AGR-AGR2 Interaction (Monomer vs. Homodimer)

The nature of AGR2 protein structure was not fully understood until the development of AGR2 crystal structure, which gives a clearer picture of AGR2 protein architecture. The nuclear magnetic resonance (NMR) crystallography showed that AGR2_41–175 can exist in homodimer structure through a motif EALYK at position E60–K64 (40). The salt bridge between the carboxylate group of E60 and the ammonium group of K64 forms the major interaction between the two AGR2 subunits in a reverse parallel fashion. Mutagenesis at this dimerization motif (E60A) abolishes the homodimeric structure. It is important to note that the dimerization interface of AGR2 is held far from the catalytic CXXS motif (opposing faces), where enhanced disulfide exchange in client proteins might occur. Although AGR2 has only one active cysteine residue (pseudo-thioredoxin domain) (Fig. 1), the homodimeric structure of AGR2 might provide equivalent redox capacity.

Alternatively, AGR2 can form a shaped homodimeric structure in which oxidation-dependent homodimerization occurs through C81-mediated disulfide bond formation (Fig. 1), which would re-orientate the homodimer into a different formation (49). Indeed, Clarke et al. (8) showed that AGR2 can homodimerize in a similar fashion via the covalent bond (C81–C81) oxidation upon chemical peroxide exposure. The mutation of the single cysteine (C81) to alanine prevents hydrogen peroxide-catalyzed homodimerization of AGR2. Additionally, mass spectrometry analysis reveals that low levels of a chemical oxidant promote an intermolecular disulfide bond through the formation of a labile sulfenic acid intermediate. However, higher levels of oxidant promote sulfinic or sulfonic acid formation, thus preventing covalent homodimerization of AGR2. The single cysteine (C81) of the AGR2 therefore acts as an oxidant-responsive moiety that regulates its tendency for oxidation and its monomeric-homodimeric state.

AGR2 homodimerization is also associated with its NH₂ terminus. The NMR study revealed that NH₂-terminal amino acid region 21–40 (Fig. 1) is unfolded (40). Deletion of the NH₂-terminal disordered region (AA 21–45) (Fig. 1) creates more stable AGR2 homodimer (18). Indeed, recombinant AGR2, having this deletion, demonstrated a spontaneous homodimer in solution, and this mutant showed enhanced binding to its client protein Reptin compared with wild-type AGR2 (18). AGR2 was also shown to be capable of promoting cell adhesion by its NH₂-terminal domain (AA 21 to 44) (Fig. 1). The monomeric AGR2 (E60A) or the native homodimeric AGR2 showed relatively similar cell adhesion properties, suggesting that homodimerization is not important for this property. As such, the NH₂ terminus could be an anchor, for example to the plastic surface (40), while the other part of AGR2 interacts with cell receptors through thioredoxin motif or the facilitation of cell adhesion.

Physiologically, this suggests that the monomer to homodimer equilibrium is important in AGR2 signaling. It also dictates the protein geometry and synergy between the homodimer interface and the catalytic thioredoxin motif, given that AGR2 has the sole C81 for thiol-disulfide interchange to assemble correctly folded disulfide bridges in client cargo proteins. High levels of homodimeric AGR2 might exist in the ER, which is important for cysteine-rich client protein folding. Recombinant AGR2 readily exists in homodimeric state in solution (8) consistent with the calculated monomer-homodimer equilibrium dissociation constant, K_d of 8.83 μM (40). This provides an indication that homodimeric AGR2 might be a predominant form in physiological condition. In addition, AGR2 can respond to oxidizing conditions, which could affect the homodimer and monomer equilibrium. The ER environment, especially in the ER of cancer, is enriched in redox signaling mediators that in turn modulate the reactive oxygen species (31). This creates a redox imbalance environment that can affect the monomeric and homodimeric equilibrium of AGR2 protein, and this could give rise to AGR2 oncogenic functions.

We recently showed that the control of AGR2 homodimerization participates in the establishment of pro-inflammatory responses, which suggests that the AGR2 homodimer would behave as a sensor that contributes to the regulation of ER proteostasis (34). Using a protein–protein interaction screen to identify cellular regulators of AGR2 homodimerization, we discovered TMED2 as a specific enhancer of AGR2 homodimerization. Furthermore, we have characterized the interaction in vitro and in vivo and shown that upon ER stress, AGR2 and TMED2 dissociate. Finally, by a molecular modeling approach, we identified K66 and Y111 amino acid residues are involved in the TMED2-AGR2 interaction [Table 1 (34)].

