A fast and tuneable auxin‐inducible degron for depletion of target proteins in budding yeast

Abstract The auxin‐inducible degron (AID) is a useful technique to rapidly deplete proteins of interest in nonplant eukaryotes. Depletion is achieved by addition of the plant hormone auxin to the cell culture, which allows the auxin‐binding receptor, TIR1, to target the AID‐tagged protein for degradation by the proteasome. Fast depletion of the target protein requires good expression of TIR1 protein, but as we show here, high levels of TIR1 may cause uncontrolled depletion of the target protein in the absence of auxin. To enable conditional expression of TIR1 to a high level when required, we regulated the expression of TIR1 using the β‐estradiol expression system. This is a fast‐acting gene induction system that does not cause secondary effects on yeast cell metabolism. We demonstrate that combining the AID and β‐estradiol systems results in a tightly controlled and fast auxin‐induced depletion of nuclear target proteins. Moreover, we show that depletion rate can be tuned by modulating the duration of β‐estradiol preincubation. We conclude that TIR1 protein is a rate‐limiting factor for target protein depletion in yeast, and we provide new tools that allow tightly controlled, tuneable, and efficient depletion of essential proteins whereas minimising secondary effects.


| INTRODUCTION
A common approach to study the function of an essential gene in vivo is to investigate the consequence of conditionally repressing its expression.
Ideally, this approach should result in a fast and specific repression to minimise secondary and off-target effects. The auxin-inducible degron (AID; Nishimura, Fukagawa, Takisawa, Kakimoto, & Kanemaki, 2009) is a technique that can fulfil this goal as it allows fast depletion of the target protein by incubating the cell culture with a small molecule (auxin) that does not perturb cell metabolism. The physiological role of auxin is to regulate growth and development in plants, through a pathway that leads to the proteasome-mediated degradation of the Aux/IAA family of transcriptional regulators (reviewed in Teale, Paponov, & Palme, 2006). This pathway can be artificially transferred to a nonplant organism by expressing the TIR1 gene, which encodes a plant auxin-binding receptor that interacts with the conserved E3 ubiquitin ligase SCF complex. The target gene is fused with a transcriptional repressor Aux/IAA protein that functions as the degron tag (domain to induce degradation). When TIR1 protein is bound to auxin, it interacts with the degron, allowing the E3 ubiquitin ligase SCF complex to polyubiquitinate the target protein, thereby directing its degradation by the proteasome.
In the first version of the AID system for yeast, which was developed by Nishimura et al. (2009), expression of Oryza sativa TIR1 (OsTIR1) was driven by a galactose-inducible (GAL1-10) promoter or the constitutive ADH1 promoter. The galactose-inducible expression system can result in metabolic perturbations caused by the shift from glucose to galactose as carbon source (Bergkessel, Whitworth, & Guthrie, 2011;Kresnowati et al., 2006;Ronen & Botstein, 2006). Here, we tested the constitutive expression of OsTIR1 (from here on referred to asTIR1), encoded by a codon-optimized sequence from the Kanemaki lab, under the control of a strong (PADH1-701) or weak (PADH1-409) promoter (Santangelo & Tornow, 1990) and show that constitutive expression of this TIR1 is problematic. High expression can result in uncontrolled target protein depletion (even in the absence of auxin).
Conversely, a weak promoter can lead to slow depletion when TIR1 is expressed at low levels. This suggests that it is important to express TIR1 in a tightly controlled manner in order to achieve optimal results. McIsaac et al. (2013) developed a β-estradiol-inducible budding yeast expression system that makes use of an artificial transcription factor, ZnEV, made by fusing a human estrogen receptor (ER), with the viral transcription activator VP16 and three or four (where n = 3 or 4) zinc finger DNA-binding domains that recognise a specific promoter, ZnEVpr. In the absence of estrogen (β-estradiol), the Hsp90 chaperone complex inhibits ZnEV by interacting with its ER domain. Binding of βestradiol to ER releases it from the Hsp90 complex, allowing ZnEV to transcriptionally activate ZnEVpr, through direct interaction between the zinc-finger array of ZnEV and its target DNA sequence within ZnEVpr (reviewed in Pratt & Toft, 1997). McIsaac et al. (2013) report ∼100-fold increase in the level of a green-fluorescent reporter protein that is produced under control of ZnEVpr only 30 min after addition of β-estradiol. More importantly, they show that the system is specific, as incubating cells with β-estradiol does not affect the global transcription profile of the yeast cells.
