detectably reduced. Differences  (Fig. 2f, Supporting Information Fig. S1c). However, the simple and Lys17) were replaced by Arg (Supporting Information Fig. S1b).
When analyzing the influence of the pH on PRT1 function as E3 Ub ligase, we docu-577 mented bell-shaped forms of the pH dependence for the highest initial reaction rate (pH 578 8.0) and determined the maximum substrate polyubiquitination rate (pH 7.5). These indi-579 cated two competing processes that generate a local maximum (Fig. 3c). In the light of re-580 cently discussed mechanisms of E3 ligase action (Berndsen & Wolberger, 2014)  6.8 to pH 8 range (Fig. 3c). The competing processes leading to the decrease in ubiquitina-589 tion at pH>8 could be destabilization of ionic and hydrogen bonds at alkaline pH simply 590 interfering with protein-protein interaction or ATP hydrolysis affecting the Ub charging of 591 the E2 by the E1. This could also explain the premature leveling of the kinetic curves in the 592 FP measurements at pH>8 (Fig. 3b) while in a longer reaction timescale, the maximum FP 593 values would be expected to be the same from pH 6.8 to pH 7.5.

594
The apparent catalytic rate constant (kcat) of the Ub transfer, more precisely the 595 transfer of the first Ub molecule, i.e. the rate limiting step, was found to be 1.30±0.07 s -1 .

596
This suggested that on the one hand PRT1 had a high turnover number due to a highly ac- favored. This is especially suggested by the variable occurrence of the distinct pattern of 604 triubiquitinated substrate species (Fig. 3d,f) as mentioned above and discussed in other 605 systems as well (Ye & Rape, 2009).

606
By using fluorescently labeled substrate proteins in the two described approaches, 608 that is, gel-based fluorescence scanning after SDS-PAGE and FP, we were able to investigate 609 the mechanism of PRT1-mediated ubiquitination and optimize the reaction conditions. The level. Therefore, we see potential for a broader impact for ubiquitination research as it is conceivable that the method is transferable to other E3 ligases and enzyme-substrate pairs.

639
In the course of our studies, we felt that rapid, easy and cheap protocols were lacking for 640 in-depth biochemical analysis of E3 ligase kinetics, the same holds true for non-radioactive 641 and sterically not interfering protocols and those where entire proteins and directly la-642 beled substrates can be applied.

643
In terms of further applications, the kinetic approach allowed collecting data that 644 can assist to set up high-throughput assays, e.g. for screens of inhibitors and the influence