Targeted protein degradation by Trim-Away using cell resealing coupled with microscopic image-based quantitative analysis

“Trim-Away” technology enables rapid degradation of endogenous proteins without prior modification of protein-coding genes or mRNAs through delivery of antibodies that target proteins of interest. Although this approach can be readily applied to almost any cytosolic protein, strategies for cytosolic antibody delivery have been limited to microinjection or electroporation, which require skill-dependent operation or specialized equipment. Thus, the development of antibody delivery methods that are convenient, scalable, and preferably do not require detachment of adherent cells is required to extend the versatility of the Trim-Away method. Here, we developed a cell resealing technique optimized for Trim-Away degradation, which uses the pore-forming toxin streptolysin O (SLO) to permeabilize the cell membrane and delivered the antibodies of interest into HEK293T, HeLa, and HK-2 cell lines. We demonstrated the ability of Trim-Away protein degradation using IKKα and mTOR as targets, and we showed the availability of the developed system in antibody screening for the Trim-Away method. Furthermore, we effectively coupled Trim-Away with cyclic immunofluorescence and microscopic image-based analysis, which enables single-cell multiplexed imaging analysis. Taking advantage of this new analysis strategy, we were able to compensate for low signal-to-noise due to cell-to-cell variation, which occurs in the Trim-Away method because of the heterogenous contents of the introduced antibody, target protein, and TRIM21 in individual cells. Therefore, the reported cell resealing technique coupled with microscopic image analysis enables Trim-Away users to elucidate target protein function and the effects of target protein degradation on various cellular functions in a more quantitative and precise manner.


Fig. S2 Evaluation of the targeting ability of resealed antibodies.
(A) Process for evaluating the targeting ability of resealed antibodies. HeLa cells expressing CD63-GFP or LC3-GFP were resealed with anti-GFP rabbit antibody and subjected to immunofluorescence.
Anti-rabbit IgG secondary antibody labeled with Alexa Fluor 546 was used to detect the resealed anti-GFP antibody. Anti-CD63 or LC3 mouse antibody was used for immunofluorescence; it was detected with anti-mouse IgG secondary antibody labeled with Alexa Fluor 647.
(B) Confocal microscopy images showing the targeting ability of resealed anti-GFP antibodies. Resealed anti-GFP antibodies (orange) showed punctate patterns and colocalization with GFP-tagged proteins (green). Immunofluorescence using antibodies against the exogenously expressed proteins (red) also showed colocalization of the two protein types. Normal rabbit IgG was used as a nontargeting negative control ("NC Rb IgG"), showing a diffused pattern. Scale bar: 10 μm.
(C) Quantifications of the colocalization level of images (B). Pearson's correlation coefficient between the indicated channels was calculated using the pixel intensity of the cytosol region for each cell. Each dot represents the value of one cell, and the bar indicates the mean ± standard error of the mean. Anti-GFP antibody (a-GFP Ab) displayed higher colocalization with GFP fused proteins (GFP-CD63 or GFP-LC3) as well as with the target proteins detected via immunofluorescence (CD63(IF) or LC3(IF)) compared with negative control antibodies (NC Rb IgG). Statistical significance of differences between the two groups was detected using Welch's t-test. **P < 0.01, ***P < 0.001. (A) Evaluation of the retention of resealed antibodies. Microscopy images of HEK293T cells resealed with 0.500, 0.250, and 0.125 mg/mL of fluorescently-labeled antibodies; cells were imaged 1, 3, 6, 12, and 24 h after resealing. Scale bar: 200 μm. Line graph (right) shows the mean intensity of Alexa Fluor 546-labeled antibody, which was quantified per frame. The background fluorescence quantified from the 0 mg/mL antibody resealing condition was subtracted to obtain these results. The mean intensity of the fluorescently-labeled antibodies decreased immediately after resealing (~3 h), presumably because a fraction of highly fluorescent cells that were excessively permeabilized induced cell death and detached within the first 3 h. However, fluorescence intensity was well-retained across the subsequent time course (3-24 h), with almost half of the antibodies being retained in the cells at 24 h after resealing. Data represent results from 15 frames (n = 15), expressed as mean ± SEM.
(B) Microscopy images of HEK293T cells simultaneously resealed with three different antibodies (0.2 mg/mL of each): anti-rat Alexa Fluor 488 goat IgG, anti-mouse Alexa Fluor 546 goat IgG, and anti-rabbit Alexa Fluor 647 goat IgG. Scale bar: 2,000 μm (Well view); 200 μm (Field view); 20 μm (Enlarged view). Pairwise plots of fluorescent intensities of the three different antibodies quantified per cell (right) showed high correlation between the three antibodies, indicating that although the permeabilization efficiency differed among cells, the ratio of the introduced antibodies was constant. Rb: rabbit, Ms: mouse. To evaluate the membrane resealing efficiency, the cells were incubated with propidium iodide (PI) and Hoechst for 30 min immediately after the resealing step (0 h) or 3 h after resealing. Strong PI staining indicated dead cells (denoted by blue arrowheads), and weaker staining indicated cells with inefficient resealing (denoted by white arrowheads). Cells with inefficient resealing were rarely observed (approximately 1%) in the three cell lines, and most cells were PI-negative immediately after resealing. We could not accurately quantify the percentage of cells with incomplete staining as dead cells were strongly stained with PI, which may interfere with quantification. At 3 h after resealing, we did not detect any cells with inefficient resealing (PI-positive cells). For the negative control, intact cells were incubated with PI and Hoechst for 30 min. For the positive control, the cells were incubated with PI when streptolysin O pores were present ("semi-intact cells") during the 5-min incubation step in TB buffer at 37°C. Dextran (10 kDa)-AF647 was used as the indicator of resealed cells. Scale bar: 20 μm.  TRIM21-overexpressed Piranha HEK293T cells were subjected to IKKα Trim-Away by resealing with anti-IKKα antibodies, followed by incubation for 3, 12, 24, or 48 h, and cells were then collected for western blotting. IKKα knockdown efficiency was the highest at 12 h after resealing. IKKα levels recovered to some extent after 24 h of incubation, which occurred simultaneously with the decrease in resealed antibody levels. Normal rabbit antibody levels also exhibited a time-dependent decrease, indicating that the resealed antibodies were degraded regardless of their target-binding ability. This result was different from the findings in Fig. S3A, in which fluorescently labeled antibodies were retained at least 24 h after resealing. This difference may be attributable to fluorescence labeling, difference in the antibody itself, or difference in the evaluation method. NC: negative control, Rb: rabbit. Data expressed as mean ± SEM (n = 3). *P < 0.05, **P < 0.01.

