Targeting the Primordial Chaperone to Overcome Acquired Drug Resistance in Cancer: TG2-Mediated Autophagy

Abstract

The transglutaminase family is roughly 250 million years old. Horseshoe crabs have been described as a ‘living fossil’ and have remained virtually unchanged since first appearing around the Triassic period. The horseshoe crab, a living fossil, carries a primitive form of TG that helps it defend against infection and survive. Transglutaminase 2 (TG2, EC 2.3.2.13, gene name TGM2) is mainly known as a cross-linking enzyme in vertebrates. Although TG2 is not an oncogene, its high levels are linked to worse outcomes in many cancers. However, how TG2 cross-linking activity relates to its role in promoting cancer growth remains unclear. A recent discovery sheds light on this. In ovarian cancer cells, TG2 binds directly to GSK3β, leading to its removal by autophagosomes, which activates β-catenin. Stopping this interaction allows GSK3β levels to recover, thereby decreasing β-catenin activity. Even in the absence of cross-linking, cancer cells use TG2 as a chaperone to promote growth and support metastasis. This suggests intracellular calcium levels are too low for TG2 to perform cross-linking. It also indicates that anticancer treatments may increase TG2 levels in cancer cells, helping them recover by removing tumor suppressors. As a result, TG2 plays a role in developing drug resistance, acting as a primitive systemic defense mechanism linked with survival signals. I suggest that blocking TG2 binding, combined with inhibiting autophagy or alternative signaling pathways, is essential for effectively overcoming drug resistance, since it is rooted in TG2’s primordial role.

INTRODUCTION

Transglutaminase 2 (TG2, EC 2.3.2.13, gene name TGM2) has been extensively reviewed regarding its diverse enzymatic roles and its physiological functions (Belkin, 2011; du Pre et al., 2025; Gundemir et al., 2012; Ibrahim et al., 2025; Lan et al., 2025; Li et al., 2024; Min and Chung, 2018; Yao et al., 2024; Zaltron et al., 2024). However, this review does not aim to cover all its functions. Instead, this review focuses solely on TG2-binding activity, specifically its role in cancer and its molecular mechanisms. From this specific perspective, the TG2 substrates can be generally classified into extracellular and intracellular domains, a distinction based on the fact that both are linked to cross-linking activity. However, a common observation is that cross-linking does not occur directly between the TG2-binding partners in either the extracellular or intracellular domains. The reason cross-linking does not happen is that, externally, various binding proteins form clusters that may block the TG2 cross-linking active site. Intracellularly, the calcium concentration is very low, which may block the final acyl transfer. Additionally, it has been shown that intracellular binding proteins predominantly exhibit either pro-death or pro-survival regulators. Based on these similarities, TG2 appears to have a unique role of cross-linking and distinct regulatory functions, with its N-terminus potentially serving as a critical regulatory target. This review examines its significance and considers its potential future applications.

Discovery of the cross-linking enzyme: The history of TG2

A peptide bond is a type of amide covalent bond that links two consecutive α-amino acids, connecting the C1 (carbon 1) of one amino acid to the N2 (nitrogen 2) of the next, along a peptide or protein chain (Fig. 1). Through this bond, a peptide forms a -C-C-N-C-C- (carbon-carbon-nitrogen-carbon-carbon) chain. This powerful covalent bond typically requires 2-4 kcal/mol of ATP for formation (Martin, 1998). Transglutaminase, however, attacks the γ-amido group of the glutamine residue in a protein with its active-site cysteine, cleaving off the amido group. It then catalyzes transamidation with the ε-NH₂ group of a lysine residue, forming a N-e-(γ-glutamyl)lysine cross-link (Folk, 1983; Folk et al., 1968). Therefore, the original name of TG is protein-glutamine γ-glutamyltransferase. However, in exceptional cases, hydrolysis may occur after glutamine is acylated onto TG, leading to deamination taking place immediately (Keillor et al., 2014) (Fig. 1). The fatal disease caused by this gliadin deamination is celiac disease (Dieterich et al., 1997). The bond formed between two molecules by TG surprisingly adopts the same C-C-N-C-C structure as a peptide bond; hence, it is termed as an iso-peptide bond. It is remarkable that this strong covalent bond forms without using any ATP; only calcium is required (Folk et al., 1968). These enzymatic activity results suggest we must look beyond ATP-based signaling and the kinase doctrine that has dominated research for the past fifty years. They indicate that beneath the development of complex and sophisticated signaling lies a defense system that has existed since the early days of invertebrates.

Fig. 1.

The reaction of transglutaminase (TG). TG primarily catalyzes the formation of an isopeptide bond between the γ-carboxamide group on the side chain of a glutamine residue and the ε-amino group on the side chain of a lysine residue in nature. Lysine and glutamine residues must be bound to a peptide or protein for this cross-linking reaction (between different molecules) or intramolecular reaction (within the same molecule) to occur. TG can also deamidate the glutamine residue of a protein in the presence of water, converting it into a glutamic acid residue.

