We tagged endogenous target loci with AID tag fused with Clover/eGFP by CRISPR/Cas9-mediated HDR (Natsume et al., 2016; Li et al., 2019). In total, we modified four genes (Rad21, Smc2, Ncaph, and Ncaph2) in mES cells and two genes (RAD21, SMC2) in HAP1 cells (Table 1). We used single gRNA targeting a region around the STOP codon to introduce an AID and selection cassette at the 3′-ends of the genes (Figure 1). Homology-directed integration of the AID cassette was facilitated by 600–1,300 bp homology arms (Supplementary Table S2). In our study, we used a minimal required AID (either mAID or miniIAA7) linked to Clover/eGFP (Figure 1). The donor vector was a mix of the two constructs that contained different selection markers (NeomycinR or HygromycinR). Such a strategy allowed us to select diploid cells that tagged all two alleles efficiently because only cells in which one allele is modified with a construct with a NeomycinR marker and the other allele with a HygromycinR marker will survive in a medium with both antibiotics (double selection).

In mESCs, Rad21 gene targeting resulted in 33% homozygous clones, the Smc2 gene in 40%, the Ncaph gene in 25%, and the Ncaph2 gene in 26%. In HAP1 cells, RAD21 gene targeting resulted in 11% homozygous clones and the SMC2 gene in 77% (Table 1). We also performed second round modification of one of the mESC Rad21-miniIAA7-eGFP clones to introduce Smc2-miniIAA7-eGFP for double protein degradation. In this case, we could not use Neo/Hygro selection; instead, we visually inspected clones and picked only those with eGFP spread across both the nucleus and the cytoplasm (in Rad21 clones, eGFP is found in the nucleus) (Supplementary Figure S1).

The correct insertion of the AID cassette was confirmed by long-distance PCR (Supplementary Figure S2) and Sanger sequencing. The molecular weight of fusion proteins was verified by immunoblotting (in case of Rad21 and Smc2). All analyzed clones were positive for Clover/eGFP expression and had a correct protein distribution pattern inside cells, i.e., Rad21-miniIAA7-eGFP and Ncaph2-miniIAA7-eGFP were confined to the nucleus, while Smc2-miniIAA7-eGFP and Ncaph-miniIAA7-eGFP resided predominantly in the cytoplasm (Supplementary Figure S1). At least four homozygous clones were expanded and frozen for each gene modification, and a few (2–3) clones were selected for subsequent transfection with AFB expression vectors.

As far as we know, this is the first application of the AID system in human HAP1 cells. This cell line is a powerful tool for genome engineering because most of the genes are present only in one copy, which simplifies gene modifications. The efficiency of gene targeting was comparable to that of mES cells (Table 1) but required an additional selection step because HAP1 cells have a tendency to spontaneously transit to a diploid state during proliferation (Beigl et al., 2020). Importantly, ploidy shift interferes with targeting experiments and complicates obtaining homozygous clones after AID cassette targeting. Haploid cells could be selected in two ways. First of all, the mixed population of haploid/diploid cells could be reliably sorted by cell size using FACS (Beigl et al., 2020) (Figure 2A; see also Figure 2B, where two subpopulations are highlighted with a color scheme). Alternately, haploid colonies could be found and picked after simple visual inspection of a plate after an antibiotic selection step. Figure 2C demonstrates fluorescent RAD21-miniIAA7-eGFP HAP1 colonies of different ploidy from one of the experiments. HAP1 cell diploidization at the late passages (after AFB transfections) does not seem to affect AID degradation efficiency.

Selected single homozygous tagged clones (Table 1) were transfected with either CMV-OsTIR1-hPGK-PuroR construct (Natsume et al., 2016) or EF1-AtAFB2-Cherry-IRES-PuroR construct (Li et al., 2019) and cultivated in the presence of puromycin for 1 week. Surviving clones with randomly integrated AFBs were expanded and tested for the POI degradation efficiency on FACS. At this stage, we noted that the two systems show significant degradation of the target protein at different times. Therefore, to characterize clones with the integration of AtAFB2, we performed FACS analysis after 1.5–3 h and for OsTIR1 after overnight incubation in auxin. Plot diagrams show the distribution of Clover/eGFP fluorescence in mESC and HAP1 clones (Figures 3A,B). The OsTIR1 approach using CMV-OsTIR1-hPGK-PuroR was mostly inefficient and required more time to degrade the POI without any actual significant depletion. At the same time, the AtAFB2 transgene, in which the AFB is co-expressed with PuroR and Cherry genes from a single promoter, allowed to achieve efficient and rapid degradation. Western blot analysis confirmed FACS results for selected clones (Supplementary Figure S3).

Clones with the highest level of protein degradation were selected for further studying the dynamics of this process (data for POI = Rad21 and Smc2). We also included HAP1 clones (AtAFB2 system) with intermediate and low sensitivity to auxin to investigate whether longer auxin exposure can enhance protein depletion efficiency.