Fig. S8 MG132 treatment inhibited Trim-Away-mediated target protein degradation.
Proteasomal degradation was inhibited by treating Piranha HEK293T cells with MG132. The cells were pretreated with 25 μM MG132 for 1-3 h and then subjected to resealing procedures. MG132 was supplemented during the resealing steps and during incubation after resealing. The cells were collected 1-3 h after resealing for western blotting. Data are expressed as mean ± SEM (n = 4). **P < 0.01, ns: not significant. NC: negative control.

Fig. S9 Resealing of negative control antibody does not alter the levels of IKKα and mTOR.
Piranha HEK293T, HeLa, and HK-2 cells were resealed with 0.5 mg/mL normal rabbit antibodies and collected 3 h after resealing for western blotting. For mTOR blots of HK-2 cell lysates, an extra band was observed, which likely represented mTOR that underwent glycosylation or another posttranslational modification. Data are expressed as mean ± SEM (n = 3). ns: not significant. NC: negative control. HEK293T or TRIM21-overexpressing Piranha HEK293T cells were incubated for 3 h after resealing with 0.25 mg/mL anti-IKKα antibodies or the same level of normal rabbit antibody as a nontargeting negative control (NC). The degradation efficiency of IKKα was normalized to each NC, and the difference between the cells was compared. TRIM21-overexpressing Piranha HEK293T cells exhibited higher IKKα degradation efficiency than normal HEK293T cells. Data are expressed as mean ± SEM (n = 3). **P < 0.01. NC: negative control.