The first discovery of the TG family in the human body was made through blood coagulation studies. In the late 1940s, while investigating blood clotting processes, the existence of a substance that further stabilizes the fibrin clot formed by platelets was proposed. Initially, the precise role of blood clotting factor XIII (FXIII) was unclear. It wasn’t until the 1950s that scientists discovered FXIII, an enzyme that TG chemically links fibrin strands, increasing rigidity and stabilizing the clot. In 1964, Lorand et al. isolated and purified FXIII (Lorand and Konishi, 1964). Over the subsequent 50 years, its structure and function were elucidated (Byrnes et al., 2016; Yee et al., 1994). FXIII is an enzyme involved in the final stage of blood clotting that exists in an inactive form. It must be activated by thrombin cleaving its N-terminus to become active. However, efforts to clarify the detailed fibrin cross-linking mechanism in blood clotting are still ongoing.

Later, it was found that the enzymatic activity behind these chemical reactions exists not only in blood but also in tissues. In 1966, Dr. Folk’s group first isolated it from guinea pig liver and identified it as tissue transglutaminase (Folk and Cole, 1966). An enzyme performing a similar function in liver tissue, unrelated to blood coagulation, was first discovered and named liver TG. However, as TG2 was later found in other tissues, it was designated tissue TG. Unlike extracellular FXIII, TG2 is intracellular and is activated immediately. Therefore, it was called cytosolic TG (or TGase C) (Lee et al., 2000). Subsequently, upon being the second to complete its genetic sequence analysis, it was named the TGM2 gene (TGM1, Jan 1991 (Kim et al., 1991); TGM2, Apr. 1991 (Noguchi et al., 2001)). There is a movement to standardise the protein as TG2 and the gene as TGM2. Historically, the term “TGase 2” was consistently used for the TG2 protein, following Dr. Folk’s proposal to append the “-ase” suffix to TG. However, I will henceforth standardize TG2, acknowledging its function as a chaperone protein in addition to its cross-linking enzyme activity.

The dawn of the physiological role of TG2

Observing the blood-stabilizing effect of coagulation, researchers initially theorized that TG2’s inherent physiological role was protein stabilization, and that excessive expression would promote protein coagulation, thereby causing pathological lesions. A representative phenomenon is scar formation during wound healing, and it has been reported to play a significant role in the mechanisms of senile plaque formation in dementia (Junn et al., 2003), the formation of apoptotic bodies (Tatsukawa and Hitomi, 2021; Wan et al., 2002), and cataract formation (Wan et al., 2002). Recent studies have shown that TG2, along with FXIII, is involved in blood coagulation. Furthermore, it has been discovered that this coagulation mechanism promotes extravascular fibrin accumulation in acute liver injury (Wei et al., 2025). Moreover, researchers focused on identifying the substrates of cross-linked TG2 to elucidate its physiological significance (Ruoppolo et al., 2003). Recently, the homology of the TG2 cross-linking site has also been utilised to identify TG2 substrates using bioinformatics tools (Damnjanovic et al., 2022). First, an extensive library of peptides is generated and displayed using cDNA display. Then, this library is incubated with the target transglutaminase, and the peptides that bind or react are selected. These selected peptides are amplified, and their corresponding cDNA is sequenced using NGS. Finally, bioinformatics analysis of the NGS data reveals the specific amino acid sequences and properties that the transglutaminase prefers, providing a detailed substrate profile (Damnjanovic et al., 2022). The TG2 research examined its physiological characteristics based on its substrate, and it can be broadly divided into the extracellular and intracellular domains.

PRIMORDIAL ROLE OF TG2

Horseshoe crab TG

Approximately 250 million years ago, TG in horseshoe crab lymphocytes served as a primordial defense system (Kin and Blazejowski, 2014) (Fig. 2). This system would gelify to form a barrier should the crustacean’s shell rupture (Matsuda et al., 2007) or gelify to isolate bacteria when they invaded the bloodstream (Wang et al., 2010). The blood cells circulating within the Limulus blood contain a coagulation system involved in both hemostasis and defence against invading microorganisms (Muta et al., 1990). This coagulation system is highly sensitive to lipopolysaccharide (LPS), a Gram-negative bacterial endotoxin (Muta et al., 1990). Coagulation proteins are secreted at the wound site, TG adheres, and gelation occurs immediately (Matsuda et al., 2007). At this point, TG itself becomes a component that undergoes gelation. TG2 itself is incorporated into polymers and also induces cross-linking (Stamnaes et al., 2015). This phenomenon is likely attributable to TG’s inherent defence mechanism, which has persisted since ancient times. This physiological response is still utilised within the medical field. It is known as the endotoxin test. This test method, called the LAL (Limulus Amebocyte Lysate) test, is used to verify whether a patient’s wound site, surgical instruments, or injectable preparations are contaminated (Zhang et al., 2025). This test utilises lymphocytes from horseshoe crabs. It employs LAL protein extracted from amebocytes, lymphocyte-like cells found in horseshoe crab blood. This protein includes both a pathogen-binding protein and TG, which crosslinks pathogens. When pathogens are present in this solution, they cause immediate whitening and gelation. Therefore, ancient TG likely served as a survival defense mechanism. TG appears to have evolved with an innate ability to promote survival by moving to injury or damage sites, where it encapsulates and inhibits infiltrating substances, effectively ‘suturing’ the wound. Therefore, it appears that a primitive immune system was established by binding foreign substances entering the wound site with TG, then using proteolytic enzymes to break down these protein aggregates and eliminate hazards. Consequently, TG was designed to exhibit exceptional binding affinity for extracellular matrix (ECM) proteins, and cross-linking appears to have evolved as calcium-dependent, activating only when necessary. Evidence of this classic structure is visible in its tertiary structure. A protein tertiary structure very similar to that of the FXIII active site was identified and reported. It was none other than the active site of the papain protease (Pedersen et al., 1994). Although FXIII and cysteine proteases differ considerably in size and overall protein structure, their active sites and surrounding protein regions share structural similarities, indicating a common evolutionary origin. The triad of Cys277, His335, and Asp358 in TG2 (Lee and Park, 2017) provides a nearly identical tertiary structure to the triad of Cys25, His159, and Asn175 in cysteine proteases (Pedersen et al., 1994). Cysteine proteases, such as papain, trace their origins to unicellular organisms. TG has also been found in Streptoverticillium cinnamoneum (Duran et al., 1998). Furthermore, the TG of unicellular organisms is about half the size of that in vertebrates, is calcium-independent, and exhibits low substrate specificity (Yang et al., 2011). Given these structural characteristics, it is highly likely that TGs in vertebrates evolved from the cysteine protease family. Chemically, the reaction proceeds in the exact opposite direction. The extension of both the N-terminus and C-terminus of TG2 probably enables not only acylation in addition to hydrolysis, but also substrate selectivity. Therefore, proteins binding to the TG2 N- or C-terminal regions appear to have evolved either to stabilize or enhance each other’s functions, or to facilitate protein transport. However, it is astonishing that while vertebrates have continuously developed increasingly complex immune systems over 250 million years, the horseshoe crab has survived into the 21st century without evolutionary compromise, relying solely on its primitive defense system centered around TG.