These parts of AGR2 function are yet to be fully understood. 1) What are the roles and ratios of monomeric and homodimeric AGR2 in normal and disease conditions (e.g., cancer)? 2) How does the oxidation and monomeric/homodimeric state of AGR2 impact on client protein binding and client protein maturation, especially cysteine-rich proteins like MUC2? 3) Could this affect the protein folding capacity of AGR2 and shift its function to chaperone activity? 4) How does homodimeric AGR2 contribute to the progression of diseases such as cancer?

Insight into Further AGR2 Interactions

Existing data show that AGR2 can bind to a wide range of proteins such as nuclear, cytosolic, and plasma membrane proteins. AGR2 interacting proteins validated using protein-protein interaction assays published in the literature largely belong to plasma membrane proteins (Table 1). This is not surprising since AGR2 is a PDI involved in the secretory pathway and therefore in the regulation of proteostasis, as we reviewed recently (9). Such a function is supported by the observation that AGR2 can act as a PDI molecule critical for mucin biosynthesis (39). Another example is the AGR2 binding partner, EGFR (Table 1), which can form mixed-disulfide bonds with AGR2. In this way, AGR2 controls EGFR cell surface delivery (10). This raises the question of how AGR2 can catalyze the folding, assembly, and trafficking of receptors such as EGFR to the cell surface. Moreover, one recent study showed that gene expression of EGFR ligand TGFA promotes the growth of mutant EGFR^L858R lung tumor cells. AGR2 was found to be the key regulator, as it was highly expressed in the TGFA-induced EGFR-mutant lung adenocarcinoma (55). AGR2 has also been shown to interact with cytoplasmic protein TAB2 (TGF-β activated kinase 1) (58), ER and Golgi protein TMED2 (34), mitochondrial protein UNG1 (32), and nuclear transcription factor hypoxia induced factor (HIF)-1α, and Reptin (30, 33) (Table 1). Some of AGR2 protein-protein interactions in Table 1 have been reviewed elsewhere (5, 7, 9).

Murray et al. (36) used an in silico study to expand AGR2 interactome by exploiting the sequence-specific peptide motif that binds to AGR2, which was previously identified using peptide-phage display. To our knowledge, AGR2 is the only PDI with such a specific peptide-binding activity. Attempts to define a consensus peptide binding site for AGR3 have been unsuccessful (data not shown). We hypothesize that if AGR2 can bind to this sequence-specific penta-peptide motif, it can also bind to proteins that have the same motif. To this end, a consensus AGR2 peptide motif was developed using synthetic peptide libraries. The consensus motif Tx[IL][YF[YF] was formed, and this motif has been used to mine the human protein database to search for proteins harboring a similar motif. The data revealed that this motif is enriched in membrane-associated proteins, which suggests a role for AGR2 in this type of protein. Interestingly, none of the AGR2 client proteins identified in previous studies overlaps with our database results, suggesting that this method is a first in terms of determining novel AGR2 client proteins and complements the classical protein-protein interaction (PPI) methods. One key receptor protein, EpCAM (Table 1), was validated as an AGR2 client protein using both in vitro and cell-based assays and maps of the binding interface using hydrogen deuterium mass spectrometry. The data suggest one plausible role of AGR2 in membrane trafficking, and thus in proteostasis, by ensuring the proper maturation of this class of protein type to their final destination through secretory pathway. However, it remains to be studied whether extracellular AGR2 (eAGR2) has extracellular activity for EpCAM, since the “docking motif” is located at the extracellular stalk of EpCAM. Other potential AGR2 proteins from this transmembrane database mining, including MKS3 (35) (see Fig. 3, TMEM67 (MKS3)), are yet to be fully validated. Consistent with this EpCAM link to AGR2, quantitative shotgun proteomics were used to demonstrate that AGR2 and EpCAM were among the top upregulated cancer-associated proteins, and immunohistochemistry in esophageal adenocarcinoma tissues showed high expression of both proteins, thus presenting a novel biomarker pathway for this type of cancer (38).

