Auxin-inducible degron system: an efficient protein degradation tool to study protein function

The coding and noncoding components of the genome shape the physiology and morphology of an organism. Abrogation of functions of these components is an effective way of studying their contribution to cellular integrity. The coding component of the genome encodes proteins. Classically, disrupting protein functions has been done by altering either the genomic DNA sequence or the RNA transcripts of the targets of interest. However, these techniques are not without their drawbacks.

Targeted gene knockouts have been instrumental in studying gene function. However, gene knockouts resulting in early lethality of the organism or cells make investigating the function of essential genes challenging. For example, knocking out CTCF to study its functions is impossible as it is essential for early embryonic development [1,2]. It is tricky to use gene knockouts to dissect the functions of proteins that change with developmental stages, or with different phases of the cell cycle, because of the irreversibility of the knockout and less temporal control of the process. Engineering gene knockouts via the generation of premature termination codons might lead to genetic compensation by nonsense-mediated decay, resulting in erroneous conclusions like the gene being nonfunctional [3,4]. Moreover, CRISPR-Cas9-based gene knockouts have been shown to result in complete or partially functional truncated proteins, which are hard to detect, again leading to false interpretations [5,6].

RNAi using si/shRNAs is an extensively employed technique to compromise gene function at the transcript level [7,8]. Processed siRNAs/shRNAs complementary to the target mRNA interact with Argonaute protein to form a RNA-induced silencing complex (RISC). Binding of RISC to the target mRNA results in transcript degradation. The phenotype of the target RNA degradation may not be immediately evident because proteins synthesized prior to degradation may be stable and functional. Also, not all shRNAs/siRNAs targeting an mRNA are equally efficient in the mRNA knockdown, a factor that dictates the severity of the phenotype [9]. The time taken for substantial mRNA degradation by RNAi may also affect the interpretation of results. For example, the role of proteins in anterograde and retrograde transport in the Golgi complex was difficult to decipher due to slow protein depletions with RNAi [10]. Slow depletion of target proteins provides cells ample time to adapt and either activate compensatory signaling mechanisms or accumulate secondary effects [11]. Further, proteins with longer half-lives, such as members of the nuclear pore complex, require long and continuous periods of RNA knockdown compared with most short-lived proteins [12,13].

Since most gene knockouts and RNA knockdown studies aim to disrupt the protein function, it is desirable to target the proteins directly. An early demonstration of direct protein degradation being advantageous over the conventional methods involved a DNA replication protein, MCM4. When MCM4 transcription was inhibited in yeast, protracted protein degradation resulted in cells having 2C DNA content, indicating defects in the G2 cell cycle phase or mitosis. On the other hand, protein degradation mediated by heat-inducible degron culminated in replication defects even before the G2 phase [14]. In other studies, inducible protein degradation allowed for reversible and controlled removal of proteins at desired specific stages, which helped in unraveling the spatial and temporal functions of target proteins [15–18].

Targeted protein degradation strategies broadly recruit lysosomes or proteasomes for protein degradation. Trim-Away uses antitarget antibody microinjections and recruits the proteasomal pathway to degrade endogenous proteins in both dividing and nondividing cells [19]. Proteolysis-targeting chimeras (PROTACs) and molecular glues are engineered chemical molecules that specifically bind target proteins and recruit the ubiquitin-proteasome system for degradation [20,21]. Only intracellular and membrane proteins can be targeted by PROTACs and molecular glues. Extracellular and membrane-bound proteins, which comprise 40% of the proteome, recruit the lysosomal degradation pathway, like in techniques using lysosome-targeting chimeras (LYTACs) [22] and antibody-based PROTACs (AbTACs) [23]. The systems mentioned cannot degrade large protein aggregates and defective cell organelles, which requires molecules that recruit the macroautophagy-lysosomal pathway, like autophagy-targeting chimeras (AUTACs) [24] and autophagosome-tethering compounds (ATTECs) [25] among others. These strategies can only degrade a subset of proteins known to bind small molecules. A comprehensive understanding of these targeted protein-degradation strategies can be found in reviews [26–29].

Degrons are signal sequences that target the cognate protein for proteasomal degradation. These include the constitutive degron, heat-inducible degron [30], light-sensitive degron [31], destabilizing domains [32], degradation tag (dTAG) system [33], HaloTag system via hydrophobic tagging or HaloPROTACs [34,35], small molecule-assisted shutoff (SMASh) [36] and auxin-inducible degron (AID) system [15]. When tagged to proteins, heat-inducible degrons degrade them upon shifting to higher temperatures [30]. This degron has been largely utilized in yeast but not in temperature-sensitive cells of invertebrates and mammalian models. Low-temperature degrons are now available for plants and Drosophila model systems [37]. Light-inducible degrons use blue light illumination as the inducer [31]. Destabilizing domains are inherently unstable and degrade the tagged protein. However, these domains require small molecules for in vivo stabilization [32]. Most of these techniques do not allow for tissue or cell-type-specific protein degradation.

AIDs use auxin, the phytohormone, as the inducible factor for the degradation of AID-tagged proteins. The AID system has seen extensive applications in various model systems and biological branches [17,38–41]. This review discusses AID technology, its applications, shortcomings and advances. Finally, the AID system is compared with other powerful tag-based degradation technologies.

