Proteasomal control of transcription factors: mechanisms, regulation and dysregulation

Abstract

Regulation of transcription is of key importance for essentially all aspects of biology. Accordingly, transcription factors (TFs) are often subject to rapid degradation, which in turn allows for strict regulation of gene expression and prevents uncontrolled activation or repression of target genes. This rapid turnover is a result of highly effective ubiquitin-dependent and -independent systems specifically selecting individual TFs for proteasomal degradation. The importance of life-time regulation of TFs is underscored by a broad range of diseases connected with dysfunctional TF degradation. Here, we summarize transcriptional regulation through controlled TF degradation. We focus on human TFs and highlight recent advances in our understanding of how TFs are selected for degradation through various proteasomal degradation pathways, and the implications of aberrant TF degradation.

Introduction

As principal regulators of gene expression, transcription factors (TFs) are critical for all cellular processes. While some genes are constitutively expressed, others are expressed exclusively in a subset of cells and tissues or in response to specific stimuli in a time-controlled process. Accordingly, regulating TF activity is of paramount importance [1–4], and TF dysregulation is often intimately connected with disease, including cancer, and metabolic and neurodegenerative disorders [4]. In general, TFs are subject to different modes of regulation, such as fluctuating expression levels, coregulator binding, post-translational modifications (PTMs), dynamic subcellular localization, including condensate formation [5, 6] and degradation. TFs have been shown to be cleared from the cell through multiple pathways, including proteolytic processing [7–10], extracellular vesicle-based disposal [11], but mainly through autophagy [12, 13] and the ubiquitin–proteasome system (UPS) [1–3]. In this review, we discuss the regulation of human TFs through protein degradation, mediated by the UPS, in health and disease.

In total, about 1,600 human TFs have been annotated, amounting to roughly 8% of the human protein coding genome [14]. Generally, TFs contain a structured DNA binding domain (DBD), which interacts with specific DNA sequences in the promoter regions of the genes they regulate. However, DNA binding may depend on regions outside the structured DBDs [15]. Upon DNA binding, the TFs either repress or promote transcription of the gene. Some TFs, such as the nuclear receptor family [16], contain other folded domains, i.e. the ligand binding domain. In addition, many TFs contain nuclear localization signals, nuclear export signals, and areas involved in dimerization. They also contain so-called activation domains (ADs) that are required to recruit transcriptional co-activators/co-repressors to promoter or enhancer regions. Many of these non-DNA binding features reside in regions characterized by intrinsic disorder. These intrinsically disordered regions (IDRs) [17–19], which can be very long in TFs [17], lack persistent structure and have an amino acid composition different to folded domains [20, 21], but typically harbor interaction motifs [22]. However, some regions in the IDRs, including the ADs, can in some cases fold upon binding their interaction partners [23]. Thus, despite typically being referred to as domains, many ADs function as hubs or scaffolds, carrying protein–protein interaction motifs embedded within the IDRs. Many ADs are typically enriched in hydrophobic residues interspersed between acidic residues, which has been suggested to prevent hydrophobic collapse, thus enabling their exposure to binding partners and repulsion from the negatively charged DNA [24]. While numerous ADs are conserved across substantial evolutionary distances, many typically exhibit limited sequence conservation between orthologues in line with their disordered character, even though their function is maintained [25].

Some TFs have been shown to localize to transcriptional condensates, which are temporal, localized non-membrane-bound dynamic compartments consisting of diverse molecules involved in gene transcription, including DNA, the Mediator complex, RNA polymerases and more [5]. Formation of such condensates may lead to efficient regulation of diverse steps in transcription, including repression [26], and is driven by multivalent interactions frequently provided by the IDRs of the participating proteins [27, 28]. Dysregulation of the formation and dissolution of transcriptional condensates, as well as changing biomaterial properties have been linked to disease, such as the cancerous Ewing’s sarcoma disease [29–31], and small-molecule-partitioning into condensates has been seen to impact the efficacy of cancer therapeutics [32].

Although some proteins, including a few TFs, are degraded by the proteasome independently of ubiquitin [33], most intracellular protein degradation occurs through the UPS [34–36]. Here, proteolysis is catalyzed by the 26S proteasome, and since most proteasomal degradation requires the target protein to be tagged with ubiquitin, the specificity is in general provided by the upstream ubiquitylation system. Ubiquitin conjugation of target proteins (ubiquitylation) is catalyzed by a three-step ATP-dependent enzymatic cascade. First, ubiquitin is activated and bound to an E1 ubiquitin-activating enzyme. Subsequently, the activated ubiquitin is transferred from the E1 to an E2 ubiquitin-conjugating enzyme, which finally in collaboration with an E3 ubiquitin-protein ligase catalyzes the formation of an isopeptide bond between the C-terminal carboxyl group in ubiquitin and the ε amino group in the side chain of a lysine residue in the target protein. Multiple cycles of ubiquitylation result in a target protein being conjugated to a ubiquitin chain, which in turn targets the protein to the 26S proteasome for degradation [37]. The 26S proteasome is a protease complex composed of a cylindrical 20S particle flanked by two 19S regulatory complexes. The proteolytically active sites are positioned within the hollow 20S cylinder, while the 19S complexes provide the ubiquitin receptors as well as ATPase subunits required for translocating the substrates into the lumen of the 20S proteasome [37]. Accordingly, the specificity in the degradation is primarily provided by the E3 ubiquitin-protein ligases [38] that either directly or indirectly via adaptor proteins, such as the large family of F-box proteins [39], recognize the target protein, through so-called degradation signals or degrons [40, 41].