The ability of AGR2 to interact with receptor proteins is not surprising as AGR2 was shown to be present both in ER [intracellular AGR2 (iAGR2)] and on the surface [extracellular AGR2 (eAGR2)] of pancreatic cancer cells using immunofluorescence and flow cytometry (12). Localization of eAGR2 at the surface of the cell could suggest that AGR2 is attached to a client receptor protein. Moreover, eAGR2 can be secreted in the microenvironment and interact directly with the extracellular matrix (14). eAGR2 is also present in the extracellular mediums of pancreatic (46), colon (53), and lung (14) cancer cells; in the urine of prostate cancer patients (6); in the plasma of ovarian cancer patients (13); and in the serum of pituitary adenomas (54). In addition, atypical roles for eAGR2 have been reported in the tumor microenvironment (9). For example, the addition of exogenous AGR2 has been shown to promote angiogenesis through the regulation of hypoxia induced factor-1 (HIF-1) (23). We previously demonstrated that eAGR2 performs an extracellular role independent of its ER function as an important microenvironmental pro-oncogenic regulator of epithelial morphogenesis and tumorigenesis (14). Furthermore, eAGR2 can stimulate the insulin-like growth factor-1 receptor-induced cell proliferation, migration, and epithelial–mesenchymal transition (EMT) in breast cancer cells through interaction with membrane ER-α (29). In colorectal cancer, eAGR2 can promote migration and metastasis through Wnt11 signaling (53). This extracellular role of AGR2 has shifted the paradigm and current focus of how an ER-resident protein can exit the ER, be secreted to extracellular space, and still retain its pro-oncogenic effects.

AGR2 and Proteostasis

AGR2, as a PDI, is involved in the maturation of various proteins through forming disulfide bonds with mucins (Table 1) and directly regulating their processing and secretion. PDIs function to attain the highest quality and productivity of the ER and to therefore maintain ER protein homeostasis (also called proteostasis). In a healthy intestine, AGR2 acts as a critical modulator of intestinal homeostasis, presenting within the ER of intestinal secretory epithelial cells (34) and regulating production of intestinal mucins via the formation of mixed disulfide bonds (60). Mice lacking AGR2 expression are unable to produce intestinal mucins and are highly susceptible to colitis.

Recently, AGR2 has been shown to interact with UNG1 (Table 1) (32). UNG1 is important for initiating base-excision repair in mitochondria. Due to UNG1 being located in close proximity to the respiration chain, UNG1 is subject to oxidative damage. Liu et al. (32) demonstrated that overexpression of UNG1 enhances cell resistance to oxidative stress and protects mtDNA from oxidation. Thus, it has been suggested that the interaction of UNG1 and AGR2 stabilizes UNG1, through its PDI function, and enhances its enzymatic activity in DNA repair. However, more studies are needed to examine the nature and the physiological role of UNG1-AGR2 interaction.

We previously demonstrated that AGR2 represents a mechanistic intermediate between endoplasmic reticulum quality control and tumor development (7, 22). Thus, “AGR2 proteostasis” interactome in cancer could enhance proteostasis (Fig. 2), thereby allowing tumor cells to cope with abnormal protein production and secretion and contributing to the tumor development.

AGR2, a HIF-1α-Binding Stabilizer for Chemoresistance Properties

In breast cancer, AGR2 has been shown to bind and stabilize HIF-1α (Table 1) (30), stabilize HIF-1α and delay its proteasomal degradation. It appears that AGR2 regulates the chemical hypoxia-induced doxorubicin resistance in breast cancer cells, which explains the variable levels of chemoresistance in breast cancers. This further validates its role as a potential breast cancer therapeutic target. Hence, given the important role of AGR2/HIF-1α interactome in increasing chemoresistance in breast cancer, it would therefore be interesting to investigate the AGR2/HIF-1α pathway to develop strategies to block chemoresistance AGR2 properties (Fig. 2) in breast cancer.

AGR2 and Tumor Cell Dissemination

Reptin was identified by the yeast two-hybrid screen (Table 1) and validated as an interacting protein of AGR2 in human cells. Reptin is a highly conserved member of the AAA+ family that can be found in numerous multiprotein complexes linked to transcription, DNA damage response, and nonsense-mediated RNA decay (16, 24–27, 47). This protein is a member of the RuvBl1/2 superfamily containing ATP binding motifs and has the ability to form biologically relevant protein–protein interactions with proteins involved in cancer, including Myc, Tip60, APPL1, Pontin, and telomerase holoenzyme complexes (3, 4, 45, 47, 56). The ATP binding motifs of Reptin and the proposed substrate-binding loop of AGR2 are determinants that drive a specific complex between AGR2 and Reptin proteins (33). Reptin was also shown to be overproduced in primary breast cancer biopsy specimens, suggesting a role in cancer growth in vivo. Because Reptin can also function as a prometastatic transcription protein, future research will inform whether AGR2 uses its substrate-binding loop to chaperone Reptin into inactive and activated transcriptional states and/or whether the allosteric ATP-binding motifs of Reptin regulate AGR2 function as a prometastatic factor in cancer.