AID technology: development & customization

In plants, auxin is a phytohormone that regulates gene expression. ARFs are transcription factors that regulate auxin-inducible genes. In the absence of auxin, ARFs are bound by AUX/indole-3-acetic acid (IAA) repressor proteins and prevent ARFs from regulating the transcription of auxin-inducible genes. In the presence of auxin, the AUX/IAA repressors are degraded by SCF^TIR1 E3 ubiquitin ligases, setting ARFs free to regulate the transcription of auxin-inducible genes (Figure 1). Skp1-Cullin/CDC53-F-box (SCF) E3 ubiquitin ligases are common E3 ligases that regulate multiple processes across yeasts to mammals. The substrate specificity of different SCF E3 ligases depends on the F-box proteins, which bind the substrates. In Arabidopsis, the SCF^TIR1 E3 ubiquitin ligase complex contains multiple proteins. The F-box protein, AtTIR1, binds AUX/IAA repressors, an adaptor protein that binds the F-box protein, ASK, a scaffold protein CUL1 and an RBX that helps transfer the ubiquitin moiety to the substrate (Figure 1) [42–45].

The Arabidopsis thaliana auxin-inducible transcription system was adapted by Kanemaki and colleagues to develop the AID system [15]. All the components of SCF^TIR1 E3 ubiquitin ligase, except the F-box protein TIR1, are conserved in animals. Notably, the F-box binding C-terminal region of Skp1, the adaptor protein, is highly conserved. Therefore, exogenous AtTIR1 expression in animals leads to the assembly of SCF^TIR1 E3 ubiquitin ligases with exogenous AtTIR1 as its substrate receptor. However, the substrates for SCF^TIR1 E3 ubiquitin ligases, the AUX/IAA repressors, are also absent in animals. Therefore, the AUX/IAA repressors can be used as degrons. Endogenous proteins with AUX/IAA repressor tags are subjected to proteasomal degradation by the chimeric SCF^TIR1 E3 ubiquitin ligase upon auxin addition. On auxin wash-off, the degradation stops and the target protein levels recover (Figure 2) [15]. Unlike RNAi-mediated depletion, which results in inconsistent protein depletion, the AID system in most cases results in a 90% depletion in the target protein levels. AtIAA17 was the first degron utilized from a repertoire of AUX/IAA repressors due to its short half-life in plant seedlings [46]. The AID degradation system has now been adapted to S. cerevisiae [15], S. pombe [47], Toxoplasma [41], Plasmodium [48], C. elegans [17], Drosophila [40], zebrafish [49], chicken DT40 cells [15], mouse [50] and mammalian cell lines [15,51].

In mammalian cells, TIR1 from Oryza sativa (OsTIR1) is used instead of AtTIR1. The higher incubation temperatures needed for mammalian cell culture growth probably compromise the interaction between AUX/IAA repressors and AtTIR1 [15,44]. OsTIR1 is the predominant choice for exogenous TIR1. AtTIR1 is adopted in C. elegans owing to the low temperatures at which C. elegans grow [17].

The exogenous TIR1 gene is targeted in the genome either to random loci [15] or to specific safe harbor loci such as AAVS1 [38] and Hipp11 (H11) [52] in human cells, and Rosa26, Hipp11 (H11) and Tigre in mouse cell lines [39,52]. Random insertions may lead to variable expression owing to position effects and require selecting clones expressing appropriate levels of TIR1, making it time-consuming. Therefore, safe-harbor loci are the preferred option. An alternative strategy to express TIR1 in mammalian cells is to express TIR1 fused with a low-expressed essential gene like RCC1. The two proteins are separated by a self-cleaving peptide, T2A, post-translationally [13]. Microinjection of OsTIR1 RNA in zebrafish resulted in nonuniform degradation of the AID-tagged proteins across cells, suggesting the need to generate transgenic OsTir1 lines [49]. Constitutive and conditionally expressing OsTir1 transgenic mice have been generated [52,53]. Recently, transgenic mice expressing an improved version of OsTIR1, namely OsTIR1F74G, have also been developed [50]. Since constitutive TIR1 expression caused leaky degradation of target proteins, a doxycycline-inducible system was utilized as an alternative in mammalian cells [38]. A variety of inducible promoters are available for use in yeast. In C. elegans, tissue-specific expression of TIR1 is extensively employed [17,54,55], where tissue-specific promoters control TIR1 expression. Such tissue-specific lines were also developed for Drosophila [40,56]. Further, to improve degradation in yeast, the AtTIR1 is fused to Skp1, the adaptor protein [47]. In yet another strategy, in addition to inducing protein degradation, the transcription of the AID-tagged protein is also inhibited by incorporating a repressible promoter controlling the tagged target protein [47].