Cellular proteins are continuously renewed. Since a fast protein turnover is a prerequisite to ensure a rapid change (both decrease and increase) in protein level [42], key signaling proteins are expected to be rapidly degraded. Over the years, it has been well-documented that many transcriptional regulators are short-lived, and the UPS thus ensures tightly controlled transcriptional regulation [3]. Well-described examples include the degradation of the c-JUN component of the AP-1 transcriptional activator, which depends on a 27-residue N-terminal fragment that is lacking in its retroviral counterpart v-JUN [43], and the degradation of HIF1α under normoxic conditions, mediated by the von Hippel-Lindau E3 ligase complex [44, 45]. Other examples are provided by the turnover of p53, regulated by the E3 ligase MDM2 [46, 47], and the signal-induced proteasomal degradation of the IκBα inhibitor, facilitated by the SCF^FWD1 E3 ubiquitin-ligase complex [48], which contributes to regulation of the TF NF-κB [49]. Despite their importance for regulating cellular protein abundance, the majority of the > 600 human E3 enzymes are currently without known substrate [41]. Accordingly, characterizing degrons in TFs and connecting these to their respective E3s, is critical for understanding transcriptional regulation on the cellular and molecular levels.

Naturally, for a degron to play a role in the degradation of a TF, it should be sufficiently exposed for E3 recognition. Accordingly, it is expected that most regulatory degrons are embedded within the IDRs [50]. Conversely, degrons located within the folded DBD are likely to be buried and inaccessible. However, DBDs are dynamic entities in the absence of DNA [51] and if the structure is destabilized or the protein becomes misfolded (e.g., as a result of mutations) such buried degrons will be exposed and can therefore play a role in clearing unbound and non-native TFs from the cell [52].

Regulated degradation of transcription factors

As part of a regulatory mechanism, many TFs are short-lived and thus naturally turned over rapidly via the UPS [2]. For example, p53 and c-JUN display half-lives of about 20 min and 90 min, respectively [53, 54]. This rapid turn-over is striking compared to the much longer half-lives of housekeeping proteins such as GAPDH and histone HIST1H1C that extend to weeks and months, respectively [55]. While this may seem wasteful, it ensures a low steady-state level of the TF in situations where it is not needed and allows the cell to react instantly to a signaling event with quick TF accumulation and mount a prompt transcriptional response. Accordingly, in this scenario, the cell must have a means to regulate the degradation of distinct TFs in response to specific signaling cues. In principle, this can be achieved through either regulating the E3s or other components of the degradation machinery or, as we focus on here, by activating or deactivating degrons directly in the TFs [2, 56]. For instance, PTMs, such as phosphorylation and sumoylation have been shown to regulate TF ubiquitylation and degradation [57, 58] (Fig. 1). In addition, it has recently been shown that proteasomal degradation of a subset of proteins is independent of ubiquitin (Fig. 1) [33, 59]. Finally, like other proteins, the DBDs of the TFs may —as a result of mutations or environmental stress conditions— unfold or misfold and thereby become targets for the cellular protein quality control (PQC) degradation system (Fig. 1). PQC mediated degradation also depends on the UPS but typically relies on a specialized subset of E3s that interact with molecular chaperones [60]. In the following, we provide examples of TF degradation mechanisms and the consequences of their dysregulation.

Fig. 1.

Regulated degradation of transcription factors. Transcription factors (TFs) are subjected to different modes of regulated degradation. Post-translational modifications (PTMs) such as phosphorylation or sumoylation can either: (I) induce or (II) inhibit ubiquitin-dependent proteasomal degradation. Depending on their exposure, constitutively active degrons in (III) disordered regions or (IV) protein quality control (PQC) degrons, exposed due to unfolding or misfolding of the structured DNA binding domain, similarly target TFs for ubiquitin-dependent proteasomal degradation. (V) C-terminal alanine, valine and cysteine residues function as C-terminal degrons (C-degrons) targeting the substrate for ubiquitin-independent degradation, in some cases facilitated by the UBQLN protein family. Finally, (VI) midnolin (MIDN) can mediate ubiquitin-independent degradation by directly bridging the TF and the proteasome. Created in BioRender. Hartmann-Petersen, R. (2026) https://BioRender.com/je6jd19

Phosphorylation regulates TF degradation

Protein degradation via the UPS is well known to be regulated by phosphorylation [61]. However, the phosphorylation event may have different outcomes [62]. When phosphorylation drives the degradation of a protein, phosphorylation typically generates a phosphodegron, a short linear motif (SLiM) [61] that upon phosphorylation provides a specific binding surface for an E3 ligase, resulting in ubiquitylation and degradation of the target protein. Conversely, phosphorylation may also inhibit degradation by masking a degron in the target protein [57].

