Highlights
•Trim-Away is a widely applicable method to degrade endogenous proteins
•Target proteins do not need to be modified before degradation
•Proteins are degraded within minutes of application
•Trim-Away allows efficient protein depletion in primary human cells
Summary
Methods for the targeted disruption of protein function have revolutionized science and greatly expedited the systematic characterization of genes. Two main approaches are currently used to disrupt protein function: DNA knockout and RNA interference, which act at the genome and mRNA level, respectively. A method that directly alters endogenous protein levels is currently not available. Here, we present Trim-Away, a technique to degrade endogenous proteins acutely in mammalian cells without prior modification of the genome or mRNA. Trim-Away harnesses the cellular protein degradation machinery to remove unmodified native proteins within minutes of application. This rapidity minimizes the risk that phenotypes are compensated and that secondary, non-specific defects accumulate over time. Because Trim-Away utilizes antibodies, it can be applied to a wide range of target proteins using off-the-shelf reagents. Trim-Away allows the study of protein function in diverse cell types, including non-dividing primary cells where genome- and RNA-targeting methods are limited.
Introduction
Interfering with protein expression is a powerful strategy to investigate the function of a protein. Traditionally, DNA-modifying methods have been used to knockout proteins on the gene level (Capecchi, 1989), an approach that has had a recent resurgence with the emergence of CRISPR/Cas9 technology (Doudna and Charpentier, 2014). RNA-targeting methods such as RNAi are also widely used to knockdown expression of a protein by destroying the mRNA (Elbashir et al., 2001). However, in both approaches, protein depletion is indirect and dependent on the inherent turnover of the protein. Consequently, long-lived proteins take more time to deplete, or may be resistant to DNA- and RNA-targeting depletion methods altogether (Jansen et al., 2007, Smoak et al., 2016). The long time frame that is required for protein depletion with these methods also means that cells may have enough time to activate compensatory mechanisms, which may mask phenotypes (Damke et al., 1995, Rossi et al., 2015). Furthermore, it is often impossible to determine whether or not a phenotype is an indirect consequence of an earlier defect. In an attempt to overcome these problems, several methods have been developed for conditional protein inactivation. However, these methods require that either the protein of interest is first modified (Banaszynski et al., 2006, Caussinus et al., 2011, Neklesa et al., 2011, Nishimura et al., 2009, Robinson et al., 2010, Tachibana-Konwalski et al., 2010) or can only be applied to a very small number of proteins (Deshaies, 2015, Fan et al., 2014). Chemical inhibitors can be equally problematic as they are limited to druggable proteins and prone to off-target effects (Mayer, 2003). A widely applicable protein depletion method that acts exclusively at the protein level is currently lacking. Such a method would not only allow the acute depletion of endogenous proteins, but also the study of protein function in non-dividing primary cells in which DNA-targeting must be done at the whole animal level and RNA-targeting cannot deplete stable proteins.
We sought to develop a truly posttranslational protein depletion method based on protein targeting by antibodies. Antibodies bind to proteins with high affinity and specificity; they are widely available commercially and can be produced for almost any protein with relative ease. Antibodies are therefore ideal as the basis of a protein-targeting method. Antibodies have been used previously to interfere with protein function (Morgan and Roth, 1988). However, this requires that the antibody binds to an epitope that blocks the function of a protein and can effectively and stoichiometrically compete with endogenous ligands. This method of inhibition is therefore only applicable to a very limited number of proteins. Instead, we aimed to develop a universally applicable method that would allow us to target any antibody-bound protein for degradation.
Antibody-bound pathogens can be recognized by the cytosolic antibody receptor, TRIM21 (Mallery et al., 2010). TRIM21 is an E3 ubiquitin ligase that binds with high affinity to the Fc domain of antibodies (James et al., 2007). TRIM21 is widely expressed in diverse cell types and tissues, which is a necessary requirement of its physiological role (Yoshimi et al., 2009). During infection, TRIM21 recruits the ubiquitin-proteasome system to antibody-bound pathogens, leading to their destruction (Mallery et al., 2010). TRIM21 causes degradation of diverse pathogens including RNA and DNA viruses (Watkinson et al., 2015), bacteria (McEwan et al., 2013), and the proteopathic agent Tau (McEwan et al., 2017). Both the proteasome and the AAA ATPase VCP/p97 have been identified as important co-factors for degradation, but their requirement differs between substrates (Hauler et al., 2012).