The degron tag, AtIAA17, is a 229 amino-acid protein that could affect the function of smaller proteins due to its size [15]. Subsequently, minimal versions of AtIAA17, namely AID* (71–114aa) [57], mini-AID/mAID (65–132 aa) [58] and AID⁴⁷ (63–109 aa) [59] were developed with no loss in the system's efficiency. Mini-AID has been employed in mammalian cells, whereas the C. elegans community utilizes AID* (Supplementary Table 1 [6,13,50,54,55,60–84]). Mini-IAA7/mIAA7 (37–104 aa), another AUX/IAA repressor, has been used in combination with a different F-box protein, AtAFB2 [60]. A fluorescent tag can be added along with an AID tag to track protein degradation via flow cytometry as well as for cell localization studies. Recently, an endogenous protein tagged with superfolder GFP with mIAA7 tag incorporated within loop 9 (SfGFP^l9mIAA7 ) was degraded in the presence of AtAFB2 and IAA. The incorporation of mIAA7 did not affect the fluorescence of superfolder GFP [61].

Auxins such as IAA (also known as natural auxin) or its synthetic analog, 1-naphthalene acetic acid (1-NAA), are the ligands that initiate the AID signaling cascade. While IAA is the predominant choice [15,38,39,60,62], 1-NAA is shown to perform better in zebrafish [49]. However, 1-NAA and IAA are used at very high concentrations (500 μM) [15]. Recently, alternative synthetic auxins such as 5-phenyl indole-3-acetic acid (5-Ph-IAA) and 5-adamantyl indole-3-acetic acid (5-Ad-IAA) have been developed. These work at low concentrations (1–5 μM) with modified OsTIR1 mutants (see also “Limitations and recent advancements in AID technology”) [50,85].

Applications of AID technology

Metazoans have a complex cellular organization with multiple cell types. Proteins perform many cellular functions and protein depletion may elicit abnormalities indicative of these functions. Housekeeping functions are performed by the same proteins across all cell types. In contrast, some proteins perform different functions depending on the cell type, developmental stage, cell cycle phase and cellular subcompartment. The AID system has been shown to mediate inducible, direct and rapid protein depletion useful to investigate protein functions.

Study of protein function at cellular level

Loss-of-function studies using knockouts or RNAi may activate compensatory mechanisms. Additionally, gene knockouts could produce functional truncated proteins masking the phenotype indicative of exact protein function. For example, in yeast, ORC2 is an essential protein implicated in DNA replication initiation. However, based on a CRISPR-mediated knockout study in mammalian p53^-/- HCT116 cells, ORC2 was reported to be a nonessential gene [86]. The nonessential nature of ORC2 was subsequently shown to result from the presence of a truncated functional ORC2 protein in these cells [6]. The AID system depleted ORC2 within 4 h of auxin treatment and the phenotype revealed that ORC2 is an essential gene [6]. Rapid and efficient protein depletion by AID technology can help dissect proteins' direct and indirect roles. GRASP55 is responsible for Golgi cisternal stacking in cell-free systems [87]. Analysis of GRASP55 following 72 h of RNAi-mediated knockdown in HeLa cells suggested either a disruption of Golgi lateral ribbon connectivity [88] or no function for GRASP55 [89]. Analysis post 96 h of RNAi treatment showed both lateral ribbon connectivity and cisternal stacking defects [90]. Auxin-mediated depletion of GRASP55 for 2 h showed no defects, and GRASP55 was undetectable in cells [63]. However, prolonged auxin-induced depletion of GRASP55 for 48 h resulted in a lateral connectivity defect, and this was due to the loss of Golgin-45, which is known to disrupt ribbon connectivity [63,91]. Put together, the phenotype observed during prolonged GRASP55 depletion using the AID system (48 h auxin treatment) and RNAi could be because of the cumulative accumulation of defects rather than the direct effect of GRASP55 depletion. Compared with RNAi and knockouts, AID technology can help dissect the direct and indirect roles of proteins because of rapid kinetics. However, we cannot rule out the possibility of indirect effects. NUP153, TPR and NUP50 form the nuclear basket on the nucleoplasmic face of the nuclear pore complex (NPC). Similar to GRASP55 RNAi studies, RNAi-mediated loss of NUP153 ambiguously showed TPR mislocalization [92,93] or no phenotype [94]. Auxin-mediated depletion of NUP153 did not affect the localization of the assembled TPR in the interphase cells but prevented TPR's postmitotic assembly at the NPC [13]. Thus, the temporal dissection of NUP153 function during interphase and postmitosis was feasible because of the accelerated depletion kinetics of AID technology.

Study of transcriptional dynamics

The acute depletion of SPT6, a histone chaperone, in U2OS cells using the AID system revealed novel functions for SPT6 during transcriptional elongation and termination not observed in RNAi knockdown [64]. SPT6 RNAi depletion triggered the initiation of spurious transcripts from gene bodies, probably because of the perturbed histone landscape [64]. Whether prolonged auxin depletion also leads to spurious transcripts due to SPT6 depletion remains to be determined. Likewise, XRN2, a 5′-3′ exoribonuclease, has been implicated in the transcriptional termination of transfected plasmids in human cells [95]. However, genomewide studies employing XRN2 RNAi suggested no role in transcriptional termination [96]. Transcriptional termination defects were observed when XRN2 RNAi was combined with the expression of a dominant negative XRN2, indicating that the residual XRN2 protein is sufficient for efficient transcriptional termination [97]. Recently, auxin-mediated depletion confirmed XRN2's role in transcriptional termination [65]. In HCT116 cells, auxin-induced depletion of DIS3, a 3′-5′ exonuclease of RNA exosome, resulted in the accumulation of unstable RNAs such as enhancer RNAs, promoter upstream transcripts (PROMPTs) and products of premature cleavage and polyadenylation not observed in wild-type cells or cells in which DIS3 was depleted using RNAi. These unstable RNAs accumulated within 1 h of AID-mediated protein degradation, highlighting the rapid turnover of unstable RNAs, a phenomenon unlikely to be caught in long-term depletion studies [66].