Here, we present a new version of the AID system for budding yeast that we call "β-est AID," in which expression of TIR1 is tightly regulated by the β-estradiol system. To test this system, we measured initial levels and depletion rate of several target proteins in a strain of Saccharomyces cerevisiae harbouring the β-est AID and compared them with those of strains that constitutively express TIR1. We then compared depletion rates of several nuclear target proteins following preincubation with βestradiol for different times. The results show that depletion rate directly correlates with length of preincubation with β-estradiol and, therefore, with TIR1 levels. We demonstrate that even a highly abundant target protein can be quickly depleted by extending the preincubation with βestradiol. Finally, we show that the β-est AID system works when the components are genomically encoded, but for ease of use, we constructed a plasmid that allows all the elements of the β-est AID system to be introduced into budding yeast through a one-step transformation.
Yeast were grown at 30°C on Yeast Peptone Dextrose supplemented with adenine (YPDA) or yeast minimal media (YMM) supplemented with Kaiser drop outs (Formedium). When required, β-estradiol (Sigma-Aldrich) was added at a final concentration of 10 uM and indole-3-acetic acid (auxin; Acros Organics) at a final concentration of 750 uM. Samples of yeast cultures were fixed in methanol, chilled in dry ice at a ratio of 3:2 (culture:methanol by volume).

| Western blots
Protein extracts were prepared following a NaOH/TCA precipitation Normalisation was by amount of protein loaded. Pgk1 bands were also quantified for comparison.

| High-level expression of TIR1 causes auxin-independent depletion of target protein
First, we compared strains PADH1-701-TIR1 and PADH1-409-TIR1 that constitutively express TIR1 to high or low levels, respectively ( Figure 1a,b). To this end, we C-terminally tagged splicing factor Prp22 with a fusion between a truncated auxin-dependent degron, named AID* (Morawska & Ulrich, 2013) and six tandem repeats of the FLAG epitope, in strains PADH1-701-TIR1 and PADH1-409-TIR1. To measure the rate of target protein depletion, we added auxin  where our protein targets are located, we fused an SV40 nuclear localisation signal (NLS) to the TIR1-coding sequence. The resulting strain, PZ4EV-NTIR1, produces TIR1-V5 protein rapidly after addition of β-estradiol to the culture medium (Figure 1b), and auxin was added after preinduction of TIR1 for 30 min. Under these conditions, Prp22 was depleted rapidly after auxin addition, without detectable auxinindependent depletion (Figure 1b,c).
Next, to test the hypothesis that levels of TIR1 inversely correlate with levels of the target protein in the absence of auxin, a culture of PZ4EV-NTIR1 with AID*-6FLAG-tagged PRP22 was incubated with β-estradiol but without auxin, and the levels of Prp22 were measured over time. Consistent with our previous observation, at 50 min of βestradiol incubation, the level of Prp22 had dropped significantly and reached 35% of the initial value at 2 hr of incubation, by which time the TIR1-V5 protein was well induced (Figure 2). Auxin-independent depletion of Yhc1 and Rrp44 is shown in Figure S2 and was less pronounced with the more abundant Rrp44. We conclude that high levels of TIR1 can cause auxin-independent depletion of the target protein in budding yeast.