Fig. 2.

The evolutionary conservation of transglutaminase (TG) function from the horseshoe crab (Limulus) to human cancer cells. (A) Caraxin is a protein essential for cuticle formation, which is cross-linked by TGase to form a network in arthropods (Matsuda et al., 2007). The coagulation network is triggered through hemocyte secretion induced by bacteria infection, resulting in a clot in associatetion with coagulin and caraxin polymer to protect injury site. Therefore, TG activation seals wound sites in crustaceans, protecting them from bacterial infection. In primordial life, TG-dependent protein cross-linking may participate in the initial stages of host defense beneath the epidermis, similar to mammalian skin. (B) Schematic proposing the chaperone function of TG2 in cancer cells. When cancer cells are exposed to the chemical toxicity of anticancer drugs, damaged cells induce death signals such as p53. Surviving cells suppress death signals by inducing TG2 (Kang et al., 2016), and promote cell growth by binding tumor suppressors like Glycogen Synthase Kinase3β (GSK3β) to p62 (paper under revision), transporting them to autophagosomes for removal. Therefore, cancer cells acquire drug resistance, allowing them to grow freely from anticancer drugs. To summarize, TG2 plays a role in repairing damaged tissue, but cancer cells exploit this to acquire anticancer drug resistance in animals.

Both extracellular and intracellular roles of TG2 are regulated by calcium

Calcium greatly influences the second chemical reaction of TG2, which is hydrolysis. When calcium levels are high, TG2 reacts more rapidly (Folk et al., 1968). Therefore, in vitro assays in laboratories usually use calcium concentrations of 1 mM or higher (Folk et al., 1968; Scarpellini et al., 2009; Sewa et al., 2024). In the case of FXIII, since blood calcium levels are already high enough to activate TG, this explains why FXIII remains in its inactive precursor form. Normal total serum calcium levels range from 2.20 to 2.70 mM (Goldstein, 1990). Approximately 40% of total blood calcium is bound to plasma proteins, mainly albumin. The other 60% includes ionized calcium and calcium bound to phosphate and citrate. The normal range for ionized serum calcium is usually 1.17-1.30 mM (Yu et al., 2021). This is a calcium concentration sufficient to activate TG. Therefore, these TG family members circulating outside the cells must exist as precursors or be bound to proteins to remain safe. Besides FXIII, TG1 and TG3 must also be present as precursors to form the skin’s cornified envelope outside the cell, allowing delicate cross-linking. However, active TG2 in the cytoplasm is exposed to only about 100 nM of intracellular calcium—one ten-thousandth of the concentration needed for activation—and therefore may not need to be a precursor (Piper et al., 2002). However, in organelles like the ER or lysosomes, calcium levels are usually high, approximately 500 μM—about 5,000 times that of the cytoplasm (Christensen et al., 2002; Henderson et al., 2015). Consequently, whenever the cell sustains injury or stress, even a slight increase in calcium can trigger protein-protein cross-linking (Shabir and Southgate, 2008). Transacylation of primary amine and hydrolysis of the carboxamido group of glutamine show a widely different calcium ion requirement range (Folk et al., 1968). Like all other TGs, the catalytic activity of TG2 requires calcium with apparent Ka (competitive) and ka (noncompetitive) values of ~1 mM (Piper et al., 2002). Here, it is crucial to consider lysosomes carefully, as when autophagy occurs, the autophagosome fuses with the lysosome, triggering protein clearance via autolysis. Consequently, proteins binding to TG2 in the cytosol utilise TG2 as a shuttle to move to subcellular compartments; once transported, they undergo both co-localization and degradation. We previously discovered that the C-terminus of TG2 binds to p62 (Kang et al., 2016) (Fig. 2).