Study of cell cycle-specific roles of proteins

Many cellular processes, such as cell division, differentiation and replication are dynamic in nature, with many proteins functioning at a rapid pace. These functions can be studied only by depleting proteins at specific time points during these dynamic processes, requiring properly timed, quick and stage-specific protein degradation. The rapid protein degradation inducible by the AID system makes it preferable over earlier methodologies to study dynamic cellular processes. For example, RIF1-deficient HCT116 cells are sensitive to aphidicolin-induced replication stress. It was not clear if RIF1 has functions during or after replication stress as it plays a role in replication initiation, DNA repair and mitotic chromosome segregation. Auxin-mediated depletion of RIF1 was easily performed during or after replication stress, revealing essential roles for RIF1 during both phases [67]. Likewise, staged auxin-induced depletion of ORC2 protein in HCT116 cells synchronized at the G1-early S-phase showed replication initiation and mitotic defects. It was possible to induce protein depletion 4.5 h before releasing the cells into the S-phase using the AID system [6]. During mitosis, some subcellular organelles are fragmented during prophase and reformed during telophase or cytokinesis. As already mentioned, GRASP55 exhibited no role in either Golgi cisternal stacking or lateral connectivity upon auxin-induced depletion (2 h) in contrast to the findings of RNAi studies. It was hypothesized that GRASP55 may play a role during postmitotic Golgi reformation. Auxin-mediated depletion induced 2 h before S-trityl-L-cysteine (STLC; an Eg5 kinesin inhibitor that arrests cells in prometaphase) wash-off demonstrated no role for GRASP55 in postmitotic Golgi reformation [63]. Thus, AID technology, with its properties of fast and inducible protein degradation combined with properly synchronized cell populations, makes it feasible to understand stage-specific cellular processes.

Study of tissue-specific protein functions

Since several proteins are known to have different functions in different tissues, it is desirable to deplete proteins in a tissue-specific manner. Tissue-specific RNAi has been extensively applied by employing the Gal4-UAS system in Drosophila [98–100]. The Gal4-UAS system is now utilized to achieve tissue-specific depletion of AID-tagged proteins. The UAS sequence controls TIR1 expression, and tissue-restricted expression of TIR1 depends on the choice of Gal4. In Drosophila, Vasa functions in embryonic patterning and germline specification. Endogenously tagged Vasa-AID was explicitly depleted in the germline using germline-specific Gal4 (nosGal4VP16), and the Vasa-depleted eggs failed to hatch, displaying no abdominal segments like the vasa hypomorphic alleles [56,101]. DHC1 is a major subunit of the dynein complex implicated in nuclear migration, mRNA localization and chromosome segregation. As DHC1 is ubiquitously expressed in C. elegans, DHC1 tagged with AID-GFP was depleted in the pharynx, gut and germline with tissue-specific expression of TIR1 using myo2, ges-1 and sun-1 promoters, respectively [17]. BBLN-1's role in maintaining intestinal luminal morphology in C. elegans was understood by depleting the protein using an intestine-specific promoter, elt-2, to drive TIR1 expression [55]. Recently, conditionally activable TIR1 transgenic mice generated facilitated tissue-specific depletion [52,53]. Depletion efficiencies of AID-tagged proteins vary across tissues. Therefore, there is a need to optimize the time and amount of auxin required in different tissue cell types in various model organisms. For example, compared with other cell types, auxin-induced degradation of Ncaph, the kleisin subunit of the condensin complex, is very low in spermatocytes and bone marrow erythroblasts of mice constitutively expressing OsTIR1 [53]. Similarly, degradation of the AID-tagged EGFP reporter in the brain was less efficient than in the small intestine and heart of OsTIR1F74G transgenic mice [50].