3.2 | Depletion rate can be tuned by modulating the duration of β-estradiol preincubation Next, we investigated how the rate of auxin-induced depletion is influenced by the length of β-estradiol preincubation. On the basis of the previous results, we anticipated that not only longer preincubation times would lead to faster depletion but also more auxin-independent depletion and that there may be an optimal preincubation time, which is likely to be target specific. To test this, we used as targets for depletion, Prp22 (232 copies/cell), Prp2, another essential splicing factor with similar abundance (211 copies/cell), and the more highly expressed decapping enzyme, Dcp1 (4,189 copies/cell; Kulak, Pichler, Paron, Nagaraj, & Mann, 2014). We performed a time-course depletion analysis in which cultures of these AID*-tagged strains were preincubated with β-estradiol for different times (20, 30, 40, or 60 min) prior to auxin addition. Samples were then taken for protein analysis at 5-min intervals. As the levels of TIR1 increase with time of βestradiol incubation, this allowed us to measure the relationship between TIR1 abundance and auxin-dependant depletion rate of different target proteins.
The protein quantification analysis (Figure 3), shows that different target proteins were depleted at different rates. A 20-min preincubation with β-estradiol was sufficient to reduce Prp22 and Prp2 to low levels (≤ 20%) within 15 min of auxin addition, but longer preincubations with β-estradiol resulted in auxin-independent degradation. In contrast, the more abundant Dcp1 required 60-min of βestradiol preincubation to achieve a similarly rapid and efficient depletion because more Dcp1 protein has to be degraded to achieve efficient depletion in terms of percentage of the starting amount, and this evidently requires more TIR1 protein. Notably, with all three target proteins, there was a direct correlation between the duration of β-estradiol preincubation and the depletion rate so that the duration of β-estradiol treatment should be optimised for each target protein (see also Figure S2).

| Plasmid-encoded β-est AID and the effect of TIR1 protein localization on efficiency of targeted depletion
To facilitate insertion of the β-est AID components into budding yeast, we constructed a centromeric plasmid, pZTRL, which contains PACT1-Z4EV and Z4EVp-OsTIR1 (without NLS) expression cassettes (Figure 4 a). We then tested pZTRL and, at the same time, investigated the effect of fusing an NLS to TIR1 on the efficiency of depleting a nuclear protein. To this end, we measured both TIR1 and Prp22 target protein levels in the pZTRL-bearing strain, and in two strains that contain genomically integrated TIR1, with (PZ4EV-NTIR1) or without (PZ4EV-TIR1) NLS on the N-terminal of TIR1 (Figure 4b,c).
We observed that the β-estradiol-induced level of TIR1 protein was about threefold higher in the genomic TIR1 (PZ4EV-TIR1) strain compared with genomic NLS-TIR1 (PZ4EV-NTIR1) or plasmid-encoded TIR1 strains, indicating that TIR1 is better expressed when it is genomically-integrated and lacks an NLS. Interestingly, Prp22, which is a nuclear localized protein, was quickly depleted irrespective of the suggests that, for depletion of nuclear proteins, an SV40 NLS need not be added to TIR1. Notably, targeted depletion was slower in the plasmid-encoded TIR1 strain compared with the other two strains, reaching 15% compared with 2% of Prp22 initial values 30 min after auxin addition, probably due to a lower expression of TIR1 in this strain.

| DISCUSSION
Our observation that in budding yeast high levels of TIR1 protein cause auxin-independent depletion agrees with reports that overexpression of TIR1 in Arabidopsis thaliana leads to an auxin-response phenotype (Gray et al., 1999)  for human cells and showed that incubation with the tetracycline analogue doxycycline slows growth of a DHC1-AID tagged cell culture, implying that in human cells, over-expression of TIR1 may cause leaky depletion of the target (Natsume et al., 2016). In our work, we were repeatedly unsuccessful with 10 out of 20 essential genes that we tried to AID-tag in PADH1-701-TIR1 (high-level TIR1 expression), whereas eight of these were successfully tagged in either PADH1-409-TIR1 (low-TIR1 expression) or in pZTRL. Indeed, all 22 essential genes for which AID tagging was attempted in PADH1-409-TIR1 (low-TIR1 expression), and all nine attempts to AID-tag essential genes in pZTRL were successful (data not shown), with success defined as failure to grow in the presence of auxin. This, together with our measurements of Prp22 and TIR1 levels in these strains, strongly suggests that high-level expression of TIR1 causes auxin-independent degradation that may result in target proteins falling below the level required for viability. The auxin independent activity of SCF-TIR1 E3 ubiquitin ligase complex that we and others have observed could be caused by low-affinity interaction of TIR1 with its target in the absence of auxin (Dharmasiri, Dharmasiri, & Estelle, 2005;Kepinski & Leyser, 2005;Tan et al., 2007); by traces of auxin in the media, as previously speculated (Natsume & Kanemaki, 2017); or by low levels of endogenous auxin in plants and yeast (Rao, Hunter, Kashpur, & Normanly, 2010).