The Skp, Cullin, F-box (SCF) complex is a multi-subunit E3 protein complex with a core consisting of the SKP1, CUL1 and RBX1 subunits functioning together with a variable F-box protein harboring the substrate binding activity of the complex. The human genome encodes ~ 70 F-box proteins, of which twelve, known as FBXW proteins, contain a WD40 domain, where some are known to recognize and bind phosphorylated substrate proteins and subsequently bring them into proximity of ubiquitin-conjugating enzymes [63–68]. A well-characterized example is FBXW7, which binds the phosphodegron motif S/TPPXS/T (X represents any residue) when the serine or threonine residues are both phosphorylated [69]. Noticeably, mutations affecting both the FBXW7-binding phosphodegrons as well as in the FBXW7 gene itself, have been connected with cancer [70]. According to the COSMIC database [71] of somatic cancer mutations, FBXW7 is amongst the most commonly mutated F-box protein encoding genes in a multitude of cancers [70], and the mutations often include missense variants in arginine residues that mediate interaction with the phosphorylated substrates [72]. The SCF^FBXW7 complex controls the phosphorylation-dependent proteasomal degradation of many oncoproteins, including several TFs such as KLF5, NF-κB2 and c-MYC [70]. As a Yamanaka factor [73], c-MYC is involved in reprogramming differentiated cells, but also activates transcription of growth-related genes. Therefore, c-MYC is subjected to meticulous regulation with a half-life of a mere 20–30 min [74, 75]. Mediated by GSK3, c-MYC is phosphorylated at T58 within a phosphodegron (-₅₈TPPLS₆₂-, UniProt ID: P01106-1) (Table 1) embedded in its AD resulting in ubiquitylation and degradation of the TF. Importantly, the phosphorylation of T58 is dependent on an ERK-mediated priming phosphorylation at S62 [76, 77]. This degradation pathway was first identified as v-MYC genes in avian retroviruses, and translocated MYC alleles in Burkitt lymphoma cells both present mutations of T58 predicted to disrupt the controlled degradation of c-MYC. Indeed, a T58A variant leads to slower turnover of the protein [75, 78]. Additionally, the protein kinase Aurora A has been shown to interact with c-MYC when phosphorylated at positions T58 and S62, with the binding interface spanning a 145-residue segment within the N-terminal AD including the phosphodegron and two conserved regions involved in transcriptional regulation [79]. Aurora A thus interferes with c-MYC ubiquitylation and degradation, adding an additional level to the regulation. Recent studies have identified a second phosphodegron within the C-terminus of c-MYC situated at T244 (-₂₄₄TPPTT₂₄₈-, UniProt ID: P01106-1). This second degron resembles the T58 degron, being phosphorylated following a priming phosphorylation at T248. Possibly the two phosphodegrons function in unison requiring homodimerization of FBXW7 [76].

Table 1.

Selected degrons regulating TF abundance

Degron	Protein	Cognate E3	References
Phosphodegrons
-58TPPLS62-	c-MYC	SCFFBXW7	[76, 77]
-244TPPTT248-	c-MYC	SCFFBXW7	[76]
-303SPPQS307-	HSF1	SCFFBXW7	[81]
-216DSGSAHS221-	HSF1	SCFβ−TrCP	[83]
-239TPPLS243-	c-JUN	SCFFBXW7	[85]
-167PPXY170-	c-JUN	ITCH	[86]
-32SGLDS36-	IκBα	SCFFWD1	[48, 107]
Constitutive degrons
-198QSNRILGVKRKIPLMLNDSG217-	HSF1	Unknown	[82]
-23PFLPPAACFFATAAAAAAA41-	ASCL1	Unknown	[82]
-247IKVYIF252-	ZFY	Unknown	[166]
PQC degrons
-306CGKAFRRFSHLTRH319-	ZNF140	Unknown	[82]

Other substrates of FBXW7 include HSF1, a stress-responsive TF responsible for expression of genes involved in protein folding and stress adaptation [80]. Here, HSF1 is phosphorylated under normal (non-stressed) cellular conditions at the conserved priming position S307 by ERK1 and the central phosphorylation site S303 (-₃₀₃SPPQS₃₀₇-, UniProt ID: Q00613-1) (Table 1) by GSK3 resulting in rapid degradation of the protein [81]. However, other degrons in HSF1 also play roles in lowering HSF1 abundance. For instance, the Q198-G217 region (-₁₉₈QSNRILGVKRKIPLMLNDSG₂₁₇-) (Table 1) contains a constitutive degron targeted by an unknown E3 [82], and this degron overlaps with a SCF^β−TrCP consensus site (-₂₁₆DSGSAHS₂₂₁-) recognized and targeted for degradation by the E3 complex SCF^β−TrCP following phosphorylation at position S216 by the protein kinase PLK1 [83].