The E2 enzymes Ube2W and Ube2N/2V2 have also been implicated in TRIM21-mediated degradation (Fletcher et al., 2015). These enzymes act sequentially on TRIM21 itself, first mono-ubiquitinating it (Ube2W) and then extending from this a K63-ubiquitin chain (Ube2N/2V2). TRIM21 is also modified with K48-chains but the E2 enzyme(s) involved are unknown, as is the functional importance of this modification. TRIM21 has not been reported to modify the pathogen or the pathogen bound antibodies, but it is unclear if such modifications have escaped detection or if auto-ubiquitination of TRIM21 is sufficient for pathogen destruction. TRIM21 has also been reported to interact with several other proteins including Skp2, DAXX, IRF-3, IRF-5, IRF-8, DDX41, and SQSTM1 (James, 2014). However, most of these interactions have been detected based on immunoprecipitation, which can deliver false positives because of direct binding of TRIM21 to the antibodies used in the assay.
In this study, we repurposed TRIM21 to establish a method to degrade endogenous proteins that we called Trim-Away. Trim-Away allows the study of protein function in various cell types, including non-dividing primary cells where genome- and RNA-targeting methods are not well suited. Trim-Away degrades proteins within minutes of application, making it suitable to investigate the function of a protein long after it has formed and at all stages of a cell’s life, cycle, or differentiation. Acute protein disruption assays, which in the past have required complex genetics and a lot of time, can now be done by Trim-Away within hours. Trim-Away therefore forms the basis for the systematic stage-specific analysis of protein function.
Results
Principle of Protein Degradation by Trim-Away
We reasoned that the antibody receptor and ubiquitin ligase TRIM21 could be used as a tool to drive the degradation of endogenous proteins by using a 3-step strategy: first, the introduction of exogenous TRIM21; second, the introduction of an antibody against the protein of interest; and third, TRIM21-mediated ubiquitination followed by degradation of the antibody-bound protein of interest (Figure 1A). As we outline in detail below, this strategy, which we called “Trim-Away,” is ideally suited for the acute and rapid degradation of endogenous proteins in both individual cells as well as bulk cell populations.
To test the Trim-Away strategy, we first set up a proof of principle experiment in mammalian cell culture. NIH 3T3 cells overexpressing mCherry-TRIM21 and free GFP were microinjected with anti-GFP antibody (Figure 1B). Remarkably, GFP was rapidly degraded following microinjection with a half-life of just 16 min (Figures 1B, 1C, and S3A; Movie S1). mCherry-TRIM21 colocalized with GFP during degradation, consistent with TRIM21 recruitment to GFP via the anti-GFP antibody (Figures S1A and S1B; Movie S1). GFP aggregated quickly at the site of antibody microinjection, likely because of the high local concentration of anti-GFP antibody (Figures 1B and S1A). Degradation was specifically due to targeting of GFP by anti-GFP antibody, because microinjection of a non-specific control IgG did not cause GFP degradation in mCherry-TRIM21-overexpressing cells (Figures 1B, 1C, and S1A–S1D). The degradation of GFP was dependent on TRIM21 overexpression, because anti-GFP antibody failed to trigger GFP degradation in cells that overexpressed mCherry instead of mCherry-TRIM21 (Figures S1E and S1F). The E3 ubiquitin ligase activity of TRIM21 was required for degradation, because a truncated form of TRIM21 lacking the RING E3 ubiquitin ligase and B-Box domains (mCherry-TRIM21ΔRING-Box) efficiently colocalized with GFP following anti-GFP antibody microinjection, but failed to cause GFP degradation (Figures S1G and S1H).
To test if protein degradation by Trim-Away is mediated by the proteasome consistent with the known mechanism of TRIM21 function, we treated cells with the proteasome inhibitor MG132. Strikingly, MG132 treatment prevented the degradation of GFP following anti-GFP antibody microinjection into mCherry-TRIM21-overexpressing cells (Figures 1D and 1E). Therefore, protein degradation by Trim-Away relies on the ubiquitin-proteasome pathway (Figure 1A).
Trim-Away Can Be Used to Degrade Proteins in Primary Cells
We next asked if Trim-Away could be applied to post-mitotic primary cells. We chose mammalian oocytes because they are transcriptionally silent, which precludes protein disruption by direct genome editing. Moreover, RNAi is inefficient in these cells due to large amounts of stored proteins (Clift and Schuh, 2013, Pfender et al., 2015). We first tested if TRIM21 overexpression has any influence on oocyte meiosis. Oocytes that overexpressed TRIM21 progressed through meiosis with similar efficiency and timing as control oocytes: neither the rate nor the timing of nuclear envelope breakdown or anaphase was significantly different (Figures S6A–S6C, S6E, and S6F). Also spindle morphology during meiosis I and meiosis II was not perturbed (Figures S6A, S6D, and S6G). Thus, TRIM21 overexpression does not perturb oocyte meiosis.