EMERGENCE OF TG2 ROLES IN CANCER PROMOTION

Proposed roles of TG2 in the extracellular domain

Although research into linking TG2 cross-linking activity to its cellular role has faced challenges, progress has been made in understanding the roles of extracellular TG2. Dr. Belkin’s 2011 review thoroughly explored the different roles of TG2 in the extracellular matrix (Belkin, 2011). TG2 is commonly defined as playing a role in enhancing cell adhesion and cell surface signalling outside the cell (Belkin, 2011). TG2 roles include binding to fibronectin to promote its incorporation into the ECM (Akimov and Belkin, 2001a), and TG2 binds to integrins and fibronectin without cross-linking to form a stable ternary complex, thereby participating in cell adhesion (Akimov and Belkin, 2001b). The N-terminus of TG2 (residues 202-222) binds to heparin/syndecan-4, facilitating its efficient interaction with integrins. This binding activates PKCα signalling via outside-in signalling (Telci et al., 2008; Wang et al., 2012). Syndecan-4-dependent PKCα regulation promotes focal adhesion formation and migration in cells that adhere to fibronectin via α5β1-integrin (Morgan et al., 2007). Furthermore, regarding angiogenesis, TG2 binds TGF-β1 to the extracellular matrix, thereby activating Smad and EMT signaling pathways in cells (Wang et al., 2017). This excessive accumulation of TGF-β has been identified as the primary mechanism causing fibrosis (Olsen et al., 2011).

Additionally, low-density lipoprotein receptor-related proteins (LRP) 5 and 6 bind to TG2 to activate β-catenin signalling (Faverman et al., 2008). Conventional β-catenin inhibitors block warfarin-induced β-catenin activation, calcification, and osteogenic pre-differentiation in vascular smooth muscle cell (VSMC) (Beazley et al., 2012). However, Wnt inhibitors fail to block β-catenin activation in warfarin-treated cells (Beazley et al., 2012). Vascular cells with TG2 knocked down or treated with TG2 inhibitors cannot activate β-catenin in response to warfarin (Beazley et al., 2012). This finding suggests that, in addition to contributing to signal amplification through Wnt ligand clustering, TG2 also has another regulatory mechanism that activates β-catenin signaling within cells. The interaction of TG2 with platelet-derived growth factor receptor (PDGFR) on the fibroblast cell surface indicates new implications. By binding to PDGFR and forming clusters, TG2 enhances PDGFR activation by PDGF, thereby increasing downstream signaling (Zemskov et al., 2009). TG2 promotes PDGF-induced activation of the PDGFR/Akt1 and β-catenin pathways, resulting in neointima formation in VSMC (Nurminskaya et al., 2014).

Endostatin is a C-terminal proteolytic fragment of collagen XVIII found in the vascular basement membrane zones of various organs. It inhibits angiogenesis and tumor growth by blocking vascular endothelial growth factor receptor (VEGFR) (Folkman, 2006). Among the binding proteins of endostatin, TG2 is one of the top binders (Faye et al., 2009). The GTP-binding site of residues 476-580 of TG2 is proposed as a high-affinity binding site for endostatin, with a nanomolar range (Faye et al., 2010). Therefore, if TG2 and endostatin bind with high affinity, endostatin cannot attach to VEGFR and thus cannot inhibit VEGF signalling. In other words, binding to TG2 prevents endostatin from inhibiting angiogenesis, thereby promoting VEGF signaling and accelerating blood vessel formation.

TG2 is clearly an intracellular enzyme, but most of its physiological functions have been studied outside cells or on cell surfaces. This is probably because high calcium levels outside cells activate TG2, making its cross-linking substrates more apparent. However, it is intriguing that TG2 promotes these binding molecules to form clusters rather than cross-linking. Although it is still debated, the cross-linking of two large substrates would likely require two TG2 molecules, each bringing a substrate, to form a TG2 dimer that can create the necessary cross-linking distance (Kim et al., 2017). Therefore, we proposed that the TG2 C-terminus domain is the major site for dimer formation (Kim et al., 2017). Perhaps in ECM, TG2 remains in its extended conformation due to high calcium levels, which might accelerate binding to ECM proteins after TG2 dimerization. The TG2-binding proteins are summarised in Fig. 3.

Fig. 3.

Significance of the N-terminus in four domains of TG2. We have demonstrated that small-molecule compounds such as Streptonigrin (Lee et al., 2018) and GK921 (Kim et al., 2018), which bind to the N-terminus, exert physiological effects by dissociating the protein to which they bind. Among these, p53 (Kang et al., 2016) and GSK3β (paper under revision), which are involved in survival and proliferation, are TG2 N-terminus-binding proteins that directly regulate the death and growth of cancer cells. Interestingly, the region to which they bind is the same region that fibronectin binds (residues 88-106) (Hang et al., 2005; Hoffmann et al., 2011), and in its vicinity lies the region to which heparan sulphate (syndecan) binds (residues 262-265) (Lortat-Jacob et al., 2012). There are two calcium (Ca⁺⁺) binding sites in TG2, designated as S1 and S3 (Sewa et al., 2024). S1 regulates formation of an inhibitory disulfide bond in TG2, while also establishing that S3 is essential for g-glutamyl thioester formation (Sewa et al., 2024). The β-barrel 1 domain of TG2 contains multiple binding hot spots of residues 476-583 as a GTP/GDP-binding site (Jang et al., 2014), ATP binding site (Han et al., 2010), and endostatin binding site (Faye et al., 2010). TG2 resides as a single stable folded form at 20°C, but transforms to TG2 dimeric stretched form at 37°C (Kim et al., 2017). The dimeric binding domain of TG2 was determined by the tandem mass spectrometry analysis (Kim et al., 2017). The C-terminus of residues 593-600 denotes the region facing inter-dimeric TG2 at 37°C (Kim et al., 2017). The C-terminus of TG2 also contains p62 bonding sites (residues 460-687) (Kang et al., 2016). Crystal structures of folded and unfolded TGase 2 are shown as ribbons. The N-terminal β-sandwich is shown in blue (N), the catalytic domain (Core) in green, and the C-terminal β-barrels (β1 and β2) in yellow and red, respectively. GDP-bound TG2 (left). The GK921 binding-mediated masking region of TG2 is residues 81-116 (red ribbon area) (Kim et al., 2018). 3D structure model was obtained from the original article (Pinkas et al., 2007) under open access license “CC-BY”, and the original article (Kim et al., 2018) under under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licen ses/by/4.0/), which permits unrestricted use.