Study of protein function during organismal development

Auxin-induced protein depletion can be employed to study the role of proteins during an organism's development. Noninvasive auxin treatment is straightforward when development is external. Therefore, rapid protein depletion can be executed in the larval stages of C. elegans, Drosophila and Danio rerio. In C. elegans, NHR-23, a nuclear hormone receptor, has a role in epidermal development and molting. Expression of NHR-23 changes with developmental stages and is high in the embryonic stage and between larval molts and low during molting stages [102]. Auxin-mediated depletion of NHR-23 resulted in 100% penetrance, with all the larvae arresting at the L1 larval stage [17]. In contrast, RNAi of nhr-23 caused most of the animals to arrest later during development, at the L3 larval stage, while only 2% were arrested at the L1 stage, suggesting a delay in RNAi-mediated depletion. Also, RNAi of nhr-23 resulted only in a 75% depletion of NHR-23 mRNA levels [102]. In another study in C. elegans, auxin-induced depletion of BBLN-1 during the late L1 stage or early L4 stage caused intestinal invaginations due to loss in intermediate filament integrity, suggesting its requirement throughout development [55]. Protein depletion across developmental stages of Drosophila is feasible with AID technology. However, auxin-mediated depletion would not be effective in Drosophila pupal stages as the third instar larvae stop feeding before pupation. In Danio rerio, GFP-tagged proteins were depleted using an AID-fused anti-GFP-nanobody and TIR1 RNA injection during early development [49]. However, a broad application across developmental stages in zebrafish will be facilitated by establishing transgenic TIR1 lines. In mice, degradation of Ncaph in E10.5 embryos was efficient on intraperitoneal injection of IAA into female pregnant mice expressing OsTIR1 and AID-tagged Ncaph [53]. Similarly, E9.5 and E13.5 embryos were depleted of EGFP when pregnant transgenic mice containing OsTIR1(F74G) and mAID-EGFP were injected intraperitoneally with 5-Ph-IAA [50]. Similar to tissue-specific studies, it is necessary to ascertain the depletion kinetics across developmental stages in different model systems, as AID-induced degradation was shown to vary during the developmental stages in transgenic C. elegans [17].

Isoform-specific functions of proteins

Increased complexity in organisms positively correlates with the genome size, but the number of genes does not show a corresponding increase [103]. One strategy organisms utilize to enlarge their functional repertoire of proteins is by expressing multiple isoforms from a gene via alternative splicing. Isoforms might have different protein interaction partners, enzyme activities and functions in subcellular compartments [104,105]. Gene-locus tagging with AID can be designed such that it results in the depletion of all isoforms of the encoded protein. Following depletion, isoform-specific functions can be investigated by complementation with specific target isoforms. For example, in Toxoplasma, two protein kinase G isoforms, PKG^I and PKG^II, have similar catalytic and regulatory domains. However, PKG^I localizes to the plasma membrane, whereas PKG^II is cytoplasmic. Auxin-mediated depletion of endogenous PKG tagged with mAID followed by complementation with either isoform revealed that, unlike PKG^I, PKG^II was dispensable for plaque formation on human foreskin fibroblast layers [41].

Domain-specific function of proteins

Protein families are broadly classified into single-domain families or multidomain families. During evolution, the frequency of gene fusion events has been fourfold higher than gene fission events [106]. Therefore, eukaryotes have predominantly multidomain protein architectures, which provide opportunities for functional diversification and innovation. For example, the polycomb protein in Drosophila has 3–5 vertebrate homologs. These paralogs have chromodomains and Pc boxes but vary in their lengths and the presence of other domains and motifs. The functional differences between vertebrate homologs of polycomb protein are attributed to the differences in their domains [107,108]. So, delineating their domain-specific functions is crucial. Protein domain-specific function can be deciphered by tagging the endogenous genes with AID and then complementing with truncated proteins. For example, in mouse embryonic stem cells (mESCs), SPEN binds Xist RNA, and SPEN's loss leads to X-chromosome inactivation (XCI) defects. Truncated versions of SPEN were expressed in the background of auxin-mediated SPEN depletion. Deleting RNA recognition motifs prevented SPEN's binding to the Xist RNA. In contrast, the SPOC domain deleted SPEN retained Xist RNA binding but compromised X-chromosome silencing. It was shown that the SPOC domain helps bridge Xist RNA with repressors [68]. Thus, auxin-induced depletion followed by complementation allowed the functional dissection of the various domains of SPEN in mESCs XCI. MPP8 interacts with epigenetic modulators to mediate the silencing of genes and retrotransposons. In mESCs, AID-mediated depletion followed by complementation with truncated proteins demonstrated that the N-terminal chromodomain region of MPP8 is essential for initiating LINE repression but not for maintenance of the repression. In contrast, the C-terminal region was observed to be essential during both stages [69]. Analogous complementation studies have also been performed in C. elegans where kinase activity of Aurora B kinase (AIR-2; a serine/threonine kinase critical during cell division) was outlined to be crucial for ring complex formation around bivalent chromosomes and chromosome segregation but were dispensable for kinetochore assembly [70]. Motif- and residue-specific protein functions can also be similarly studied.

Dose-dependent depletion of proteins

Protein levels vary depending on active, cell-type-specific cisregulatory elements. The phenotypes and their magnitude correlate with the levels of proteins in a cell. Examining the physiological effects of altering the protein levels in cells would be interesting. Cellular protein levels can be easily altered by varying the auxin levels in the AID system. For example, in Drosophila, AID-mediated Vasa depletion using maternal Gal4s exhibited abdominal defects whose magnitude positively correlated with auxin levels [56]. Similarly, auxin dose-dependent depletion of CTCF revealed a positive association between auxin dose, CTCF depletion and folding defects at the X-inactivation center [39]. Protein functions vary with the levels of the protein in the cell. For example, condesin is required for mitotic chromosome structure as well as for sister chromatid separation during anaphase. Auxin dose-dependent depletion of SMC2 (a shared subunit of both condensin complexes) in synchronized chicken DT40 CDK1^as cells showed that both functions were unaffected at 25% depletion. A 40% SMC2 depletion resulted in 55% of cells having defects in anaphase sister chromatid segregation, but 90% of cells formed mitotic chromosomes on mitotic entry. This observation suggests that higher levels of condensin are required for sister chromatid segregation function than for the formation of the mitotic chromosome [109]. Differences in target protein concentrations have been shown to result in different developmental fates. For example, homeotic genes are known to affect different segmental identities depending on their concentrations [110]. In Drosophila embryos, different levels of Dorsal nuclear protein instruct the formation of mesoderm, dorsal ectoderm and neuroectoderm [111]. Therefore, it would be interesting to study the effect of disruption of these concentration gradients on developmental fates. The AID system provides an opportunity to execute such experiments in a controlled manner by varying auxin levels.