We then performed a similar proof-of-principle GFP degradation experiment as described above in isolated mouse oocytes. Microinjected anti-GFP antibody triggered rapid GFP degradation in oocytes overexpressing mCherry-TRIM21. In contrast, control IgG had no effect on GFP protein levels (Figures 2A and 2B ). As in NIH 3T3 cells, mCherry-TRIM21 colocalized with GFP during GFP degradation (Figure S2A), and GFP degradation was dependent on TRIM21’s ubiquitin ligase activity (Figures S2C–S2H). Before protein degradation, we observed a transient increase in GFP intensity at the site of GFP-antibody microinjection (Figure 2A). This is likely due to a local enrichment of GFP-antibody at the microinjection site, which results in sequestration of GFP from the entire oocyte volume to this region.
We noticed that mCherry-TRIM21 was depleted concomitantly with GFP upon anti-GFP antibody microinjection (Figures S2A and S2B). Western blotting of whole cell extracts confirmed that TRIM21, antibody and target protein are all degraded during Trim-Away (Figure S2I). This is consistent with the proposed mechanism of TRIM21 function and suggests that TRIM21 and antibody levels could become limiting during Trim-Away if the target is in significant molar excess. To investigate this further, we modified relative levels of TRIM21 and GFP and found that TRIM21 needs to be present in excess of the target protein to facilitate complete protein degradation (Figures S2J–S2L).
Trim-Away Can Target Diverse Cellular Substrates
A widely applicable protein depletion method should be able to target diverse substrates within the cell. To address this, we localized GFP to different regions of the cell and tested the efficiency of degradation by Trim-Away. GFP containing an N-terminal N-myristoyl and S-palmitoyl motif (membrane-anchored GFP) localized to the plasma membrane and vesicle-like structures in the cytoplasm (Figure 2C); GFP fused to the histone H2B (H2B-GFP) was efficiently incorporated into chromatin (Figure 2E); GFP containing a nuclear localization signal (NLS-GFP) accumulated in the nucleus (Figure 2G). All three different substrates were rapidly degraded by anti-GFP antibody and TRIM21 with similar kinetics to free cytosolic GFP (Figures 2C–2H). Remarkably, the half-life for degradation by Trim-Away was as little as 9 min (Figures 2B, 2D, 2F, 2H and S3A). This is identical to the rate of protein degradation achieved by the auxin-inducible degron system (Holland et al., 2012), but with the advantage that Trim-Away requires no modification of the target protein. Altogether, these data show that Trim-Away can degrade proteins that are localized to different regions of the cell and incorporated into larger protein complexes and cellular structures.
We noticed that NLS-GFP was degraded in the cytoplasm, presumably because NLS-GFP shuttles in and out of the nucleus and can therefore be bound by antibodies in the cytoplasm (Figure 2G). We therefore tested if Trim-Away can also degrade retained nuclear proteins, such as GFP-H2B that is stably associated with chromatin within the nucleus. In contrast to NLS-GFP, GFP-H2B was not degraded when the chromatin was contained within an intact nucleus (Figures S3B and S3C). We reasoned that we might be able to overcome this limitation by fusing the Fc-domain of an antibody to a nanobody. The Fc-nanobody fusion is much smaller in size and should be able to enter the nucleus. Strikingly, H2B-GFP was quickly degraded also inside the nucleus when TRIM21 and the Fc-nanobody fusion were co-expressed (Figures 2I–2K). This demonstrates that Fc-nanobody fusions can be used to degrade proteins inside the nucleus. It also illustrates that the growing number of nanobodies is compatible with Trim-Away when these nanobodies are fused with an Fc-domain.
Rescue Experiments Confirm Trim-Away Specificity
In principle, it should be possible to degrade any endogenous protein in the cell that can be accessed by antibodies using Trim-Away (Figure 1A). To test this, we decided to target an endogenous protein in mouse oocytes called Eg5. Eg5 is a microtubule motor protein required for proper spindle assembly during mitosis and meiosis (Clift and Schuh, 2015, Mayer et al., 1999, Schuh and Ellenberg, 2007). When the function of Eg5 is disrupted, a monopolar microtubule aster forms and the spindle fails to become bipolar. We chose to Trim-Away Eg5 because we could directly compare the phenotype following Eg5 degradation to that of Eg5 inhibition with the small molecule inhibitor monastrol (Mayer et al., 1999).
