Switching toxic protein function in life cells

Toxic proteins are prime targets for molecular farming and efficient tools for targeted cell ablation in genetics, developmental biology, and biotechnology. Achieving conditional activity of cytotoxins and their maintenance in form of stably transformed transgenes is challenging. We demonstrate here a switchable version of the highly cytotoxic bacterial ribonuclease barnase by using efficient temperature-dependent control of protein accumulation in living multicellular organisms. By tuning the levels of the protein, we were able to control the fate of a plant organ in vivo. The on-demand-formation of specialized epidermal cells (trichomes) through manipulating stabilization versus destabilization of barnase is a proof-of-concept for a robust and powerful tool for conditional switchable cell arrest. We present this tool both as a potential novel strategy for the manufacture and accumulation of cytotoxic proteins and toxic high-value products in plants or for conditional genetic cell ablation.


INTRODUCTION 39
The low-temperature degron cassette (lt-degron) fused to a protein of interest (POI) 40 can efficiently control its accumulation or degradation in a temperature-dependent 41 manner. The technique is based on altering the in vivo stability of the lt-degron:POI 42 fusion protein by using a temperature stimulus and relies on the proteostatic machin-43 ery of the N-end rule pathway of protein degradation (Dissmeyer et al., 2018) for fast 44 removal from the cell. The lt-degron:POI fusion can thus be efficiently accumulated at 45 a permissive temperature or removed from the cell at a restrictive temperature gen-46 erating an artificial temperature-sensitive variant of the POI (Faden et al., 2016;47 Dissmeyer, 2017). Opposed to many other systems for conditional protein expression 48 and gene regulation, the lt-degron approach relies on posttranslational oinzerference 49 by protein degradation of the entire fusion protein and is applicable in living multicel-50 lular organisms, in both plants and animals. By this, it offers tight control over the 51 levels of active/functional protein which directly correlates to the temperature stimu-52 lus used to induce the system. 53 54 Molecular farming, the generation of pharmacologically active or biotechnologically 55 usable compounds in plants, is an emerging topic especially for the production of 56 peptides and proteins requiring special glycosylation patterns impossible to generate 57 in classical fermenter-based expression systems using microorganisms like bacteria 58 (Stoger et al., 2014). Various cytotoxic peptides are under investigation as highly po-59 tent cancer therapeutics (Boohaker et al., 2012). Some toxic proteins, such as the 60 mistletoe lectins possess anti-cancer properties when administered orally (Pryme et 61 al., 2007) and recombinantly produced immunotoxins are in clinical and pre-clinical 62 evaluation for cancer treatments (Kelly et al., 2012;Madhumathi and Verma, 2012).  The lt-degron K2 starts with a 3' tobacco Omega (Ω) leader sequence for translation 140 enhancement from pRTUB8, a derivative of pRTUB1 (Bachmair et al., 1990;141 Bachmair et al., 1993). The leader contains the 20 nucleotides upstream of the start 142 7 / 27 codon of tobacco mosaic virus strain U1 (Gallie et al., 1987) and is followed by a co-143 don-optimized synthetic human Ubiquitin gene (Ecker et al., 1987), a triplet for the 144 destabilizing bulky and hydrophobic amino acid Phe (F), a short linker peptide (trans-145 lates to HGSGI) (Bachmair and Varshavsky, 1989) to further expose the N-terminal 146 residue, a temperature-sensitive mouse dihydrofolate reductase sequences (DHFR ts ) 147 triggering the protein unfolding response, a triple hemagglutinin (HA)-tag (HAT) for 148 immunodetection. Phe is used as destabilizing N-terminal residue, because it was 149 shown to be recognized by the Arabidopsis N-end rule pathway (Bachmair et al., 150 1993;Potuschak et al., 1998;Stary et al., 2003;Faden et al., 2016;Mot et al., 2018  (w/w) clay, 15% (w/w) block peat, 20% (w/w) coco fibers; 10-00800-40, Einheitser-178 dewerke Patzer); 25% (w/w) Vermiculite (grain size 2-3 mm; 29.060220, Gärt-179 nereibedarf Kamlott); 300-400 g of Exemptor (100 g/kg thiacloprid, 802288, Hermann 180 Meyer)) per m³ soil mixture or aseptically in vitro on plastic Petri dishes containing ½ 181 Murashige & Skoog (MS) (2.16 g/l MS salts, Duchefa Biochemicals, M0221; 0.5% 182 Glu sucrose; 8 g/l phytoagar, Duchefa Biochemicals, P10031; pH to 5.6 -5.8 using 183 KOH For aseptic culture, dry seeds were sterilized with chlorine dioxide gas produced from 198 extend where the barnase protein is unable to elicit a phenotype (Figure 1b). 327 328