Overall, AtAFB2 cell lines displayed more rapid and efficient depletion of the degron-fused proteins (Figures 3C–E). For instance, Rad21 levels significantly diminished after 30 min of auxin exposure (Figures 3C,D), which led to the collapse of nuclear architecture and cell death after 24 h in accordance with earlier results. Although AtAFB2-Cherry protein was mostly localized in the cytoplasm due to weak NLS, it was enough to rapidly degrade nuclear chromatin factors (Li et al., 2019). Moreover, extended time of auxin exposure does not seem to improve protein degradation efficiency: the most pronounced changes in proteins levels occurred during the first 30–120 min and then reached a plateau (Figure 3E).

OsTIR1 and AtAFB2 systems differ in several components; in particular, they use different AFB proteins, tag sequences, and the expression of the AFB protein driven by different promoters (Figure 1). Because of these differences, it is challenging, if not impossible, to compare these systems in parallel under the same conditions accurately. However, we decided to assess whether the promoter of AFB protein affects the degradation efficiency. To do this, we cloned the coding part of the OsTIR1 gene instead of AtAFB2 (Figure 4A). Then, we integrated the resulting construct into a random place in the genome of mESC clones with the tagged Smc2 or Rad21 gene. For each gene, we picked three random colonies and tested the degradation efficiency. In one clone of Rad21 and in two clones of Smc2, there was no degradation at all. For the rest, we examined the dynamics of degradation. None of the clones tested showed more than 75 percent degradation (Figure 4B). These data may indicate that the difference between the systems from the two articles is not due to the promoter providing greater expression in the AtAFB2 system.

The AtAFB2 construct contains an in-frame Cherry fluorescent marker which is extremely useful to select suitable clones, since Cherry levels reflect AtAFB2 expression. On FACS, we noted a strict correlation between Cherry expression and POI degradation efficiency for the eGFP signal (data not shown).

Although our initial experiment with the AtAFB2 AID system demonstrated satisfactory results (Figure 3), we observed that a subset of cell population remained refractory to auxin and its proportion increased with passages in culture (data not shown). Apparently, the decline of POI degradation activity stems from the silencing of AFB constructs. Rad21-miniIAA7-eGFP mES cells and SMC2-miniIAA7-eGFP HAP1 cells bearing randomly integrated AtAFB2 constructs were seeded with low density to generate single-cell founder colonies. After overnight treatment with auxin, we observed both eGFP-positive and -negative colonies (Figure 5B). Furthermore, we detected a significant number of eGFP/Cherry mosaic colonies (Figures 5A,B), even though all colonies were derived from single progenitors. Generally, cells with silenced AtAFB2-Cherry construct retained eGFP fluorescence (Figures 5A,B). For example, in Figure 4A, Cherry^pos part of the mESC colony successfully degraded Rad21-miniIAA7-eGFP and started to die, while remaining cells kept expressing Rad21-miniIAA7-eGFP in the absence of AtAFB2-Cherry.

It is well known that random integration of transgene leads to variable expression due to position effects and silencing. Integration of the OsTIR1 and AtAFB2 transgenes in the AAVS1 safe-harbor locus in human cells is considered a sufficient measure to provide stable transgene expression (Natsume et al., 2016; Li et al., 2019; Sathyan et al., 2020). We inserted the AtAFB2 transgene into the AAVS1 locus in one of the HAP1 lines with Rad21-miniIAA7-eGFP modification. The AtAFB2 construct has an in-frame Cherry selection marker, and thus, we were able to analyze AtAFB2 expression levels using FACS. Interestingly, all selected Cherry^pos clones with correct AtAFB2 insertion were mostly mosaic in expression with 16% Cherry^pos cells at best (Supplementary Table S1). None of the six clones showed efficient POI depletion in further tests (data not shown). We repeated the experiment with one-cell founder colonies for one of these clones and observed the same outcome with mosaic colonies (Figure 5C).

Thus, in our experiments, targeted integration of the AFB into the AAVS1 safe-harbor locus did not yield any significant advantages over random integration of AFBs. In addition, cells were heterogeneous in AFB expression and the ability to degrade AID-tagged protein after auxin exposure.

Our results show that some cells fail to facilitate degradation of target proteins due to the low expression of AtAFB2 protein. Therefore, it is important to keep cells under constant puromycin selection during multiple passages in culture or re-freezing. To demonstrate this, we cultured mES cells bearing Rad21 and Smc2 modifications (AtAFB2 and OsTIR1 systems) and HAP1 cells with RAD21 modifications for the OsTIR1 system and SMC2 modifications for the AtAFB2 system in the presence of 2 µg/ml puromycin for several passages and then analyzed target protein degradation dynamics after auxin treatment (Figure 5D). As expected, the efficiency of protein degradation was extremely improved in puromycin-treated cells but only for the AtAFB2 system (Figure 5D).