Discovery of the intracellular roles of TG2

TG2-mediated NF-κB activation: We discovered that TG2 activates NF-κB via a novel pathway (Lee et al., 2004). TG2 induces the polymerisation of I-κBα, an inhibitor of NF-κB, rather than promoting its phosphorylation and degradation (Lee et al., 2004; Park et al., 2006). This polymerisation induces the release and nuclear translocation of NF-κB, which can upregulate the expression of numerous inflammatory genes in the nucleus, including inducible nitric oxide synthase and tumour necrosis factor-α (TNF-α) (Kim, 2011). This discovery introduces a new paradigm, distinct from the traditional understanding of gene expression based on signaling pathways. It shows that a system capable of controlling vital biological signals without signaling pathways already exists.

Numerous reports have indicated that TG2 is closely associated with inflammation, yet establishing the functional link proved challenging. While searching for this connection, we found that TG2 activates NF-κB by removing I-κBα. This is usually explained by the presence of an NF-κB transcription factor binding site on the TG2 promoter; thus, when NF-κB increases in response to inflammatory signals, TG2 expression rises (Gundemir et al., 2012). However, the crucial point is that when NF-κB is activated, the first protein it produces is I-κBα. Consequently, within 30 minutes of NF-κB activation, NF-κB is inactivated by I-κBα (Park et al., 2011). However, NF-κB activity actually persists for several hours and gradually diminishes (Park et al., 2011). This phenomenon is termed the “resolution phase” of inflammation and occurs as I-κBα decreases through TG2 expression. I previously proposed this as the “Ouroboros theory” (Kim, 2011; Park et al., 2011). This resolution phenomenon is vital for a normal inflammatory response. Without this system, the body cannot defend against external pathogens, leading to potential death. An experimental example has documented this. In Drosophila larvae, those with the Toll and imd pathway-dependent antimicrobial pathway (De Gregorio et al., 2002) were tested for bacterial infection after TG deficiency alone (Wang et al., 2010). Larvae with decreased TG levels showed a significantly higher mortality rate after septic shock (Wang et al., 2010). The larvae with TG depletion were also more vulnerable to natural infections caused by insect-pathogenic nematodes and their symbiotic bacteria (Wang et al., 2010). These findings clearly demonstrate that TG is crucial in early innate immunity, helping to prevent septic infections. This delayed inflammatory phase is part of the resolution process. However, if this phase lasts too long, it can develop into a disease in vertebrates. When inflammatory cytokine TNF-α is applied to tissue exhibiting elevated TG2 expression due to causes other than inflammation, NF-κB does not switch off but instead burns fiercely (Park et al., 2011). This phenomenon is observed in chronic inflammatory diseases such as sporadic inclusion body myositis (Choi et al., 2000). These discoveries revealed that TG2, once believed to be a protein at the bottom of the signaling cascade, actually controls transcription factors at the top of the signaling hierarchy.

TG2 is p53 suppressor – TG2 is autophagy activator: RCC exhibits an abnormally high incidence of wild-type p53, with a low mutation rate of less than 4% (Girgin et al., 2001). It is known that mouse double minute 2 homolog (MDM2) prevents cancer cell death by promoting p53 degradation. However, we discovered that p53 protein levels in renal cell carcinoma are regulated by autophagy-mediated degradation (Ku et al., 2013). TG2 has been confirmed to be involved in p53 degradation via this pathway (Kang et al., 2016). TG2 knockdown increased p53-mediated cell death in RCC. Furthermore, p53 is rescued by TG2 knockdown but is not rescued by MDM2 knockdown (Kang et al., 2016). We determined that the process by which p53 is eliminated in RCC cells occurs entirely through TG2-mediated autophagic degradation, rather than via MDM2-mediated proteasomal degradation (Kang et al., 2020). That is, in RCC, the p53 suppressor is TG2. MDM2 is the p53-induced feedback protein. At this point, TG2 was observed to bind to p53 at its N-terminus, while p62 binds to its C-terminus, leading to its transport to the autophagosome (Kang et al., 2016). MDM2-binding domain (residues 15-26) of p53 interacts with the N-terminus of TG2 (residues 1-139). In the intracellular domain, p62, which induces autophagy, binds to the TG2 C-terminus. The PB1 (Phox and Bem1p-1) domain of p62 (residues 85-110) directly interacts with the β-barrel domains of TG2 (residues 592-687) (Kang et al., 2016). Thus, TG2 binding to p53 results in the loss of p53 function of inhibiting autophagy among its various functions (Kang et al., 2019). In addition to RCC, knockdown of TG2 in lung, colorectal, and breast cancer cells with wild-type p53 leads to p53 reactivation, rendering these cells susceptible to anticancer drugs (Kang et al., 2019). However, when TG2 expression increases, we observe decreased p53 levels, increased LC3 levels, and increased anticancer drug resistance (Kang et al., 2019).