Fig. S11 Trim-Away of IKKα in HEK293T cells with TRIM21 knockdown (KD).
(A) Efficiency of siRNA-mediated TRIM21 KD was tested in HEK293T cells. Cells were collected at 24, 48, or 72 h after transfection and assessed via western blotting. The estimated endogenous TRIM21 band is indicated by the black arrow, and extra bands are indicated by asterisks. T21: TRIM21.
(B) HEK293T cells were transfected with TRIM21 siRNA or negative control siRNA, incubated for 48 h, and subjected to IKKα Trim-Away. Antibody delivery via cell resealing was performed using the standard resealing procedure except that the cytosol used in the resealing step was obtained from TRIM21 KD HEK293T cells. The cells were collected 3 h after resealing for western blotting. The degradation efficiency of IKKα was normalized to each negative control (NC Ab), and the difference between TRIM21 KD and negative control cells was analyzed. IKKα degradation was inhibited in TRIM21 KD cells compared with negative control cells (NC siRNA). Data are expressed as mean ± SEM (n = 4). **P < 0.01, ***P < 0.001, ns: not significant. NC: negative control.

Fig. S12 Cell resealing-mediated Trim-Away of IKKα in HK-2 and HeLa cells.
Western blotting of HK-2 and HeLa cells incubated for 3 h after resealing with 0.25 mg/mL of anti-IKKα antibody (clone Y463, abcam, Cat# ab169743) or the same amount of normal rabbit antibody as a negative control (NC).
(B) In HeLa cells, resealing of anti-IKKα antibody (clone Y463, abcam, Cat# ab169743) led to a 58% reduction in IKKα protein level; IKKβ protein level slightly decreased. Phosphorylation of NF-ĸB p65 was reduced under anti-IKKα-resealed conditions. Data expressed as mean ± SEM (n = 3). *P < 0.05, ***P < 0.001.  HEK293T cells were resealed with normal rabbit IgG, after which they were incubated for 1 h and fixed for analysis. Compared to the cells resealed with antibody alone (left panel), cells resealed simultaneously with antibody and TRIM21 protein showed a decreased amount of introduced antibody. Scale bar: 50 μm. Rb: rabbit.

Fig. 3C).
Western blotting bands presented in Fig. 3C were quantified, and the degradation efficiency was compared between conditions with or without exogenous TRIM21 ("His-Lipoyl-TRIM21"). Anti-IKKα antibody-resealed conditions were normalized to those of their respective negative controls (conditions resealed with normal rabbit antibody). Data are expressed as mean ± SEM (n = 4). *P < 0.05.

Fig. S16
The presence of cytosol in the resealing step enhances resealing efficiency.
(A) Piranha TM HEK293T cells were resealed with normal rabbit IgG in the presence or absence of L5178Y cytosol in the antibody introduction step and the resealing step. Cells resealed in the presence of L5178Y cytosol showed an enhanced retention rate of the introduced antibody (left panel) compared to the cells resealed in the absence of cytosol (right panel). Cells resealed in the absence of cytosol showed higher susceptibility to cell death as indicated by decreased cell number and nuclear shrinkage (white arrowhead), likely due to inefficient resealing. The mean intensity was quantified per cell (n = 2945, 2525). Scale bar: 50 μm. Rb: rabbit.
(B) HeLa cells were resealed with normal rabbit IgG in the presence or absence of L5178Y cytosol in the antibody introduction step and/or the resealing step. The presence of cytosol during the resealing step enhanced the retention rate of the introduced antibody (comparing the right and middle panels), whereas the presence of cytosol during the antibody introduction step did not further enhance the retention rate of the introduced antibody (comparing the middle and left panels). Scale bar: 50 μm. The mean intensity per cell was quantified (n = 808, 997, 740). Scale bar: 50 μm. Rb: rabbit.

Fig. S18 Comparison of the antibody resealing efficiency (Related to
Comparison of the antibody resealing efficiency between cells that were incubated without cytosol (TB), with L5178Y cytosol, or with cytosol obtained by the new freeze-thaw protocol, during the resealing step. HEK293T cells or HeLa cells were resealed with normal rabbit IgG, after which they were incubated for 1 h and fixed for analysis. The images shown in Fig.3D were quantified per cell (n = 10138, 10445, 10538 for HEK293T cells, n=2322, 2888, 2638 for HeLa cells).

Fig. S19 mTOR inhibition by Torin led to decreased phosphorylation of Akt and S6rp.
Piranha HEK293T cells were treated with 25 ("low") or 250 nM ("high") Torin for 3 h and collected for western blotting. Data are expressed as mean ± SEM (n = 3). *P < 0.05, ***P < 0.001. The bar graph presents the quantification of the bands in Fig. 5B. The total levels of Akt and S6rp were not altered in the mTOR Trim-Away experiment. Data are expressed as mean ± SEM (n = 3). ns: not significant. Rb: rabbit, Ab: antibody.