The c-JUN TF, which dimerizes with c-FOS to form the AP-1 transcription factor complex involved in cell proliferation and differentiation [84], provides another example of an FBXW7 target. The half-life of c-JUN is about 90 min [54]. FBXW7 recognizes and targets c-JUN for degradation through a C-terminal phosphodegron (-₂₃₉TPPLS₂₄₃-, UniProt ID: P05412) (Table 1) requiring a priming phosphorylation at S243 followed by GSK3-mediated phosphorylation at T239 [85]. However, JNK-catalyzed phosphorylation upstream at S63 and S73, and to a lesser degree at T91 and T93, efficiently blocks ubiquitylation resulting in substantial stabilization of the protein [54]. These phosphorylation sites are interestingly not located near the degradation-promoting phosphorylation site; however, the exact mechanism behind this JNK-mediated stabilization is yet to be uncovered but may involve structural effects on the overall ensemble [62]. Additionally, c-ABL-dependent phosphorylation of c-JUN at Y170 disables a P-P-X–Y degron (-₁₆₇PPXY₁₇₀-, UniProt ID: P05412) (Table 1), blocking the interaction between c-JUN and the E3 ubiquitin ligase ITCH [86].

Finally, another well-described example of phosphorylation-dependent degrons involved in transcriptional control is the regulation of the transcriptional co-activator and key switch in the WNT signaling pathway, β-catenin [87]. Here, cellular levels of β-catenin are kept at a constant low by the E3 complex SCF^β−TrCP1 following phosphorylation at residues S33 and S37 (-₃₃SGIHS₃₇-, UniProt ID: P35222) and binding to the WD40 domain of β-TrCP1 [66]. Upon activation of the WNT signaling pathway, phosphorylation of β-catenin is inhibited, resulting in increased cellular abundance and nuclear translocation of the co-activator, allowing it to control activation of the TCF/LEF (T-cell factor/lymphoid enhancer-binding factor) TFs, crucial for embryonic development [66, 87–89].

As evident from the given examples, phosphodegrons critically regulate the turnover of several key TFs in response to various signaling cues. However, many TFs have multiple degrons that contribute to TF degradation, and since some of these can be blocked by phosphorylation, this provides a means for different signaling pathways to differentially regulate the abundance and activity of TFs.

SUMO-dependent degradation of TFs

The small ubiquitin-like modifier (SUMO) has been established as a key regulatory protein of several cellular processes including regulation of gene expression [58, 90, 91], and TFs constitute some of the most frequently identified targets of sumoylation [92]. SUMO has mainly been described as a repressor of transcription, with downstream effects of sumoylation rooted in the modification of protein surfaces, thereby impacting DNA binding, localization and steady-state levels of TFs [91]. Similar to ubiquitylation, sumoylation involves the covalent attachment of SUMO to a specific lysine residue in the target protein, preferentially located in IDRs [92]. SUMO is a small, ~ 10 kDa, modifier protein with a structure resembling that of ubiquitin, yet the two proteins share only ~ 20% sequence identity [93].

SUMO is activated and conjugated to target proteins in an ATP-dependent three-step enzymatic cascade resembling that of ubiquitylation, but dependent on its own specific set of E1, E2 and E3 enzymes [94, 95]. Target specificity is largely determined by the SUMO E2 conjugating enzyme UBC9 that primarily recognizes targets through a sumoylation consensus SLiM, ΨKXE (Ψ represents a bulky aliphatic residue) [90]. Notably, the isopeptide bond between SUMO and the substrate protein can be cleaved by SUMO-specific proteases making the modification reversible and ideal as a molecular switch for signaling [96].

Sumoylation can induce degradation of substrate proteins by marking them for recognition by SUMO-targeted ubiquitin ligases (STUbLs) resulting in subsequent ubiquitylation and proteasomal degradation. One of the most extensively studied STUbLs is the human RNF4 protein [97]. RNF4 inactivation leads to inadequate response to DNA damage [98, 99] with implications such as chromosomal translocations, as well as cell death and mutations that increase the risk of a cancer [100]. One target of RNF4 is the TF SP1, which regulates the expression of a wide range of genes and is subject to different PTMs, including phosphorylation-dependent sumoylation. In addition to its role as a TF, SP1 also regulates the DNA damage response by facilitating the recruitment of repair factors to DNA breaks, followed by dissociation of SP1 from the chromatin and sumoylation of the TF at position K16 (-₁₅VKIE₁₈-, UniProt ID: P08047-1) in the S-phase [101]. In turn, this leads to RNF4-mediated ubiquitylation and subsequent proteasomal degradation [101, 102]. Additional targets of RNF4 include the ETS-domain TF PEA3, involved in neuronal pathfinding and associated with tumor metastasis. This TF is SUMOylated on K96 (-₉₅IKKE₉₈-, UniProt ID: A2A5C2) in a phosphorylation-dependent manner, and on K222 (-₂₂₁FKQE₂₂₄-, UniProt ID: A2A5C2) and K256 (-₂₅₅IKQE₂₅₈-, UniProt ID: A2A5C2) following activation of the ERK MAP kinase pathway [103]. Additionally, the previously discussed c-MYC has ten sumoylation motifs and is subject to sumoylation by the SUMO E3 ligase PIAS1 and subsequently recognized and ubiquitylated by RNF4 [104].