Strikingly, microinjection of anti-Eg5 antibody into oocytes overexpressing TRIM21 caused the formation of monopolar spindles; precisely the phenotype expected if Eg5 is degraded and identical to oocytes treated with the Eg5 inhibitor monastrol (Figures 3A–3C ). Neither control IgG microinjection into TRIM21-overexpressing oocytes, nor anti-Eg5 antibody microinjection alone caused monopolar spindles (Figures 3B and 3C), confirming that Eg5 degradation requires both TRIM21 and anti-Eg5 antibody. Eg5 protein was completely degraded as shown by immunoblotting with two different anti-Eg5 antibodies (Figure 3D).
To confirm that the monopolar spindle phenotype was due to Eg5 degradation, we performed a rescue experiment by microinjecting mRNA encoding Eg5-mEGFP into Eg5 Trim-Away oocytes (Figures 3A and 3E). Eg5-mEGFP expression transformed the monopolar spindles back to a bipolar state (Figures 3E and 3F; Movie S2). Eg5-mEGFP can be recognized by anti-Eg5 antibody, but likely rescues because it is expressed in excess. Altogether, these data show that endogenous proteins can be degraded using Trim-Away, and rescue assays can be used to confirm the specificity of a Trim-Away phenotype.
Trim-Away Is Suitable to Degrade Long-Lived Proteins Acutely
Long-lived intracellular proteins are surprisingly common (Toyama et al., 2013), but studying their function is challenging. The lack of protein turnover means that depletion by RNAi is ineffective. Gene knockouts are also often not suitable, because long-lived proteins are frequently especially important in non-dividing primary cells and essential for viability (Toyama et al., 2013). We reasoned that because Trim-Away acts exclusively at the protein level, it should be suitable to degrade even very long-lived proteins and thereby study their function long after their synthesis.
To test this, we targeted the long-lived Rec8 protein in mouse eggs that were arrested in metaphase of the second meiotic division. Rec8 is part of the cohesin protein complex that mediates sister chromatid cohesion in oocytes from birth until ovulation (Tachibana-Konwalski et al., 2010). Rec8 does not turnover (Tachibana-Konwalski et al., 2010), but remains stably associated with chromosomes for months in mice and possibly decades in humans.
We asked if we could use Trim-Away to degrade endogenous Rec8 acutely in metaphase II arrested eggs, long after it has been incorporated into chromosomes. Microinjection of anti-Rec8 antibody (Eijpe et al., 2003) into eggs overexpressing mEGFP-TRIM21 triggered the premature separation of sister chromatids (Figures 4A–4C ; Movie S3), producing 40 single chromatids (Figure 4D), which is indicative of Rec8 degradation and complete loss of sister chromatid cohesion (Tachibana-Konwalski et al., 2010). No separation was observed when control IgG was microinjected into mEGFP-TRIM21-overexpressing oocytes or when oocytes overexpressing antibody-binding-deficient TRIM21 (mEGFP-TRIM21ΔPRYSPRY) were microinjected with anti-Rec8 antibody (Figures 4B and 4C). This confirmed that sister chromatid separation was due to Rec8 degradation triggered by TRIM21 recruitment to Rec8 via anti-Rec8 antibody (Figure 4A). Previous functional studies of Rec8 required complex mouse genetics to replace the Rec8 gene with a version that expresses a TEV protease-cleavable Rec8 protein (Tachibana-Konwalski et al., 2010). The Rec8 Trim-Away experiments described here provide the first evidence that endogenous, unmodified Rec8 protein is responsible for sister chromatid cohesion in mouse eggs. Remarkably, sister chromatids began to separate on average just 11 min after anti-Rec8 antibody microinjection into mEGFP-TRIM21-overexpressing eggs (Figures 4B and 4E), implying Trim-Away can degrade even very long-lived proteins with unprecedented speed.
Trim-Away Can Be Used to Degrade Specific Protein Variants Selectively
Given that Trim-Away capitalizes on the high specificity of antibodies, it should also be suitable to selectively degrade splice variants, posttranslationally modified or disease-causing protein variants while preserving the healthy protein, as long as antibodies are available that are variant-specific. To test this, we asked if Trim-Away can be used to selectively degrade the disease-causing variant of the protein huntingtin, which causes Huntington’s disease. We co-expressed the normal or the disease-causing variant of huntingtin protein together with mCherry-TRIM21 in NIH 3T3 cells. We then microinjected the cells with an antibody (3B5H10) which specifically binds to the disease-causing variant of huntingtin (Miller et al., 2011) (Figure S4A). The disease-causing variant of huntingtin protein was rapidly degraded following microinjection of the antibody, whereas normal huntingtin protein was preserved (Figures S4A and S4B). Similar results were obtained in oocytes that expressed both the normal and the disease-causing variant of huntingtin simultaneously: the disease-causing variant was degraded, while the normal variant was preserved (Figures S4C and S4D). This demonstrates that Trim-Away can be used to degrade specific protein variants selectively.