TG2-mediated GSK3β suppression in ovarian cancer – β-catenin activator: TG2 is recognized as one of the genes most significantly upregulated during the epithelial-mesenchymal transition (EMT) process in ovarian cancer (Kumar et al., 2010; Lentini et al., 2013). Drug-resistant ovarian cancer cells exhibit activated EMT, allowing them to evade chemotherapy, a phenomenon linked to higher TG2 expression (Blaheta et al., 2025; Kumar et al., 2010). Dr. Matei and colleagues reported on the diverse roles of TG2 and its physiological regulatory functions through extensive research over a long period; however, the direct regulatory factors of TG2 within cells remained unidentified. Recently, we discovered that TG2 promotes epithelial-mesenchymal transition by directly binding to glycogen synthase kinase-3β (GSK3β) and thereby stabilizing β-catenin (paper under revision). The TG2 N-terminus interacts with the middle domain of GSK3β, inducing the autophagic degradation of GSK3β. Within cells, GSK3β was found to bind to the N-terminus of TG2 (residues 1-139) (paper under revision). Following binding to the TG2 N-terminus, p62 binds to the TG2 C-terminus, leading to its transport into autophagosomes for degradation (paper under revision). Consequently, treatment with an autophagy inhibitor, chloroquine rescues GSK3β. When GSK3β binds to TG2, β-catenin signaling is activated, which increases EMT. In essence, TG2 acts as a chaperone protein that suppresses GSK3β, thereby activating β-catenin and promoting cellular EMT. Therefore, identifying and using a small molecule that binds to the TG2 N-terminus to inhibit this interaction could free GSK3β and reduce β-catenin activity.

TG2-mediated pVHL suppression under normoxia – HIF-1α activator: The activation of hypoxia-inducible factor-1α (HIF-1α) under hypoxic conditions was well explained. Under normal oxygen levels, HIF-1α is hydroxylated by prolyl hydroxylase, which is recognized by the E3 ligase von Hippel–Lindau tumor suppressor (pVHL), leading to ubiquitination. The proteasome then degrades HIF-1α. However, under low-oxygen conditions, reduced hydroxylation activates HIF-1α, prompting it to produce essential factors for vascular growth. As a result, HIF-1α is recognized as a key regulator of angiogenesis. This raises the question: Does HIF-1α activation also occur under normal oxygen levels? A notable example is seen in cancer cells. Treatment with anticancer drugs has shown a significant increase in HIF-1α within these cells (Cao et al., 2013). Can the administration of anticancer drugs induce hypoxia only in localised cancer cells with pinpoint accuracy? If not, what mechanism could activate HIF-1α? Is HIF-1α simply a regulatory factor required solely for angiogenesis? No, it is not. Downstream of HIF-1α regulation lie angiogenesis, alongside factors related to cell survival such as c-Myc, cyclin D1, VEGF, Insulin-like growth factor 2 (IGF-2), TGF-α, and metabolic processes including glucose transport and glycolysis (Kao et al., 2023). If such diverse senescence phenomena are regulated, hypoxia alone does not regulate HIF-1α . The TG2 promoter contains a position at -367 where a hypoxia response element can bind (Gundemir et al., 2012). Hypoxia directly regulates TG2. So, what role does TG2 play? It naturally activates NF-κB for survival or reduces p53 to prevent cell death. However, the most crucial vascularisation occurs via HIF-1α. Several reports have documented that TG2 induces HIF-1α activation, which has been explained either through NF-κB-mediated activation (Kumar and Mehta, 2012) or through regulation of p53. (Lee et al., 2020). We discovered that TG2 depletes pVHL under normoxic conditions, thereby activating HIF-1α, and demonstrated this mechanism (Kim et al., 2011). At normal oxygen levels, when HIF-1α is induced by anti-cancer drug treatment, TG2 is activated to remove pVHL (Kim et al., 2011). In various cancers, TG2 levels have been reported to increase following administration of anticancer drugs (Antonyak et al., 2004; Cao et al., 2008; Kang et al., 2019; Kim et al., 2006). We demonstrated in vitro that TG2 polymerizes pVHL upon calcium addition, but it is not clear whether TG2 transfers pVHL to the autophagosome without cross-linking. This is because, regardless of how much TG2 and pVHL are overexpressed, no polymer forms, whereas its downstream target HIF-1α increases (Kim et al., 2011). Furthermore, the polymerisation observed upon treatment with a proteasome inhibitor suggests that these two molecules bind at a specific site to form a polymer (Kim et al., 2011). That is, TG2 and pVHL may bind at the N-terminus, subsequently translocating to autophagosomes or proteasomes, where they polymerise before undergoing degradation.