Importantly, nearly a quarter (22.4%) of the identified sumoylation sites overlap with described ubiquitylation sites [105]. Hence, sumoylation may also stabilize specific proteins by temporarily shielding ubiquitylation sites (Fig. 1). This mechanism is critical for the regulation of NF-κB and its regulation of immune responses [106]. In unstimulated cells, NF-κB is retained in the cytosol in an inactive state by IκB inhibitor proteins. Activation of NF-κB requires signal induced degradation of IκBα leading to translocation of NF-κB to the nucleus, where it binds and initiates transcription of its target genes. Central to its regulation, under normal cellular conditions IκBα is conjugated to SUMO by UBC9 primarily at K21 (-₂₀LKKE₂₃-, UniProt ID: P25963), a known ubiquitylation site. Accordingly, this protects the protein from ubiquitylation and degradation. Interestingly, ubiquitylation of IκBα requires prior phosphorylation of S32 and S36, located in a phosphodegron (-₃₂SGLDS₃₆-, UniProt ID: P25963) (Table 1), which, conversely has an inhibitory effect on IκBα sumoylation [107].

Collectively, the above-described examples illustrate the intricate crosstalk between different signaling pathways and different PTMs required for regulation of gene expression through targeted degradation, or stabilization, of TFs.

Other examples of ubiquitin-dependent TF degradation

As described, signaling pathways leading to PTMs of TFs can lead to their degradation via the UPS. Several other examples exist where ubiquitin-mediated TF degradation is critical for keeping TF levels at bay, thus allowing a rapid increase in TF abundance in response to signaling cues [108]. An example of this is the pioneer transcription factor [109] ASCL1, involved in early development of neural and neuroendocrine progenitor cells [110]. Contributing to keeping the ASCL1 protein levels low is its degradation mediated by the E3 ubiquitin ligase HUWE1. Under normal conditions, HUWE1 polyubiquitylates ASCL1 at one or more lysine residues placed within the basic helix-loop-helix DBD of the protein when located in the cytosol [111]. This controlled degradation of ASCL1 is in place to minimize its nuclear localization and chromatin association, thus preventing untimely neuronal differentiation. When neuronal differentiation is required, degradation of the protein is inhibited by TCF3 binding [112], resulting in a rapid increase in ASCL1 abundance [111]. Consistent with this, recent work has identified a degron placed in the N-terminus of ASCL1 at P23-A40 (-₂₃PFLPPAACFFATAAAAAAA₄₁-, UniProt ID: P50553) (Table 1). This degron is predicted to be located in an IDR, thus making the degron readily available for interaction with the ubiquitylation machinery [82].

Another example highly dependent on intricate regulation by ubiquitin and of particular importance for disease is the tumor suppressor p53, that critically operates during stress conditions and in response to DNA damage by inducing cell cycle arrest and apoptosis [113]. Under normal conditions, the steady-state protein level of p53 is kept low through its rapid proteasomal degradation. p53 ubiquitylation is primarily catalyzed by the E3 ubiquitin ligase MDM2 [46, 47, 114], which binds an N-terminal motif between residues ~ 20–30 within p53 adopting an α-helical structure upon interaction. Here, residues F19, W23 and L26 (-₁₉FSDLWKLL₂₆-, UniProt ID: P04637-1) interact with a hydrophobic cleft in MDM2 [115, 116]. Degradation of p53 additionally relies on a region located further downstream (-₉₂PLSSSVPSQKTYQGSYGFRLG₁₁₂-, UniProt ID: P04637-1) (Table 1) [117]. However, more than 15 different E3 ligases and other modulators have been connected with p53 ubiquitylation [118], including PIRH2 [119], which has been shown to target a different set of lysine residues than MDM2. While both PIRH2 and MDM2 have been shown to ubiquitylate residues K101, K292, K305 and K357, the lysine residue at position 164 was shown to be exclusively ubiquitylated by PIRH2 and the lysine at position 319 exclusively by MDM2 [120]. Accordingly, the regulation of p53 levels is exceedingly complex. However, in its simplest iteration, MDM2-mediated regulation of p53 is controlled by p53 itself via a negative feedback loop where p53 controls the expression of the MDM2 gene. At low cellular levels of p53, MDM2 levels will also be low, leading to p53 monoubiquitylation and nuclear export, likely to be mediated by monoubiquitylation-dependent exposure of a C-terminal nuclear export signal [121]. Conversely, when p53 levels are high this will result in an increased expression of MDM2, in turn leading to p53 polyubiquitylation and degradation [122].

In addition to the structured p53-MDM2 interaction, evidence further supports that some degrons function as structurally defined elements, including folded domains, which was recently reviewed [115]. A remarkable example includes BACH1 that functions as a transcriptional repressor in heme metabolism and forms homodimers via its BTB domain. Increased heme levels promote BACH1 release from DNA thereby leading to exposure of the BTB dimer which in turn interacts with the E3 ligase FBXO22, triggering BACH1 proteasomal degradation [123, 124].

An example of a context dependent degron includes NRF2, a transcription factor that plays a key role in the oxidative stress response, which contains two degrons recognized by the E3 ligase KEAP1 [125]. While either degron can bind KEAP1, both are required for efficient ubiquitylation. Given the abundance of E3-encoding genes in the human genome, such structural or context dependent degrons are likely to be common.