Fig. S21 Additional expression of exogenous TRIM21 is required for efficient degradation of mTOR.
The degradation efficiency of mTOR in TRIM21-overexpressed Piranha TM HEK293T cells was compared with that in normal HEK293T cells. Cells were resealed with 0.5 mg/mL of anti-mTOR antibody (clone 7C10, CST, Cat#2983BF) or the same amount of normal rabbit antibody as a negative control (NC), after which they were incubated for 12 h and collected for western blotting analysis. Endogenous TRIM21 was not sufficient to degrade a significant amount of mTOR in HEK293T cells, whereas cells with additional expression of exogenous TRIM21 showed significant mTOR degradation.
Data expressed as mean ± SEM (n = 3). *P < 0.05, ns: not significant. NC: negative control.  Single-cell multidimensional data containing the mean intensities of the stained proteins [obtained from the mTOR Trim-Away experiment (see Fig. 6B)] were plotted. The upper panels show correlations between variables (Corr: correlation of both NCRb and mTOR conditions combined, NCRb: correlation of negative control condition, mTOR: correlation of anti-mTOR antibody resealing condition), the lower panels show scatter plots of the variables, and the diagonal shows the density plots of the variables. Histograms of fluorescence intensity indicate that the anti-mTOR antibody resealing condition (color-coded pink) had lower mTOR and pAkt intensities compared with those of the negative control (color-coded blue). Scatterplots showing the fluorescence of resealed anti-mTOR antibodies vs. immunofluorescence-stained mTOR fluorescence indicate a weak negative correlation between the variables. Similarly, a weak negative correlation existed between pAkt and the resealed anti-mTOR antibody. RbAb: rabbit antibody (Normal rabbit IgG or anti-mTOR antibody), NC: negative control, Rb: rabbit, Corr: correlation. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. S24 Single-cell analysis of adherent electroporation-mediated Trim-Away targeting mTOR.
The cyclic immunofluorescence analysis framework described in Fig. 6A was applied for single-cell analysis of adherent electroporation-mediated Trim-Away targeting mTOR. Anti-mTOR antibodies (clone 7C10, CST, Cat#2983BF) were introduced into TRIM21-overexpressed Piranha TM HEK293T cells grown in 96-well plates using adherent electroporation; the cells were then incubated for 12 h and fixed for cyclic immunofluorescence. Images were acquired from the well center to obtain images of well-electroporated cells. After repeated rounds of immunofluorescence staining, imaging, and bleaching, the resultant images were aligned and analyzed. RbAb: rabbit antibody (Normal rabbit IgG or anti-mTOR antibody), NC: negative control, Rb: rabbit.
(A) Pairwise plots of multivariate data obtained from the adherent electroporation-mediated mTOR Trim-Away experiment. The upper panels show correlations between variables (Corr: correlation of both NCRb and mTOR conditions combined, NCRb: correlation of negative control condition, mTOR: correlation of anti-mTOR antibody resealing condition), the lower panels show scatter plots of the variables, and the diagonal shows the density plots of the variables. Histograms of fluorescence intensities indicate that anti-mTOR antibody-electroporated conditions (color-coded pink) had lower mTOR and pAkt intensities compared with those in the negative control (color-coded blue).
Scatterplots showing the fluorescence of anti-mTOR antibody vs. immunofluorescence-stained mTOR fluorescence show a weak negative correlation between the variables. A weak negative correlation was also detected between pAkt and the introduced anti-mTOR antibody. Corr: correlation. *P < 0.05, **P < 0.01, ***P < 0.001.
(B) Density plots showing mTOR intensity, pAkt/Akt, and pS6rp/S6rp of cells electroporated with mTOR or normal rabbit IgG (NCRb). The mean intensities of the phosphorylated forms were normalized against the total level of the respective proteins, represented as pAkt/Akt and pS6/S6rp.
Single-cell data were filtered by the mean fluorescence intensity of the introduced antibody. Differences in the distribution between anti-mTOR antibody-electroporated cells and the negative control became more pronounced as the data were narrowed down to the most efficiently electroporated cells; thus, mTOR was efficiently degraded and the downstream targets were inactivated by the introduction of anti-mTOR antibody. AU: arbitrary unit.