We also inspected whether pH level changes can affect POI degradation efficiencies. Acidification of culture medium pH (6.8 and below) is typically observed in prolonged incubations or in culture mediums with low buffering capacities (Michl et al., 2019). We were gradually changing pH values of the medium used for cultivating HAP1 RAD21/SMC2-miniIAA7-eGFP clones and later detected eGFP fluorescence on FACS, similar to previous experiments. Normal and mildly acidic pH (6.3–7.4) did not affect POI degradation efficiency, while higher pH values (8.6–8.8) effectively blocked POI degradation (Figures 6A,B). The auxin analog NAA (1-naphthaleneacetic acid) showed similar pH sensitivity (Figure 5B). Therefore, physiological pH changes in the cell culture medium do not affect auxin-dependent protein degradation efficiency.

Nevertheless, low pH might influence basal degradation in the AtAFB2 AID system. We inspected the mosaic HAP1 clone with miniIAA7-tagged SMC2, which consists of mixed Cherry^pos/Cherry^neg subpopulations, caused by the silencing of the AtAFB2 expression in some cells (Figure 5B). Generally, Cherry^pos cells have a slightly weaker eGFP fluorescence profile [blue (Figure 6C)] with a shift between two peaks, probably reflecting minor basal degradation. At normal pH or high pH (no AtAFB2 activity), eGFP level shifts remain similar. However, changing the pH value to 6.7 led to a slight decrease in eGFP fluorescence in Cherry^pos cells, but not in Cherry^neg cells (Figure 6C, middle), which might hint at increased basal degradation at low pH. We observed the same slight shift for the OsTIR1 AID system while culturing cells in the acidic pH medium, including culture media in which cells were growing overnight (old media) (Figure 6D).

All cell lines were grown at 37°C with 5% CO₂ and passaged every 2–3 days. HAP1 cells (Horizon Discovery) were cultured in IMDM (Thermo Fisher) containing 10% FBS (Gibco), 1 mM L-glutamine (Sigma), 0.5 mM NEAA (Gibco), and penicillin/streptomycin (100 U ml⁻¹ each). Mouse ES cells (DGES1 line) established in our laboratory (Kontsevaya et al., 2017) were cultured on a gelatin surface under 2i conditions (1 μM PD, 3 μM CHIR) in DMEM (Thermo Fisher), supplemented with 7.5% ES FBS (Gibco), 7.5% KSR (Gibco), 1 mM L-glutamine (Sigma), NEAA (Gibco), 0.1 mM β‐mercaptoethanol, LIF (1000 U/ml, Polygen), and penicillin/streptomycin (100 U ml⁻¹ each). 3-Indoleacetic acid (IAA, auxin) (Merck, I2886) and 1-naphthaleneacetic acid (NAA) (Merck, N0640) were dissolved in NaOH (1M) solution to a concentration of 500 mM, aliquoted, stored at -20°C, and used immediately after thawing. For all experiments, the cells were incubated with 500 μM indole-3-acetic acid (IAA) (Sigma) for the indicated times.

For testing the pH-responsive auxin-inducible degradation of POI, cell culture media with a pH range of 6.3–8.8 were prepared and used immediately. The cells were harvested no longer than 30 min post auxin exposure because alkaline pH rapidly returns to neutral values in CO₂-rich atmosphere.

Donor vectors with homology arms, degron tag (mAID or miniIAA7), Clover/eGFP, and selection cassette (either NeoR or HygroR) were assembled in one reaction using the NEBuilder HiFi kit (NEB) in the pMK289 backbone (Natsume et al., 2016). Homology arms for RAD21 (NCBI Entrez Gene ID: 5885), SMC2 (NCBI Entrez Gene ID: 10592), Rad21 (NCBI Entrez Gene ID: 19357), Smc2 (NCBI Entrez Gene ID: 14211), Ncaph (NCBI Entrez Gene ID: 215387), and Ncaph2 (NCBI Entrez Gene ID: 52683) genes were PCR amplified from human and mouse genomic DNA with Q5 polymerase (NEB). The lengths of homology arms are featured in Supplementary Table S2. Other functional elements were amplified from plasmids pMK289 (mAID-mClover-NeoR) (Addgene #72827), pMK290 (mAID-mClover-HygroR) (Addgene #72828), and pSH-EFIRES-B-Seipin-miniIAA7-mEGFP (Addgene #129719) (Natsume et al., 2016; Li et al., 2019). For auxin receptor F-box protein (AFB) overexpression, OsTIR1 and AtAFB2 plasmids were used (AtAFB2) (pMK232 CMV-OsTIR1-PURO, Addgene #72834; pSH-EFIRES-P-AtAFB2-mCherry-weak NLS, Addgene #129717) (Natsume et al., 2016; Li et al., 2019). For gene tagging and AAVS1 locus modification, the cells were transfected with donor vectors, Cas9 (Addgene #41815), and corresponding sgRNA-expressing plasmids (Addgene #41824, backbone). In order to precisely replace the AtAFB2 gene in the pSH-EFIRES-P-AtAFB2-mCherry-weak NLS (Addgene #129717) vector with the OsTIR1 gene, we assembled two PCR-amplified fragments (the OsTIR1 gene and the Cherry gene) with the BglII/NotI linearized backbone using NEBuilder HiFi DNA Assembly Master Mix. Sequences of sgRNAs are provided in Supplementary Table S2.