Like most PTMs, ubiquitylation is a reversible modification, and conjugated proteins can thus be deubiquitylated by proteases appropriately named deubiquitylases (DUBs). DUBs are critical in various regulatory processes and may rescue TFs otherwise destined for degradation. An example includes p53 that can be deubiquitylated by several DUBs such as USP3 [126], USP7 [127], and USP10 [128]. Moreover, another target of USP7 is MDM2, which, during cellular stress conditions, is phosphorylated by the ATM kinase, thus lowering its affinity to USP7, resulting in MDM2 destabilization and thereby accumulation of p53. Other substrates of USP7 include GATA1, a key TF involved in regulation of erythroid maturation and function [129]. Here, USP7-mediated deubiquitylation of polyubiquitin chains on GATA1 is involved in erythroid differentiation by stabilizing protein levels of GATA1 [130].

The above-described examples display the critical role of ubiquitin-dependent degradation for controlling the cellular abundance of TFs, and thus in turn their transcriptional activity.

Ubiquitin-independent TF degradation

It is evident that ubiquitin is key to proteasomal degradation of most intracellular proteins. Still, early studies showed that certain disordered proteins can be degraded independently of ubiquitin directly by the 20S proteolytic proteasome particle [131]. However, the mechanistic details on how these proteins are able to bind and enter the 20S proteasome are lacking. In addition, the transcriptional co-activator NCOA3 has been shown to be degraded in both a ubiquitin-dependent [132] and -independent [133] manner, where the latter depends on the proteasome-activator REGγ. Ubiquitin-independent proteasomal degradation has been reported for several other proteins, including multiple TFs such as yeast Rpn4 [134], and human p53 [135] and FOS [136]. Results of in vitro studies suggest that ubiquitin-independent degradation could be relevant for up to 20% of cellular proteins [137]. A recent high-throughput systematic approach identified peptides that function as degrons for ubiquitin-independent proteasomal degradation primarily mediated by sequence specific degrons located in the C-terminal part of the peptides (Fig. 1) [33]. More specifically, A, V and C residues at the three last positions in certain targets were shown to lead to ubiquitin-independent degradation. About 70 full-length human proteins were identified as substrates, including the TF SOX30, a member of the SOX family involved in embryonic development [138] and the previously identified ubiquitin-independent degradation target FOS [136]. The degradation of a subset of the identified substrates was shown to depend on members of the UBQLN family of proteasome co-factors [33]. These co-factors are known to facilitate degradation by interacting with both the substrate and the proteasome [139].

Recently, another novel ubiquitin-independent proteasomal degradation pathway was described [59]. A CRISPR screen was employed to identify regulators of the TFs FOSB and EGR1 [59], where FOSB had previously been proposed to undergo ubiquitin-independent degradation [140]. Both TFs are members of the immediate-early gene (IEG) family that function as first responders to cellular stimuli and mediate the transcription of late-response genes [141]. A top hit from the screen was MIDN, a gene encoding the UBL-domain protein midnolin [59]. This protein, found mainly to localize to the nucleus, had early on been suggested to contribute to mouse development [142] and regulation of the protein Parkin [143] involved in mitophagy and Parkinson’s disease [144]. However, the midnolin protein had remained largely uncharacterized, but now shown to target both IEGs and other TFs such as GATA1, IRF4 and PAX8 for degradation [59]. Within the nucleus, midnolin directly interacts through its C-terminal α-helix with the PSMD2 subunit of the 26S proteasome, also known to recruit ubiquitylated substrates during the canonical ubiquitin-dependent proteolytical pathway [145, 146]. Next, midnolin uses its Catch domain to interact with aliphatic disordered regions in the target protein, which may form β-strands upon interaction. This is followed by binding of the UBL domain of midnolin to the PSMD14 subunit of the 26S proteasome [145, 146], normally involved in deubiquitylating substrates prior to their degradation. This positions the Catch domain with the bound TF substrate directly over the proteasomal substrate entry site directing the TF into the proteasome for degradation [145, 146]. Additionally, midnolin was found to be downregulated in malignant plasma cells from patients suffering from multiple myeloma, leading to abnormally high cellular levels of the TF IRF4 [145]. This uncovers a previously unknown clinical relevance of midnolin and demonstrates a new and fascinating mechanism for rapid ubiquitin-independent proteasomal degradation.

Protein quality control degradation of TFs

It is well established that proteins which become structurally unstable as a result of environmental stress conditions or mutations are rapidly degraded via the UPS [147]. Accordingly, computational predictions of all possible missense variants in all human proteins indicate that about half of all pathogenic missense protein variants are structurally destabilized with a ΔΔG of at least 2 kcal/mol [148]. In turn, this destabilization is expected to render most of these protein variants to become targets of the PQC degradation pathways [149, 150]. Typically, PQC degradation involves various E3 ubiquitin-ligases that often in collaboration with molecular chaperones engage the misfolded or partially unfolded target, leading to its ubiquitylation, degradation, reduced cellular abundance and a loss-of-function phenotype [151].