Antibody Electroporation Allows Trim-Away in Bulk Cell Populations
While microinjection is suitable for studying individual cells, most assays require the manipulation of bulk cell populations. We therefore aimed to develop a method of antibody and TRIM21 delivery that would make Trim-Away applicable to large numbers of cells simultaneously. Recent work has shown that antibodies can be delivered into the cytoplasm by electroporation (Freund et al., 2013), suggesting that antibody electroporation could be used to apply Trim-Away to large cell numbers.
Thus, we sought to develop an optimized antibody electroporation protocol. To determine the most efficient electroporation conditions, we used fluorescently labeled antibodies and quantified the number of fluorescent cells upon electroporation (Figures 5A–5C ; STAR Methods). The optimized electroporation protocol resulted in highly efficient delivery of antibody into the cytoplasm without cell death (Figures 5A–5C and S5A; STAR Methods). This method of antibody delivery is also compatible with the rapid observation of phenotypes: live imaging revealed that adherent cells started to re-adhere within 30 min of electroporation, with adherence completed by 4 hr (Figures S5B and S5C; Movie S4). Electroporated cells showed no evidence of stress or damage and rapidly re-entered the cell cycle with similar division rates as non-electroporated cells (Figures S5D–S5F; Movie S4).
We also generated stable NIH 3T3 (mouse) and HEK293T (human) cell lines overexpressing mCherry-TRIM21. These cell lines proliferated at a similar rate to wild-type cells and behaved indistinguishably after multiple passages suggesting that TRIM21 overexpression is stable and nontoxic (Figures S6H–S6J; STAR Methods). We also analyzed the transcriptome of TRIM21-overexpressing and wild-type NIH 3T3 cells either with or without electroporation of a non-targeting antibody, BSA or PBS. The transcriptomes were closely related under all conditions, with the vast majority of genes being expressed at similar levels (Figures S6K–S6M; Tables S1, S2, and S3). Notably, transcripts encoding previously described ligands of TRIM21, IRF-3, IRF-5, Skp2, DAXX, DDX41, and SQSTM1 were not significantly increased or decreased upon overexpression of TRIM21 (Figures S6N–S6P). Consistent with this result, neither TRIM21 overexpression, nor Trim-Away of an endogenous protein caused changes in the protein level of IRF-3 (Figure S6Q). Together, these results suggest that cells overexpressing TRIM21 behave similar to wild-types.
We reasoned that by electroporating antibodies into TRIM21-overexpressing cell lines it should be possible to target endogenous proteins in bulk cell populations for degradation. To test this, we targeted two proteins that are non-essential for cell viability: ERK1, a redundant serine/threonine kinase in the MAPK/ERK signaling cascade (Frémin et al., 2015), and IKKα, a catalytic subunit of the IκB kinase (IKK) complex (Hu et al., 1999, Takeda et al., 1999). Strikingly, both ERK1 and IKKα were depleted within 1–2 hr of antibody electroporation into TRIM21-overexpressing cells, with depletion lasting for 3–4 days (Figures S5G and S5H). The reappearance of ERK1 and IKKα coincided with the depletion of antibody from the cells (Figure S5H). This indicates that antibody availability limits the duration of protein depletion and suggests that the depletion period could be modified by varying the amount of electroporated antibody.
We next tested if Trim-Away using antibody electroporation is compatible with quantitative analysis of cellular phenotypes. To this end, we targeted the centrosomal protein pericentrin in NIH 3T3 cells. Strikingly, electroporation of rabbit polyclonal anti-pericentrin antibody (ab4448) into mCherry-TRIM21-overexpressing cells led to a complete loss of pericentrin signal at the centrosome in ∼95% of cells (Figures 5D and 5E). The high efficiency of pericentrin loss likely reflects the fact that ∼95% of cells receive antibody following electroporation (Figures 5A and 5C) and ∼99% of cells overexpress TRIM21 (Figure S6H). Pericentrin degradation by Trim-Away was confirmed by immunoblotting with two different antibodies (Figure 5F). Pericentrin is proposed to have an important role in the localization of Cdk5rap2 to the centrosome (Chen et al., 2014, Lee and Rhee, 2011). However, previous studies relied on prolonged, indirect depletion of pericentrin by RNAi or gene knockout (Chen et al., 2014, Lee and Rhee, 2011). Thus, it has been impossible to distinguish between a role for pericentrin in Cdk5rap2 recruitment during the course of the centrosome cycle or in maintenance of Cdk5rap2 at the centrosome. We used Trim-Away to address this question and investigate how Cdk5rap2 localization is altered upon acute removal of pericentrin. Repeating Trim-Away and staining for Cdk5rap2 revealed that loss of pericentrin from the centrosome leads to a concomitant loss of centrosomal Cdk5rap2 (Figures 5G and 5H).