Transient expression of lt-barnase in tobacco leaves 329
The lt-degron, as a conditional protein 'shut-off' technique, relies on a temperature-330 sensitive switch between POI depletion and accumulation on protein level without 331 altering transcript levels (Faden et al., 2016;Dissmeyer, 2017). Therefore, we tested 332 conditional accumulation of lt-BAR upon shift of ProTRY::lt-BAR expressing Ara-333 bidopsis plants to permissive temperature. Although the cell-ablation phenotype was 334 visible (Figure 1a,b), detection of lt-BAR protein with standard western blotting tech-335 niques gave only variable and unsatisfactory results. Because also detection of lt-336 BAR transcript of ProTRY::lt-BAR by semi-quantitative RT-PCR proved to be difficult, 337 which is possibly due to very low levels of transcript as an effect of the additional 338 RNAse activity from barnase, we transiently expressed the lt-BAR construct under 339 the control of the UBIQUITIN10 promoter (ProUBQ10) in Nicotiana benthamiana (to-340 bacco) leaves. Surprisingly, the expression led to an almost temperature-341 independent phenotype with developing necrosis and chlorosis of the leaves at both 342 permissive and restrictive temperature. The phenotype even appeared to be stronger 343 at the restrictive temperature. This might be explained through an override of the 344 degradation system in regard to lt-BAR-dependent degradation, most likely being a 345 result of expression from the strong ProUBQ10 combined with a possibly higher cata-346 lytic activity of the lt-BAR protein at the higher temperature (Figure 2a). We tried 347 once more to demonstrate the connection between the phenotype and the abun-348 dance of lt-BAR fusion protein by western blotting and semi-quantitative RT-PCR and 349 could obtain transcript data clearly linking the observed phenotype to the expression 350 of lt-BAR (Figure 2b). 351 352

Conditional toxin function can be shifted and is active at ambient temperature 353
We then elucidated the sensitivity of the lt-BAR module. For this, plants were grown 354 at permissive temperature, shifted after three weeks to restrictive temperatures for 355 nine days, and finally shifted back to permissive temperature. During this process, 356 trichomes on newly forming leaves were monitored. The trichome phenotype of shift-357 ed plants followed the temperature stimulus, exhibiting glabrous leaves at the per-358 missive temperature, developing wild type-like leaves containing trichomes when 359 grown under restrictive temperature, and again showing glabrous leaves when shift-360 ed back to the permissive temperature (Figure 3a). This shows that also the function 361 of lt-BAR follows the temperature stimulus exhibiting efficient ON/OFF transitions as 362 previously described for the lt-degron for non-toxic proteins (Faden et al., 2016;363 Dissmeyer, 2017). Next, we asked whether lt-BAR expressing plants could success-364 fully elicit their glabrous leaf phenotype when grown at standard greenhouse condi-365 tions. Stability of lt-BAR at ambient temperature, thus the possibility to evoke a phe-366 notype, would greatly facilitate its application as only one controlled growth environ-367 ment, used for restrictive conditions, would be necessary in order to be applied suc-368 cessfully. Indeed, lt-BAR was able to robustly result in glabrous leaves when plants 369 were grown in the greenhouse, suggesting sufficient stability of the fusion protein 370 even at a temperature that is higher than the standard permissive temperature of 371 14°C (Figure 3b). Remarkably, the lt-BAR expressing plants maintain the glabrous 372 leaf phenotype under greenhouse conditions even though the greenhouse represents 373 a significantly less controlled environment with higher temperature fluctuations and 374 initially higher then permissive temperatures then the previously used growth cabi-375 nets. 376 377 Conditional toxin causes growth arrest rather than cell death 378 Then, we further elucidated the effect of lt-BAR on trichome development. Trichome 379 spacing and therefore TRY activity begins in the leaf primordia (Larkin et al., 1996). 380 Since TRY is not important for formation of the trichome itself but crucial for its spac-381 ing and patterning by suppressing trichome initiation in neighboring cells, ablation or 382 arrest of a TRY-expressing cell could allow its expression in the neighboring cell po-383 tentially resulting in its own ablation and starting a chain reaction. To address this 384 hypothesis, we prepared agarose imprints of the leaf surface of plants grown at the 385 permissive temperature. Analysis revealed that the trichome forming cells went into 386 an early arrest and were not able to form the typical trichome structure (Figure 3c). 387 This could be explained through two different mechanisms. Either the TRY promoter 388 is only active for a limited period which allows to express only a level of lt-BAR that is 389 sufficient to arrest cell growth but which is not sufficient to kill the cell. Or an equilibri-390 um is reached where synthesis of active lt-BAR and possible degradation due to a 391 slight leakiness of the degron is reached. Such degradation at the permissive tem-392 perature has been demonstrated previously for a lt-degron::TTG1 protein fusion 393 (Faden et al., 2016). Also, growth arrest of a single cell, including its functional dis-394 ruption, does not seem to trigger trichome formation in neighboring cells indicating 395 that trichome spacing is a restricted process only happening for a defined period dur-396 ing leaf formation. To test whether the cells underwent cell death or resided in a state 397 of growth arrest, nuclei were fluorescently stained with DAPI. All cells showed normal 398 nuclei indicating indeed cell arrest rather than cell death (Figure 3d). 399