Notably, the plasmid-encoded, β-est AID (pZTRL) depleted the target protein more slowly than genomically integrated TIR1 (Figure 4), likely because the plasmid-encoded TIR1 protein was FIGURE 3 Depletion rate can be tuned by modulating the duration of β-estradiol pre-incubation. Western blot of AID-targets (a-b) Prp22-AID*-6FLAG, (c-d) Prp2-AID*-6FLAG, and (e-f) Dcp1-AID*-6HA, from cultures preincubated with β-estradiol (β-est) for 20, 30, 40, or 60 min prior to auxin addition. Equal amounts of total protein were loaded in each lane, and Pgk1 is included as a visual loading control, except for panel E, where Pgk1 and Dcp1 comigrate. Quantifications of protein bands in panels A, C, and E are shown in panels B, D, and F. As a measure of depletion rate, the slope (m) was calculated for the linear section (from 100 to 30% of initial values) of each curve. Error bars represent sd of two biological replicates. For each experiment, only one representative blot is shown produced at a lower level. This could be due to transcriptional interference between PACT1-Z4EV and Z4EVpr-OsTIR1 that are convergently transcribed in pZTRL, for example, if their shared transcriptional terminator TTEF1 does not act bidirectionally-even though it was proposed that most transcriptional terminators in S. cerevisiae function bidirectionally (Uwimana, Collin, Jeronimo, Haibe-Kains, & Robert, 2017). Thus, the plasmid-based β-est AID may require longer βestradiol preincubation times and may not be ideal for protein targets that are more abundant. In addition to pZTRL, we created pZTRK by replacing LEU2 in pZTRL with KanMX. pZTRK can be propagated in rich media, which is useful given that high concentrations of auxin can impair growth in minimal media (T.S. Turowski, S. Bresson and D. Tollervey, personal communication).
We targeted only nuclear proteins for depletion, which is why we tested the effect of an NLS sequence at the start of the TIR1 protein.
However, as previously reported for budding yeast (Tanaka, Miyazawa-Onami, Lida, & Araki, 2015), the presence of the NLS had little effect on depletion rate of a nuclear protein (Prp22). We speculate that TIR1 protein when N-terminally fused to an NLS may have reduced stability, but that this may result in similar amounts of nuclear localised TIR1 protein in the presence or absence of an NLS, such that the target protein depletion rates are similar. Nishimura et al. (2009) showed that the AID system allows the rapid and efficient depletion of proteins present either in the nucleus or the cytoplasm. Therefore, we anticipate that our non-NLS TIR1 would also support efficient depletion of cytoplasmic targets although we have not tested this.
In summary, we present evidence that the level of expression of TIR1 has a profound influence on how rapidly and completely the AID system depletes its target protein. In order to enhance this system for the molecular biology community, we developed a suite of strains and plasmids in S. cerevisiae. The strain PADH1-409-TIR1 allows a slower, more linear depletion of the target protein (Figure 1c), which is useful where full depletion is undesirable, and may facilitate kinetic studies of the effects of protein depletion. The β-estradiol-induced TIR1 strains can have their preincubation time optimised to produce an extremely rapid and effective degradation of the target protein whereas minimising auxin-independent degradation. Although this tunable βestradiol expression system has been developed specifically for use in budding yeast, the data and principles we present are likely to apply to other organisms and to be useful for the wider scientific community interested in making the most out of the powerful AID technique.
for the gift of rabbit anti-OsTIR1 antibody and for helpful discussion.