TG2-mediated induction of stemness and prevention of cell death: Dr. Park found that TG2 expression rises in the perinecrotic zone of glioblastoma (GBM), prompting mesenchymal transdifferentiation of glioblastoma stem cells (GSCs) by controlling key transcription factors (TFs) such as C/EBPβ, TAZ, and STAT3 (Yin et al., 2017). TG2 expression was induced through NF-κB activation by TNF-α from macrophages and microglia. TG2 promoted mesenchymal transcript expression by removing DNA damage-induced transcript 3 (GADD153), thereby inducing C/EBPβ expression (Yin et al., 2017). This study did not explore which TG2 domain interacts with GADD153; however, since GK921 reactivates GADD153, GADD153 and TG2 are likely to bind to the TG2 N-terminus. In TG2 knockout mice, cell death proceeds normally, and the phenotype remains unchanged. However, when exposed to cell death stress induced by tumor necrosis factor-α (TNF-α), embryonic fibroblasts (MEFs) from TG2 knockout mice showed a stronger response to cell death compared to MEFs from wild-type mice (Kim et al., 2013). This study identified cathepsin D (CTSD) as the TG2-binding protein via LC/MS (Kim et al., 2013). The interaction between TG2 and CTSD caused CTSD depletion through cross-linking in vitro and in MEFs, which decreased cell death. In other words, TG2 acts as a barrier to prevent cell death.

Substrate specificity: beyond cross-linking target: TG family enzymes do not randomly cross-link glutamine and lysine residues in all proteins. Even when glutamine and lysine constitute one-tenth of the total amino acids, as in BSA, TG2 does not form any cross-links (Scarpellini et al., 2009). No matter how much calcium is added and how long the incubation with TG2 is prolonged, cross-linking does not occur (Scarpellini et al., 2009). However, heparan sulfate is a vital step in TG2 cross-link formation because it binds to TG2. Within this bound structure, glutamine and lysine residues must be accessible to the active site. While the sequence similarity near Q or K in the secondary structure is significant, it indicates that the tertiary structure, especially the accessibility of Q or K after TG2-substrate binding, is even more critical.

A good example illustrating the distinctiveness of this mechanism is available. It has been reported in several studies that NF-κB is activated when I-κBα is polymerised by TG2 (Kim et al., 2006; Lee et al., 2004; Park et al., 2006). However, it was demonstrated that I-κBβ, despite possessing very high sequence homology with I-kBα, does not become a target of TG2 (Kim et al., 2008). It specifically indicates that glutamine and lysine, which are the TG2 targets of I-κBα, are substituted with other amino acids at their respective positions (Kim et al., 2006). However, there may be a fact we are overlooking. The ankyrin domain, which repeats six times in I-κBα (Park et al., 2006), is deleted approximately four times in I-κBβ (Kim et al., 2008). This suggests that TG2 and I-κBα may bind to each other via this domain (Kim et al., 2006). Regrettably, we did not analyze which region of TG2 I-κBα binds to. If substrate binding is the first step in TG2 cross-linking activity, then N-terminus binding substrates seem especially important. This is because we hypothesize that the binding occurs at the N-terminus, pushing the substrate toward the active site located in the middle domain. Additionally, even if cross-linking does not happen, N-terminus-binding proteins become targets that TG2 transports to autophagosomes. Therefore, targets whose protein levels are controlled by TG2, such as I-κBα, are very likely to bind to the N-terminus of TG2. This common physiological process of TG2 could lead to a new understanding of TG2-mediated protein transfer.

N-TERMINUS DOMAIN OF TG2

Let us now summarise once more only those TG2-binding proteins and TG2 inhibitors reviewed earlier that bind to the N-terminus (Fig. 3). Notably, FXIII, a member of the TG family, possesses a fibronectin binding site at its N-terminus (Hoffmann et al., 2011). The extracellular domain of TG2 also binds to fibronectin (Hang et al., 2005; Hoffmann et al., 2011). The residues 88-106 of TG2 are proposed to be the major binding site of fibronectin (Hang et al., 2005). Fibronectin is a crucial component of ECM formation, stabilizing its attachment to the ECM by binding to integrins and various growth receptors. Extracellularly, the N-terminus of TG2 (residues 202-222) binds to syndecan-4, facilitating its effective interaction with integrins. The N-terminus of TG2 (residues 202-222) binds to heparin/syndecan-4, facilitating effective integration with integrins. This interaction activates PKCa signalling via outside-in signalling (Telci et al., 2008; Wang et al., 2012)

Among the proteins that bind to TG2 within cells, p53 and GSK3β bind to the N-terminus of TG2. Both p53 and GSK3β have been shown to bind to the N-terminus of TG2 (residues 1-139). p53 and GSK3β are among the proteins that bind to TG2 in the cell, both binding to the TG2 N-terminus (Kang et al., 2016). Furthermore, proteins bound to the TG2 N-terminus are transported to autophagosomes for degradation via p62 binding to the TG2 C-terminus (Kang et al., 2016). Consequently, p53-related downstream cell death is reduced, and β-catenin signalling is activated, leading to increased epithelial-mesenchymal transition (EMT).