Study of essential proteins

Essential proteins are required for the viability of cells or organisms. Conditional knockout strategies are imperative in these cases, as traditional knockouts cannot be applied. The AID system helps generate conditional mutants of essential genes. For example, knockouts of kleisin subunits of condensin I or II result in lethality and require the generation of conditional knockouts. The conditional mutant of kleisin subunits generated using the AID system showed that condensin II is responsible for the formation of 400 kb outer loops, which are subdivided into 80 kb loops by condensin I during mitotic chromosome formation [112]. Some proteins have tissue-specific and developmental-stage-specific roles, with at least one of these roles being essential. For example, during Drosophila development, the loss of homeotic genes, which specify the segmental identity along the anterior-posterior axes, causes embryonic or larval lethality [113]. However, homeotic genes also have noncanonical functions in different tissues during postembryonic development [114–116]. Investigation of these noncanonical functions is impossible with constitutive gene knockouts because of their early lethality. AID technology allows user-controlled protein depletion at any desired stage of development. In metazoans, PAR-6 is required for determining the apical-basal polarity of a wide variety of cells. In C. elegans, PAR-6 has been extensively studied during embryonic development, but its postembryonic function was an enigma. During larval development, depletion of PAR-6 with the AID system showed that this essential protein is required for larval epidermal epithelial development via its interaction with NOCA/Ninein-1 to maintain noncentrosomal microtubule organization [54]. Likewise, AID technology can also enable the discovery of moonlighting roles of essential genes in other model systems.

Limitations & recent advances in AID technology

Direct depletion of proteins using AID technology has uncovered several novel protein functions and resolved ambiguous results. However, it has a few drawbacks. Unlike other tag-based degradation technologies, the AID system requires two gene-editing events, one for tagging the endogenous protein with the degron AID, and another for degron-binding F-box protein, TIR1. Recently, in chicken DT40 cells and human cells, endogenous protein tagging and TIR1 targeting were done simultaneously by a single transfection [85,117]. In this study, the endogenous target gene is disrupted by the insertion of a construct that codes for both the TIR1 gene and AID-tagged target gene driven by a common CMV promoter. A T2A (self-cleaving peptide sequence) and internal ribosomal entry site (IRES) sequences separate the TIR1 gene and AID-tagged target gene [85]. After translation, self-cleavage by T2A separates the TIR1 and AID-tagged protein. One disadvantage of this system is that the expression of the AID-tagged protein is not in the physiological range due to the CMV promoter used. However, the IRES sequence results in a range of clones each having a different level of target gene expression, which allows for the selection of an appropriate clone [85]. Known functions of tagged protein may need to be verified as attachment of the auxin degron can sometimes affect their activity.

Small molecules used to elicit degradation, such as IAA and 1-NAA, are utilized at very high concentrations of 500 μM, which result in cellular defects. IAA treatment of unmodified HEK293T cells has been shown to trigger the upregulation of genes with aryl hydrocarbon receptor binding sites [62]. Recent studies in C. elegans report that IAA usage or the expression of AtTIR1 and degrons increased the lifespan of C. elegans irrespective of the target protein depleted [118]. IAA also prevents endoplasmic reticulum (ER) stress by functioning through the XBP-1/IRE-1 pathway of unfolded protein response in C. elegans [119]. These observations suggest that AID technology should be used cautiously in aging and ER homeostasis studies.

The AID system is also prone to leaky protein degradation via proteasome even before induction with auxin [39,50,60,62,85,120]. AID-tagged CTCF had threefold lower protein levels than endogenous CTCF without auxin induction [39]. In human cells, a similar problem made it difficult to generate AID-tagged lines of essential proteins like DHC1 [38] and CPSF73 [65]. AID technology is known to induce rapid protein degradation within a few hours, but the recovery of protein levels after auxin wash-off depends on the target protein and may take several hours [17,39]. It would be very informative to investigate the recovery of dynamic processes on auxin wash-off if the protein replenishment duration can be reduced to a few hours. Some recent developments have helped overcome some of the drawbacks of the AID system, especially the leaky degradation of proteins in the absence of auxin-mediated induction.

Use of conditional promoter for TIR1 expression

DHC1 is an essential gene that is a component of the cytoplasmic dynein complex and is involved in multiple processes. In HCT116 cells, tagging of endogenous DHC1 with mAID was unsuccessful in the presence of constitutive OsTIR1 expression because of leaky degradation. However, the use of a doxycycline/tetracycline-inducible promoter to drive OsTIR1 expression helped generate clones of DHC1 tagged with mAID. Although the generation of clones was feasible, there was mitotic arrest even before auxin addition indicating leaky degradation [38]. This leaky degradation was circumvented by using auxinole, a competitive inhibitor of OsTIR1 [120,121]. The use of auxinole improved the recovery kinetics of the tagged protein during auxin wash-off [120].