PQC degradation of missense protein variants has been identified as one of the primary underlying molecular determinants in a range of rare genetic diseases [152–161]. As long stretches of most TFs are intrinsically disordered, TFs are rarely PQC targets. However, missense variants resulting in structural destabilization of folded DNA binding domains may lead to PQC degradation. Accordingly, we recently found that a somatic missense mutation linked to skin cutaneous melanoma [162] in the ZNF140 C2H2 zinc finger (ZnF) motif results in local unfolding of the folded ZnF structure. Consequently, this leads to exposure of a PQC degron (-₃₀₆CGKAFRRFSHLTRH₃₁₉-, UniProt ID: P52738) (Table 1) buried within the native C2H2 fold and its subsequent degradation [82]. Additionally, mutations in the GATA1 N-terminal ZnF have been postulated to disrupt folding and thus diminish binding to the GATA1 interaction partner, FOG1 [163]. Another clinically relevant PQC target is FOXC1, where missense mutations have been linked to Axenfeld-Rieger syndrome 3, a disease resulting in developmental defects in the anterior segment of the eye [164]. Here, substitutions in the DBD have been shown to result in altered DNA binding, reduced activation ability and, in the case of the I87M mutation, reduced protein stability and increased degradation, all of which contribute to loss-of-function [165].

High-throughput degron mapping in human TFs

Due to the significance of TF degradation in transcriptional regulation, it is important to identify degrons in TFs and understand the mechanistic details of their function. In recent years, several reports have described the results of various high-throughput screens in both yeast and human cells aimed at mapping degrons [33, 52, 166–174]. Some of these studies define and characterize N- and C-degrons [33, 167, 168, 173], which are likely relevant for many TFs. For instance, the BTG2 transcriptional coactivator, which is a short-lived ubiquitin–proteasome target [175, 176], is also targeted and degraded by the proteasome independently of ubiquitin, likely via a C-terminal degron [33]. In addition, the N- and C-degrons may also contribute to the turnover of TF fragments resulting from prior cleavage by caspases or other proteolytic enzymes [7–10].

Recent proteome-wide screening performed by Zhang et al. [166] identified thousands of degron peptides, which—together with CRISPR ablation of E3s—led to the discovery of novel E3-degron pairs. These include the Zn-finger proteins ZNF19 and ZNF510, which are targeted by the E3 ligase complexes CUL4^AMBRA1 and CUL4^DCAF5, respectively, and the ZFY transcriptional activator recognized and targeted by the BAG6 chaperone, and likely its associated E3 RNF126, by an exposed internal degron (-₂₄₇IKVYIF₂₅₂-, UniProt: P08048). Recently, we applied such screening technologies to characterize degrons in all human TFs creating a comprehensive degron map of the TFs [82]. The majority of the degrons were located in the structured DBDs of the proteins, hence likely constituting PQC degrons. However, about 40% of degrons were located in regions predicted to be disordered. Assuming no regulation by PTMs or other regulatory mechanisms, degrons embedded within an IDR are expected to be exposed and therefore constitutively active, thus contributing to keeping the abundance of the TFs low. In accordance with degron screens targeting other groups of proteins, the degron activity of internal PQC degrons does not seem to rely on specific SLiMs, but rather the overall features emerging from amino acid composition of a region [168–170, 174]. This observation may, however, in part stem from the approaches used in large-scale studies, where e.g. PTM-, structure- or context-dependent degrons may not be detectable. Another limitation of using short fragments is an increased exposure that may not be readily available in e.g. collapsed chains of long IDRs (Fig. 2).

Fig. 2.

Multiple roles of acidic residues in activation domains. Acidic residues in an activation domain (AD) play multiple roles: they repel and inhibit unfavorable interactions with DNA, they reduce the inherent degron potency of the AD and thereby prevent degradation, and they prevent hydrophobic collapse, leaving the AD exposed, allowing interaction with co-activators/co-repressors required for activation/repression of transcription. Created in BioRender. Hartmann-Petersen, R. (2026) https://BioRender.com/are4akh

The results from large-scale degron mapping experiments have enabled the development of sequence-based degron prediction methods [166, 174, 177]. While the methods and data differ, the overall results show that the abundance of many short polypeptides can be predicted relatively accurately by amino acid composition, where hydrophobic residues promote degradation, while acidic residues reduce degron activity, with additional model complexity needed to capture effects at the N- or C-termini. These prediction models can in turn be applied to study the relationship between TF function and degradation, and to understand the effects of missense variants in TFs. For example, using a computational predictor of protein abundance, it was shown that ADs in general overlap with degrons [82]. Accordingly, systematic studies of ADs have shown that most of these are enriched for hydrophobic residues but also contain acidic residues [178, 179]. The presence of hydrophobic residues in ADs is consistent with them operating as degrons [180], while the acidic residues are not. Early studies have shown that the acidic residues only have a limited effect on activation compared to the bulky hydrophobic residues [181–183], and it has therefore been suggested that the acidic residues are required to prevent hydrophobic collapse of the ADs [24] ensuring exposure of the AD allowing co-activator/co-repressor recruitment and binding (Fig. 2). This is likely the case, but the degron predictions indicate that the acidic AD residues also protect the TFs from degradation. Accordingly, during evolution, the UPS may have contributed to the enrichment of acidic residues in ADs, so that the role of acidic residues in TF may in part be to prevent hydrophobic regions from functioning as degrons [82, 177]. In agreement with this, computational predictions of mutations resulting in degron formation or disruption correlate with the results from deep mutational scanning on the CRX TF [82, 184], involved in development of the central nervous system, pancreas and eye [184]. Supporting these observations that a major role for acidic residues in ADs is to prevent degradation, a recent mechanistic model for AD predictions showed that the importance of acidic residues for AD function could be largely ascribed to their stabilizing role rather than a direct effect in transcription [185].