To further confirm the specificity of pericentrin degradation, we repeated Trim-Away with a second mouse monoclonal anti-pericentrin antibody (BD611815). Pericentrin Trim-Away with BD611815 also caused pericentrin degradation and Cdk5rap2 mislocalization (Figures 5F and S5I–S5M). Notably, pericentrin degradation and Cdk5rap2 mislocalization was observed just 3 hr post-antibody electroporation (Figures S5I–S5M), further demonstrating that pericentrin is required to maintain Cdk5rap2 at the centrosome. Together, this data shows that Trim-Away is suitable to rapidly degrade endogenous proteins and observe cellular phenotypes in bulk cell populations.
Trim-Away Can Rapidly Activate Signal Transduction Pathways through Selective Degradation
The above experiments show how Trim-Away can be used to investigate loss-of-function phenotypes. Next, we attempted to use Trim-Away to target proteins involved in active signaling pathways to induce both pathway inactivation and activation. First, we targeted the protein kinase mTOR as it regulates several essential cellular functions (Laplante and Sabatini, 2009) (Figure 6A). Electroporation of anti-mTOR antibody into HEK293T cells overexpressing mCherry-TRIM21 lead to a reduction in mTOR protein levels (Figure 6B). Degradation was not complete, perhaps because a subset of mTOR resides within intracellular compartments inaccessible to antibody (Betz and Hall, 2013). Nonetheless, mTOR degradation by Trim-Away was sufficient to reduce phosphorylation of ribosomal protein 6 at Ser235/236, consistent with a loss of mTORC1 activity and similar to mTORC1 inhibition by rapamycin (Figures 6A and 6B). Importantly, mTOR also functions in a second complex, mTORC2, that induces phosphorylation of Akt. However, this activity cannot be investigated by rapamycin as it is a poor inhibitor of the mTORC2 complex (Sarbassov et al., 2005) (Figure 6A). In contrast, mTORC2 function could be efficiently ablated using Trim-Away. Phosphorylation of Akt at Ser473 was lost upon mTOR degradation by Trim-Away, but only slightly reduced upon rapamycin treatment (Figure 6B).
Next, we attempted to use Trim-Away to induce a change in a transduction pathway that activates rather than inhibits downstream signaling. We chose the nuclear factor κB (NF-κB) pathway as this system is constitutively repressed by the inhibitor IκBα. Upon stimulation, IKK phosphorylates IκBα leading to IκBα ubiquitination and degradation, allowing nuclear translocation of the heterodimeric p65 and p50 complex and transcription of target genes (Gilmore, 2006). Strikingly, electroporation of anti-IκBα antibody into HEK293T cells overexpressing mCherry-TRIM21 triggered both the degradation of IκBα and the substantial induction of NF-κB activity (Figures 6D and 6E). Thus, acute degradation of IκBα by Trim-Away is sufficient to relieve inhibition of the NF-κB pathway and trigger downstream signaling. NF-κB activation following IκBα Trim-Away was not indirectly due to TRIM21 activation (McEwan et al., 2013), because Trim-Away of a control endogenous protein, mTOR, triggered only minor NF-κB activity (Figures 6D and 6E). Electroporation of anti-IκBα antibody alone triggered a small induction of NF-κB, although to a much lesser extent than in cells expressing mCherry-TRIM21 (Figures 6D and 6E). It is possible that this small induction by anti-IκBα antibody alone was due to minor disruption of the NF-κB-IκBα interaction, because IκBα protein was not degraded in this condition (Figure 6E).
Collectively, these data establish that Trim-Away can be used to perform rapid protein depletion in bulk cell populations upon electroporation of specific antibodies. Furthermore, Trim-Away allows specific gain-of-function and loss-of-function phenotypes to be observed within just a few hours.