Interestingly, among the TG2 inhibitors discovered, GK921 (Kim et al., 2018) and Streptonigrin (Lee et al., 2018) possess no reactive moiety capable of attacking the cysteine within TG2’s active site (Fig. 3). Moreover, it appears to inhibit TG2 primary activity by directly binding to it rather than its cross-linking activity, and to alter TG2 structure, thereby suppressing its cross-linking activity (Kim et al., 2018; Lee et al., 2018). GK921 appears to bind to TG2 at positions 81-116 of TG2 (Kim et al., 2018) (Fig. 3). GK921 alone, by interfering with TG2 and p53 binding in a renal carcinoma animal model, activated p53 and demonstrated a highly favourable anticancer effect (Ku et al., 2014). Streptonigrin was found to bind to TG2 residues 95-116. In a renal carcinoma animal model, it showed strong anticancer effects when given alone by disrupting TG2-p53 binding, which activates p53 (Lee et al., 2018).

As summarized in Fig. 4, the proteins bound and degraded by TG2 within the cell are either suppressors that inhibit transcription factors promoting cell growth, EMT, or survival, or proteins that promote cell death, such as p53 or cathepsin D. Therefore, the N-terminus of TG2 may bind to numerous survival suppressor proteins, thereby enhancing the survival signal triggered by primary signaling pathways over an extended period. Even focusing only on the tumor suppressors identified so far that bind to the TG2 N-terminus, these are highly significant targets. However, many more targets may bind to the N-terminus beyond these. Furthermore, cancer cells seem to utilize TG2 to gain this functional advantage, creating pathways to evade the toxicity of anticancer drugs and persist.

Fig. 4.

Key intracellular functions of TG2 in cancer. TG2 induces EMT by β-catenin activation through GSK3β suppression. TG2 activates NF-κB by depleting I-κBα (Lee et al., 2004). TG2 induces HIF-1α and IGF-1R activation by suppressing pVHL (Kim et al., 2011). TG2 protects itself from cell death by reducing p53 (Ku et al., 2013) and cathepsin D (Kim et al., 2013). TG2 promotes stemness by inhibiting GADD153 (Yin et al., 2017). TG2 promotes angiogenesis through p53 suppression mediated by HIF-1α induction (Lee et al., 2020). GSK3β, Glycogen Synthase Kinase3β; NF-kB, Nuclear Factor k-light-chain-enhancer of activated B cells; I-kBa, NF-kB inhibitor a; pVHL, von Hippel–Lindau protein; IGF-1R, Insulin-like Growth Factor 1 Receptor; HIF-1a, Hypoxia-Inducible Factor-1a; GADD153, Growth Arrest and DNA Damage-inducible Protein GADD153; C/EBPb, CCAAT enhancer binding protein.

FUTURE DIRECTIONS OF TG2 RESEARCH IN CANCER THERAPY

Upon review, it is apparent that TG2’s intracellular role is more similar to that of a molecular chaperone than a cross-linking agent. Additionally, proteins bound to TG2 are involved in degradation processes, such as autophagy. Therefore, research on the role of TG2 in cancer should focus on studying its binding partners. First, proteins that bind to the TG2 N-terminus must be identified in cells from various tumor types. The main point is that TG2-binding proteins should be analyzed after cancer cells are treated with the primary therapeutic agent for that tumor type. This is because anticancer drug treatment not only increases TG2 expression but also alters signaling pathways, thereby altering the binding of different proteins, depending on the tumor. To understand TG2 role in primary treatment resistance in a specific cancer type, it is essential to identify the proteins binding to TG2 in that context. How TG2 promotes cancer can be complex and vary across different cancer types. However, given each cancer’s unique circumstances, TG2 is likely to have a primary, key function in response to anticancer drug treatment within each cancer type. For example, it might act as a p53 suppressor in renal cell carcinoma or as a GSK3β suppressor in ovarian cancer. Therefore, the most urgent task is to identify which proteins bind to the TG2 N-terminus across different cancer types.

CONCLUSION

The implications of this regulatory role are detailed in Table 1. TG2 intrinsically suppresses cell death and inhibits survival pathway suppressors—beneficial activities that cancer cells have hijacked. These cells have developed a complex drug resistance signaling system that exploits TG2’s natural function, enabling them to evade the effects of anticancer drugs and persist. This results in drug resistance. Consequently, combining an anticancer drug with a TG2 N-terminus-binding inhibitor could be a breakthrough in cancer treatment, overcoming drug resistance and greatly enhancing survival rates.

Table 1.

Intracellular TG2-binding proteins in cancer

Binding partner	TG2 binding site	Biological effect domain	Cancer type	Ref
GSK3β	81-116	β-catenin activation	Ovarian cancer	In revision
I-κBα	N/D	NF-κB activation NF-κB activation NF-κB activation NF-κB activation	Breast cancer Mantle cell lymphoma Non-small cell lung cancer Ovarian cancer	(Kim et al., 2006) (Zhang and McCarty, 2017) (Jeong et al., 2013) (Cao et al., 2008)
pVHL	N/D	IGFR, HIF-1α activation	Breast, Ovary cancers	(Kim et al., 2011)
p53	1-139	Survival increase	Renal cell carcinoma	(Kang et al., 2016)
p53	1-139	HIF-1α activation	Renal cell carcinoma	(Lee et al., 2020)
Cathepsin D	N/A	Survival increase	Normal cell	(Kim et al., 2013)
GADD153	N/A	Stemness increase	Glioblastom	(Yin et al., 2017)
p62	460-687	Autophagy activation	Renal cell carcinoma	(Kang et al., 2016)
TG2	593-600	TG2 dimer	Normal cell	(Kim et al., 2017)