Outlook

TFs are implicated in essentially all cellular processes and are thus fundamental for life. It is therefore critical that TFs themselves are strictly regulated, and controlled TF turnover is key to generate the appropriate response to cellular stimuli [3]. Understanding the different mechanisms of targeted TF degradation is important not only to provide a mechanistic explanation for the multitude of diseases rooted in the dysregulation of TFs [4], but also to comprehend the intricate network of TF regulation. In this review, we have described the different mechanisms of TF degradation and elaborated on the importance of degrons in TFs. While some degrons are regulated, many are described as constitutively active. However, for both it is important to consider the dynamic structure of TFs and the ways in which local structures may impact the exposure or availability of the degron (Fig. 3). Accordingly, events such as intramolecular and dynamic binding of the AD to the DBD [186–191], interaction with binding partners such as co-activators or co-repressors [192], TF dimerization [193] and PTMs may all shield or block the degron and thereby inhibit degradation. In their unbound state, IDRs, which are abundant in TFs, have been shown to form transient structures, and varying levels of compaction important for TF functions [194, 195]. These dynamic conformational changes may also affect the availability of the degron and thereby contribute to regulating the abundance of the proteins (Fig. 3). Additionally, it has been shown that some TFs undergo phase separation with co-activators/co-repressors and the transcriptional machinery at genomic loci to facilitate gene expression, suggesting condensation as a possible additional mechanism to regulate degron availability [5, 196, 197].

Fig. 3.

The dynamics of transcription factors impact degron availability. Constitutively active degrons may be shielded from the degradation machinery (Pac-Man) by: (I) intramolecular interactions between the intrinsically disordered regions (IDRs) and the structured DNA binding domain, (II) interaction with binding partners, (III) transcriptional condensate formation, (IV) post-translational modifications (PTMs), (V) transcription factor dimerization or, (VI) due to the structural ensembles of the IDR. Created in BioRender. Hartmann-Petersen, R. (2026) https://BioRender.com/pu1znrj

Dysregulation of TFs is implicated in a wide range of diseases [4] but due to their lack of enzymatic activity and ligand binding pockets, TFs are difficult drug targets. One approach towards their targeting is by generating so-called TF-specific deubiquitylase targeting chimeras (TF-DUBTACs). Here, a DNA oligonucleotide targeted by the TF of interest is covalently linked to a ligand of the deubiquitylase OTUB1 [198], leading to recruitment of the DUB to the target TF, resulting in stabilization and an increased cellular activity of the TF. In contrast, a similar approach, known as a TF-specific proteolysis targeting chimera (TF-PROTAC), links an E3 ligase ligand to a DNA oligonucleotide specific for the TF of interest, thus allowing targeted degradation of the TF [199]. These two methods provide a generalizable platform to selectively regulate the cellular abundance of TFs by small molecules and provide new possibilities for drugging TFs. However, none of these approaches target the IDR where 40% of degrons are located [82], and while both TF-PROTACs and TF-DUBTACs depend on the availability of non-DNA bound TF, targeting degron sites in the IDRs will be possible also in the DNA bound state of a TF. Thus, detailed understanding of degrons and TF degradation is highly warranted.

Development of high-throughput methods to study transcriptional activity and protein abundance has contributed significantly to the knowledge we now possess regarding TF regulation and dysregulation. However, there are still major gaps in our understanding, both in relation to the detailed molecular mechanisms on how ADs operate in transcriptional activation and repression, but also regarding the degradation of TFs and how this is regulated. The transient nature of the interaction between E3 ligases and substrate makes substrate identification challenging, leaving the majority of the ~ 600 human E3 ligases without known substrates. Identifying the substrates of E3 ligases is important since studies have shown that ~ 19% of all cancer driver genes affect UPS function [200], and presumably, TFs constitute a large fraction of the UPS targets. Recently, work on multi-scale classification of all human E3 ligases [201], as well as development of methods to identify substrates of E3 ligases at scale [202] have emerged, making it an exciting time to study transcriptional regulation through TF degradation.

Author contributions

F.B.L. prepared the figures. F.B.L., B.B.K., K.L.-L., P.O.H., and R.H.-P. wrote and edited the paper.

Funding

Open access funding provided by Copenhagen University. R.H.-P. is funded by the Novo Nordisk Foundation grants NNF18OC0033950, NNF18OC0033926, NNF21OC0071057, and NNF0102348 and the Danish Council for Independent Research (Det Frie Forskningsråd) https://doi.org/10.46540/2032-00007B and https://doi.org/10.46540/5284-00009B. B.B.K. is funded by the Novo Nordisk Foundation grant NNF18OC0033926. P.O.H. is funded by the European Research Council (ERC StG 101040601-PIONEER). The funders had no role in the preparation or the decision to publish the manuscript.

Data availability

No data were generated as part of this paper.

Declarations

Competing interests

The authors declare no competing interests.