Co-electroporation of TRIM21 Protein and Antibody Facilitate Rapid Protein Degradation in Unmodified Cell Lines
While Trim-Away performed efficiently when TRIM21 is ectopically overexpressed, this requires prior modification of a cell line. We therefore investigated whether TRIM21 could be supplied exogenously in protein form, eliminating the necessity for prior transfection or transduction. We envisaged that Trim-Away could be achieved in one step by co-electroporating antibody together with recombinant TRIM21 protein. To this end, we established a protocol to purify functional recombinant TRIM21 (STAR Methods). We tested this method by co-electroporating recombinant TRIM21 and antibodies against IKKα or ERK1 into six commonly used cell lines. Remarkably, in all cell lines, this led to the efficient degradation of IKKα or ERK1 (Figures 7A and S7A). Interestingly, in some cell lines, IKKα and ERK1 were at least partially degraded when antibody was electroporated without recombinant TRIM21 (Figures 7A and S7A). This degradation was linked to a marked decrease in the levels of endogenous TRIM21, suggesting that the endogenous TRIM21 was mediating protein degradation.
These data show that Trim-Away can be used to rapidly degrade endogenous proteins in unmodified cells by simply electroporating antibody and TRIM21 protein simultaneously. It also shows that some cell lines have sufficient endogenous active TRIM21 to mediate protein degradation, although this is likely to be dependent upon target expression levels.
Trim-Away by Endogenous TRIM21 in Human Primary Cells
The possibility to directly degrade endogenous proteins by Trim-Away is potentially transformative for studies of primary human cells, in which gene editing and RNAi are challenging. We therefore tested if we can use Trim-Away to degrade proteins in normal human lung fibroblasts (NHLFs), primary cells used to study respiratory infection. Co-electroporation of IKKα or ERK1 antibodies together with recombinant TRIM21 led to efficient protein degradation in these cells (Figures S7B and S7C). Even endogenous levels of TRIM21 were sufficient to mediate protein degradation, as suggested by concomitant degradation of TRIM21 together with the target protein upon antibody electroporation alone (Figures S7B and S7C).
Studying protein function in primary human immune cells has long represented a difficult challenge (Chow et al., 2011). This is particularly true for primary macrophages, as their active nucleotide sensing machinery makes classical plasmid and siRNA transfection approaches unfeasible and likely to indirectly induce stimulation (Hornung and Latz, 2010). Because Trim-Away acts exclusively at the protein level, this method offers a unique opportunity to investigate protein function in primary human macrophages without perturbing them. NLRP3 is an intracellular signaling molecule expressed in immune cells that triggers the formation of a multiprotein complex called the inflammasome in response to diverse pro-inflammatory stimuli. The inflammasome directs interleukin-1β (IL-1β) maturation and secretion as part of a linked inflammatory and cell death response called pyroptosis (Schroder and Tschopp, 2010). The requirement for the NLRP3 protein in inflammasome activation has so far only been shown in mouse macrophages (Sutterwala et al., 2006) and a human leukemia-derived THP-1 cell line (Papin et al., 2007). We used Trim-Away to test whether NLRP3 is required for inflammasome activation in ex vivo human monocyte-derived macrophages (HMDMs). Electroporation of anti-NLRP3 antibody into HMDMs triggered rapid depletion of NLRP3 (Figure 7C). This was likely TRIM21-dependent, because NLRP3 was degraded upon anti-NLRP3 antibody electroporation into macrophages derived from wild-type mice, but not in macrophages from TRIM21 knockouts (Figure 7D). We did not observe co-degradation of TRIM21 and anti-NLRP3 antibody (Figures 7C and 7D), presumably because they were present in large excess to endogenous NLRP3. Strikingly, Trim-Away of NLRP3 caused a reduction in IL1-β secretion in response to stimulation with lipopolysaccharides (LPS) and the pore-forming toxin Nigericin in HMDMs from all four blood donors (Figure 7E). Neither electroporation of an anti-GFP antibody nor Trim-Away of IKKα had any effect on IL1-β secretion (Figure 7E). These results demonstrate for the first time that NLRP3 has a nonredundant and crucial role in inflammasome activation in primary human macrophages. Moreover, taken together the above data demonstrate that Trim-Away is suitable to degrade proteins in a wide range of cell types, including primary human cells that are intractable to other methods.