400
In very rare cases, cells were able to partially overcome the toxic effect of lt-BAR. 401 This was indicated by started initiation and formation of a trichome in these cells 402 (Figure 4a). This observation indicates that indeed the TRY promoter activity was 403 reduced and that, potentially, cells are able to degrade the lt-BAR over an extended 404 period of time even at the permissive temperature. However, it is noteworthy that ma-405 ture trichomes where never found on any lt-BAR expressing plant at the permissive 406 temperature. Surprisingly, the phenotype evoked by lt-BAR strikingly resembled a 407 mutant of the transcription factor GLABRA2, the glabra2 (gl2) mutant (Rerie et al., 408 1994;Schwab et al., 2000). The ProTRY::lt-BAR-expressing line shows similarly ar-409 rested cells that may grow out into basic trichome precursors (Figures 3c,d and 4a) 410 as described for the gl2 mutant. GL2 is thought to be mainly important for trichome 411 formation, however, also playing a role in trichome patterning was discussed (Rerie 412 et al., 1994;Hulskamp, 2004).

N-end rule E3 ubiquitin ligase PRT1 is required for modulating lt-BAR toxicity 421
We recently showed that in Arabidopsis, lt-degron fusions, for technical reasons, are 422 mainly degraded by the N-end rule E3 ligase PROTEOLYSIS1 (PRT1) (Faden et al.,423 2016) which has a high preference for aromatic N-terminal amino acids (Potuschak 424 et al., 1998;Stary et al., 2003;Naumann et al., 2016;Dong et al., 2017;Reichman 425 and Dissmeyer, 2017;Mot et al., 2018). To generate a stable genetic situation with 426 constitutively stabilized lt-BAR and to test whether the observed phenotype was a 427 result of the fusion to the lt-degron cassette or rather a response of barnase activity 428 to the changed temperature, we introgressed ProTRY::lt-BAR into the prt1 mutant. 429 Indeed, crossing the lt-BAR into the prt1 mutant background resulted in a complete 430 and temperature-independent stabilization of the phenotype evoked through lt-BAR 431 showing that presence of PRT1 and therefore altered activity of the N-end rule path-432 way of targeted protein degradation are responsible for the temperature-responsive 433 phenotype (Figure 4c). 434 435 436

DISCUSSION 437
Conclusively, we have shown that lt-BAR is able to efficiently mediate cell arrest and 438 thereby influence organ fate in Arabidopsis. The lt-BAR is superior to many systems 439 used due to a few of its characteristics. First of all, the control over the lt-BAR protein 440 is easy to execute. Since it follows a temperature stimulus, stabilization as well as 441 degradation and tuning of active protein are easy to carry out. Additionally, exoge-442 nous addition of compounds and/or other stimuli for induction of the system are not 443 needed hence eliminating issued connected to infiltration, uneven induction of the 444 system due to uneven perfusion of inducing agents as well as possible toxic side ef-445 fects on the system by used compounds and solvents. Additionally, the temperature 446 stimulus is easily controllable enabling for easy tuning and regulation of the system 447 therefore rendering maintenance of the lt-BAR transgene simple and straightforward. 448 Summing up, here we present a highly versatile and easy system for conditional or-  Analysis of lt-BAR transcript and protein levels demonstrates that it is expressed but protein amounts are below the detection limit. An lt-control protein expressed from the same promoter served as a positive control during western blot analysis. Control plants were transformed with Agrobacteria carrying the p19 plasmid only.