HAP1 cells were electroporated at conditions of 1200 V, 30 ms, 1 pulse using the Neon Transfection system (Invitrogen), according to the manufacturer’s instructions. The day after, electroporation cells were seeded in 100-mm dishes with low density and expanded for 24 h, followed by selection with 500 μg/ml G418 (for RAD21 targeting) or 800 μg/ml hygromycin B (for SMC2 targeting). Mouse ES cells were transfected by Lipofectamine 3000 (Invitrogen) in OPTI-MEM (Thermo Fisher), following the supplier’s recommendations. After 24 h, transfection cells were seeded in 60-mm dishes in the culture medium containing 300 μg/ml G418 and 200 μg/ml hygromycin B for selection of homozygous tagged clones (Natsume et al., 2016). After 10–14 days of selection, resistant colonies of HAP1 cells and mouse ESCs expressing green fluorescent protein (eGFP) were manually picked for subcloning and further analysis. Most of the surviving clones were confirmed to harbor the correct insertion of degron tags by genomic PCR, Sanger sequencing, and western blotting. In addition, the resulting HAP1 clones were stained with propidium iodide (PI), and their DNA ploidy was determined by FACS (Beigl et al., 2020). Several clones for each target gene were chosen for integration of AtAFB2- or OsTIR1-expressing vectors and transformed using identical settings described for the previous step. Selection of the AFB-positive clones was performed using 1 μg/ml puromycin for 1 week. Finally, several mESCs and HAP1 clones with homozygous degron tags and AFB insertions were tested for efficiency of auxin-induced protein depletion, using microscopy, FACS, and western blotting.

The cells were lysed in PBND lysis buffer (10 mM Tris-HCl, 50 mM KCl, 2.5 mM MgCl₂, 0.1 mg/ml gelatin, 0.45% NP-40, and 0.45% Tween 20, pH 8.3) containing proteinase K (200 μg/ml) for 1 h at 55°C followed by proteinase K inactivation for 10 min at 95°C. The target regions were amplified by PCR with either Taq-HP DNA Polymerase (Biospecifika Novosibirsk, Russia) or LongAmp Taq DNA Polymerase (NEB) (primers sequences could be found in Supplementary Table S3). The parameters were as follows: 95°C for 30 s, then 35 cycles of 94°C for 10 s, 60°C for 30 s, 72°C for 1 min (1 min/kb), and a final step at 72°C for 5 min for Taq-HP DNA Polymerase and 94°C for 30 s, then 30 cycles of 94°C for 10 s, 60°C for 30 s, 65°C for 3–6 min (1 min/kb), and a final step at 65°C for 5 min for LongAmp Taq DNA Polymerase. The amplified products were analyzed by agarose gel electrophoresis.

The cells were washed twice with PBS and scrapped from the surface in the presence of RIPA buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, and 0.1% SDS) containing the protease inhibitor cocktail [1x Complete ULTRA (Roche), 1x PhosSTOP (Roche), 5 mM NaF (Sigma)]. After that, the cells were sonicated by three 10 s pulses at 33–35% power settings. Lysates were centrifuged at 18,000 g for 20 min at 2°C, frozen, and stored at -80°C. The protein concentrations in cell lysates were quantified using Pierce BCA Protein Assay Kit (Thermo Fisher). Equal amounts (25 μg) of total protein were separated on 10% SDS-PAGE and then transferred onto the Immun-Blot PVDF membrane (Bio-Rad). After blocking in 5% milk/TBST for 2 h, the membrane was incubated with primary antibodies against RAD21 and SMC2 (#12673/#8720, Cell Signaling Technology) at 4°C overnight. On the following day, membranes were incubated with horseradish peroxidase–conjugated secondary antibodies (#7074, Cell Signaling Technology) for 2 h at 25°C. Detection was performed with SuperSignal West Pico PLUS Chemiluminescent Substrate (#34580, Thermo Fisher) and ChemiDoc XRS Imaging System (Bio-Rad).

GraphPad Prism 6.0 was used to design graphs and analyze quantitative data. The results are presented as mean values ± standard deviations.