AtAFB2 as an alternative to OsTIR1

An alternative F-box protein, AtAFB2, used in place of OsTIR1, decreased the leaky degradation of mAID-tagged proteins fivefold. However, the use of AtAFB2 also decreased the overall degradation kinetics of mAID-tagged proteins in the presence of auxin. Replacing mAID with mini-IAA7 (37–104 aa), another AUX/IAA repressor, improved auxin-induced degradation kinetics threefold in this system [60].

Expression of PB1 domain of ARFs

In plants, AUX/IAA repressors bind ARFs and prevent ARFs' regulation of auxin-inducible genes. AUX/IAA repressors bind to the Phox & Bem1 (PB1) domain of ARFs to repress ARFs function. Expression of the ARF16-PB1 domain along with AID-tagged proteins and OsTIR1 inhibits leaky but not auxin-induced degradation of target proteins in animals [62].

Improved mutants of OsTIR1

Improved versions of OsTIR1 (used with inducer IAA), OsTIR1F74G (with 5-Ph-IAA) and OsTIR1F74A (with 5-Ad-IAA) were designed by two independent groups. These systems utilize about 500-fold less synthetic auxin. They have also been shown to decrease the leaky degradation of proteins [50,85]. An improved version of the original AID technology has also been developed in C. elegans using AtTIR1F79G with 5-Ph-IAA, which is the analogous version of OsTIR1F74G in Arabidopsis [122]. The improved version in Saccharomyces pombe utilizes the other improved mutant OsTIR1F74A with 5-Ad-IAA [123].

Although these advances address the AID system's leaky degradation problem, they could also alleviate other problems. For example, 5-Ph-IAA usage and OsTIR1(F74G) could boost the application of the AID system in aging and ER stress homeostasis studies. Also, a 500-fold reduction in the amount of synthetic auxins (5-Ph-IAA and 5-Ad-IAA) makes in vivo applications of these systems in higher organisms feasible.

A brief comparison between various advances made to the original AID technology would help users choose the appropriate technology for their research. A comparison in mammalian cells between auxin-induced degradation of mIAA7-tagged proteins by OsTIR1 and AtAFB2 suggested that AtAFB2 is better than OsTIR1 [60,124]. In HCT116 cells, genetic engineering of DHC1 with mAID in the background of OsTIR1 was unsuccessful. However, DHC1 was successfully tagged with mIAA7 in the background of OsTIR1(WT), OsTIR1(F74G) or AtAFB2, but not with the OsTIR1-ARF system [50]. Leaky degradation was least in OsTIR1(F74G) followed by AtAFB2. However, the protein degradation efficiency of OsTIR1(F74G) in combination with mAID was better than with mIAA7. Additionally, OsTIR1(F74G) uses 5-Ph-IAA at a 1-μM concentration, which prevents cellular defects resulting from high concentrations of small-molecule inducers [50]. These reports suggest that AtAFB2, OsTIR1(F74A) and OsTIR1(F74G) and their appropriate degrons are better than the original degron in mitigating leaky degradation and improving degradation efficiencies. A comparison utilizing OsTIR1(F74G), OsTIR1(F74A), AtAFB2 and OsTIR1 with the ARF16-PB1 domain to assess the degradation efficiencies for multiple proteins in different model organisms would be very informative.

Comparison of AID technology with other tag-based degradation technologies

Several tag-based degradation technologies similar to AID technology have been developed. These include the dTAG system, HaloTag system, HaloPROTACs and SMASh. The dTAG system uses the FKBP12^F36V tag. When tagged to intracellular proteins, it induces proteasome-mediated degradation by recruiting endogenous CRL4 E3 ubiquitin ligase with substrate receptor CRBN in the presence of the small molecules dTAG-13, dTAG-7 or their equivalents [33]. A novel small molecule, dTAG^v-1, degrades FKBP12^F36V-tagged proteins by recruiting CRL2 E3 ubiquitin ligase with VHL as its substrate receptor [125]. This degradation tag system has been used extensively in mammalian cell lines.

Following translation, proteins fold by burying their hydrophobic residues in the interior. HaloTag technology is based on the fact that proteins with hydrophobic residues exposed to the exterior are subjected to proteasomal degradation. HaloTag fusion proteins are covalently modified by hydrophobic tag (HyT) molecules. HyT molecule consists of two moieties, the HaloTag reactive linker, which covalently modifies the HaloTag fusion protein, and the hydrophobic moiety. After modification by the HyT molecule, the HaloTag fusion protein becomes hydrophobic and is degraded [34]. HaloPROTACs use small-molecule-based PROTACs to target HaloTag fusion proteins to E3 ubiquitin ligases [35].