Orphan glycosyl-phosphatidyl-inositol-linked receptor, C4.4A, has been reported to interact with eAGR2 to induce cell proliferation, migration, and chemoresistance in pancreatic ductal adenocarcinoma (2). The secreted metastasis-associated GPI-anchored C4.4A protein and the extracellular domain of α-dystroglycan have been shown to directly interact with eAGR2, indicating potential mechanisms for eAGR2 in the promotion of tumor metastasis via the regulation of receptor adhesion and the interaction with the extracellular matrix (59, 60).

Interestingly, C4.4A is a structural homologue of the urokinase-type plasminogen activator receptor (uPAR), which, along with CD59, is the most closely related human homologue of Prod1. In salamander limb regeneration, eAGR2 interacts with a cell surface receptor on adjacent cells identified as Prod1. Prod1 is a member of the three-finger protein (TFP) superfamily that is GPI-anchored at the cell surface. Whether CD59 is responsible for transducing the effect of eAGR2 is at present unknown. Prostate basal cells are immuno-stained by anti-CD59 and AGR2 (31). Therefore, it is possible to speculate that cells without CD59 might be resistant to the effect of eAGR2. Thus, eAGR2 interacts with the cell surface to modulate adhesion and promote tumor cell dissemination.

A better understanding of the molecular mechanisms that promote the dissemination of tumor cells is essential for the development of novel detection and therapeutic strategies. For lung cancer, we demonstrated that AGR2 induced the EMT and contributed to cancer dissemination (14). As AGR2 promotes the dissemination of tumor cells, we therefore suggest that “AGR2 tumor cell dissemination” interactome targeting may permit new strategies for therapeutic management against tumor cell dissemination capacity.

AGR2 and Regulation, the Trafficking of Proteins Destined for the Secretory Pathway

It has been established that AGR2 interacts physically with the epithelial growth factor receptor (EGFR) within the ER and that this interaction is necessary before the receptor can progress to the Golgi apparatus and the plasma membrane (10). In pancreatic disease, it has been hypothesized that this AGR2-induced EGFR signaling forms a common and essential link between injury-induced tissue regeneration and the development of pancreatic cancer. AGR2 expression represents an early initiating event necessary for EGFR signaling, and thus serves as a potential target for the treatment of chronic and neoplastic pancreatic disease. In pancreatic ductal adenocarcinoma, AGR2 is found activated downstream of mutant Kras^G12D (11), and its important role in pancreatic ductal adenocarcinoma development was shown using both orthotopic and Kras-driven pancreatic cancer mouse model with the deletion of Smad4, a tumor suppressor gene frequently lost in the late stage (PanIN-3) of pancreatic carcinogenesis (11).

Screening the human proteome for proteins that harbor a specific peptide motif (Tx[IL][YF][YF]) revealed an EpCAM interaction (35). AGR2 binding to EpCAM seems to be in the ER, since the recombinant AGR2 and EpCAM binds in vitro. This recombinant protein is made in the bacterial system, which lacks glycosylation—an event before post-translational modification that supports the hypothesis that AGR2-EpCAM binding is in the ER. EpCAM forms a model client protein for AGR2, which suggests that AGR2 is involved in regulating the trafficking of specific proteins destined for the secretory pathway, as has been demonstrated for EGFR.

Thus, AGR2 not only plays an essential role in the secretory pathway via the “AGR2 proteostasis” interactome (see AGR2 and Proteostasis), but also via its protein trafficking function. Thus, a protein trafficking network controlled by AGR2 could be a future target for anti-cancer therapy.

AGR2-Enhanced VEGF and FGF Signaling

We demonstrated that AGR2 is a signaling molecule that is found outside cancer cells (eAGR2) and one that makes these cells more aggressive. Indeed, eAGR2 has several extracellular functions (9), including the promotion of angiogenesis- and fibroblast-coordinated tumor cell invasion. Indeed, eAGR2 has been reported to bind directly to VEGF and FGF2, are the major players in tumor angiogenesis, and enhance their activities (19). This enhancement is dependent on both the AGR2 self-homodimerization region and the signaling pathways of VEGF and FGF2. It has been shown, both in vivo and in vitro, that a monoclonal antibody targeting the AGR2 self-homodimerization domain can partially abrogate eAGR2 activity. Thus, eAGR2 acts at multiple signaling crossroads, suggesting that eAGR2 is potentially a therapeutic target.