Specificity of Trim-Away Assays
The successful application of Trim-Away to 9 different endogenous proteins in 10 different cell types demonstrates that the Trim-Away method is widely applicable and cell and substrate independent. Importantly, Trim-Away did not lead to the degradation of proteins in close spatial proximity of the target proteins: Trim-Away of IκBα did not cause degradation of NF-κB (Figures 6D and 6E), Trim-Away of individual nucleoporins did not lead to degradation of the entire nuclear pore complex (Figure S7D), Trim-Away of pericentrin did not cause degradation of other centrosomal proteins (Figures 5F and S5M), and Trim-Away of H2B-GFP did not cause degradation of H2A (Figures S3D–S3G). However, the fate of multiprotein complexes following Trim-Away of individual components may depend on the biology of the complex in question. For example, rapid degradation of CENP-A using the auxin-inducible degron system also causes degradation of the CENP-A-interacting protein HJURP (Hoffmann et al., 2016). Similarly, degradation of mTOR by Trim-Away also led to a reduction of Raptor and Rictor, which form complexes with mTOR (Figures 6A and 6B).
The Trim-Away mechanism itself is highly specific, as shown in rescue experiments (Figures 3A, 3E, and 3F) and by using multiple antibodies raised against different regions of the same protein (Figures 5D–5F and S5I–S5M). These approaches can be readily employed to confirm the specificity of Trim-Away phenotypes. Non-specific antibodies should be avoided because Trim-Away with these antibodies can lead to the simultaneous degradation of several proteins (Figure S7D). Thus, the specificity of the Trim-Away approach will be determined by the antibody used.
Discussion
Gene knockouts and RNAi are widely used methods to study protein function. However, both methods act indirectly by blocking the expression of a protein and require that the target protein turns over. This can result in delayed protein disruption, an accumulation of non-specific defects and the activation of compensatory mechanisms. Current methods that act directly at the level of the protein either require that the endogenous protein is first replaced by a modified protein variant or are only applicable to a very small number of proteins.
Here, we have developed a widely applicable technique to acutely and rapidly degrade endogenous proteins in mammalian cells without the need for prior modification of the protein-coding gene or mRNA (Figure 1A). To our knowledge, this is the first posttranslational protein knockdown approach that can be readily applied to almost any intracellular protein.
We have shown that Trim-Away can degrade diverse protein substrates in both mammalian cell culture and primary mouse and human cells. Importantly, because the TRIM21-antibody interaction is highly conserved and maintained both within and between species (Keeble et al., 2008), the Trim-Away approach should be applicable to any mammalian cell using antibodies produced in any mammalian species. Trim-Away may also facilitate loss-of-function experiments in species that are not genetically tractable such as various marine species, or mammals that have long generation times.
Given that antibodies have been used over decades for various assays that require native protein recognition such as immunoprecipitation or immunofluorescence there is a comprehensive resource of antibodies with confirmed specificity readily available for use in Trim-Away assays. The growing pool of recombinant nanobodies (Chirichella et al., 2017, Helma et al., 2015, Pleiner et al., 2015) can also be utilized for Trim-Away simply by fusion to the Fc domain of conventional antibodies to allow TRIM21 binding (Figures 2I–2K).
By utilizing specific antibodies, Trim-Away is also suitable for selectively degrading posttranslationally modified proteins, splice or mutant protein variants while preserving the unmodified/wild-type protein; an approach that is not possible using current DNA- and RNA-targeting methods. Trim-Away could also be applied to study the function of proteins in selected cellular compartments. For instance, proteins that function both in the cytoplasm but also in other membrane-enclosed compartments or organelles can now be studied selectively in the cytoplasm without perturbing their function elsewhere, where they are shielded from antibodies. Vice versa, one could consider to target antibodies, nanobody-Fc fusions or TRIM21 to specific cellular sites for the local degradation of proteins.
Our observation that the endogenous levels of TRIM21 are sufficiently high for protein degradation in several cell types demonstrates that Trim-Away may be applicable without introducing excess TRIM21. This further simplifies the application of Trim-Away, especially in primary cells. If endogenous TRIM21 levels are insufficient for protein degradation, antibodies can easily be co-electroporated with recombinant TRIM21 protein to facilitate the complete degradation of target proteins.
In summary, this study provides unprecedented tools for studying protein function. First, it allows protein function to be studied in non-dividing primary cells where DNA- and RNA-targeting methods are not suitable. Second, it allows the functional analysis of long-lived proteins that are resistant to current knockdown methods that rely on protein turnover. Third, removal of essential endogenous proteins can now be achieved without the introduction of protein modifications such as degrons. Fourth, the remarkable speed of Trim-Away means that phenotypes can be observed immediately following degradation of the endogenous protein at any stage of a particular biological process. Finally, aberrant protein expression or activation is a hallmark of many human diseases such as neurodegeneration (Aguzzi and O’Connor, 2010) and cancer (Hanahan and Weinberg, 2011). It may become possible in the future to adapt the Trim-Away method to develop novel therapeutics that target disease-causing proteins for degradation….