The SMASh degradation system is adapted from the hepatitis C virus (HCV). The SMASh tag consists of an NS3 protease-recognition site followed by NS3 protease domain and NS4A hydrophobic sequence. The SMASh-tagged protein of interest (POI) is cleaved at the NS3 recognition site by the NS3 protease domain immediately after synthesis, and the POI is stable without the tag. However, in the presence of asunaprevir, an inhibitor of the NS3 protease, the POI with the SMASh tag is degraded because of the NS4A hydrophobic sequence. One disadvantage of this system is that it degrades only the newly synthesized proteins in the presence of asuprenavir, and therefore, the degradation kinetics depends on the half-life of the protein. This system, however, can be utilized to investigate protein half-lives [36].

The advantages and disadvantages of these systems are summarized in Table 1. The availability of multiple systems makes it feasible to study the synergistic or additive effects of proteins on biological phenomena. Among the systems described here, the degradation efficiencies of the dTAG system and AID system are comparable and better than other systems. A study comparing the degradation of a highly abundant protein LMNA in HEK293A cells and A431 cells using dTAG and AtAFB2-mIAA7 systems showed that LMNA degradation and recovery kinetics are better with AtAFB2-mIAA7 than dTAG system [60]. However, the degradation times of RAD21 [126,127], SPT6 [64,128], RIF1 [67,129] and CTCF [39,130] were comparable in both systems. It is important to note that these comparisons except for CTCF are based on target protein degradation in different cell lines by different groups (see Supplementary Table 1 for the AID system's degradation times [6,13,50,54,55,60–84]). The recovery rate of the reporter construct after auxin/dTAG-13 wash-off is 1 h with both AID technology as well as the dTAG system. This recovery rate is known to vary for different endogenous proteins. Overall, this suggests that the dTAG system is comparable to the AID systems with regard to degradation kinetics and depletion efficiencies. However, a more accurate comparison would require depleting multiple proteins using both techniques in a range of cell types. Also, the dTAG system has the added advantage of requiring a single genetic modification by utilizing endogenous E3 ligase. This advantage, however, does not allow for tissue-specific degradation of proteins. The dTAG system also awaits applications in other model systems.

Comparison of AID technology with small-molecule inhibition

Small-molecule inhibitors have been developed for many proteins and are utilized for therapeutic purposes [135,136]. Many biological phenomena have been understood with the use of small molecules [137–139]. However, not every protein molecule can be inhibited with small molecules because of the difficulty of finding protein-specific small-molecule inhibitors. Sometimes, the small molecule inhibiting a protein may also inhibit other related proteins and lead to secondary nonspecific effects. These secondary effects, if any, would differ depending on the small molecule [140–142]. Secondary effects in cells treated with auxin have been reported. For example, IAA has been shown to upregulate genes with aryl hydrocarbon receptor binding sites in unmodified HEK293T cells [62]. However, the secondary effects of auxin would be similar in a cell type irrespective of the target protein degraded. Another advantage of auxin-mediated depletion is that while small molecules may have domain-specific inhibitory effects, auxin-mediated degradation depletes the whole protein, thereby eliminating phenotypes resulting from the unaffected domains of the proteins. Thus, auxin-inducible depletion offers an advantage over small-molecule inhibition in some cases. In Plasmodium berghei, calcineurin is a multifunctional calcium-regulated phosphatase with expression in the cytoplasm of merozoite/schizont, sporozoite and male gametocyte stages. The irreversibility of knockouts allows functional investigation only at the initial stages of the parasite life cycle. Another approach is to use small molecules, such as cyclosporin A or FK-506, to inhibit calcineurin at various stages. Cyclosporin A and FK-506 are known to interact with prolyl isomerases such as cyclophilins or FK-506 binding proteins, respectively, to inhibit calcineurin function at various stages [143,144]. Since prolyl isomerases have several substrates, inhibiting prolyl isomerases may cause secondary effects unrelated to calcineurin inhibition. Previous studies demonstrated calcineurin to be essential for parasitic development in red blood cells [144]. In contrast, auxin-mediated depletion revealed that calcineurin was not essential for parasitic development, and the merozoite numbers in red blood cells were not affected. However, calcineurin depletion compromised the parasitic invasion of red blood cells due to the parasite's inability to attach to erythrocytes [18].

Conclusion

With its rapid and inducible protein depletion, AID technology has proved useful in dissecting the direct and indirect functions of proteins. The technology also provides an excellent handle to study essential proteins and dynamic processes, which were previously challenging. AID-mediated investigations of isoform-specific and domain-specific functions of proteins have and can help us understand the complexity of organisms at an unprecedented level. The newer advancements that prevent leaky degradation and synthetic auxins mitigating secondary effects in cells have and will aid us in investigating protein functions at their physiological concentrations. In the present CRISPR-Cas9 era, we anticipate that AID and other targeted protein-degradation technologies will become a major tool in researchers' repertoire.

Future perspective

AID technology can help us ask if different levels of proteins are responsible for different functions. If protein levels indeed dictate protein functions, understanding how this is brought about and regulated would be an important area of research. Combining different targeted protein-degradation technologies, such as AID technology with dTAG technology, would help us understand if two proteins contribute to biological phenomena additively or synergistically. Antibodies generated against degradation tags can be used to perform protein pull-downs and ChIP-sequencing experiments without the need for additional tagging experiments. Comparison of various AID technologies as well as other tag-based degradation technologies will provide a better understanding of these technologies with respect to their degradation kinetics as well as recovery kinetics and would also help the user make an informed choice about which technology to use.