Novel PROTACs targeting tissue transglutaminase (TG2) suppress tumorigenicity of ovarian cancer cells

Abstract

Tissue transglutaminase (TG2), a multifunctional enzyme involved in protein crosslinking through transamidation, fibronectin–integrin interactions and GTP hydrolysis, is upregulated in cancer. Due to its diverse functions, TG2 has been a challenging therapeutic target. Here, we investigate the use of PROteolysis TArgeting Chimeras (PROTACs) to degrade TG2 and inhibit its tumor-promoting functions in ovarian cancer models. We describe a novel family of VHL based PROTACs using a ligand that binds to the TG2 fibronectin interacting domain and a thiol ether PEG linker. Three structurally related PROTACs—P374, P404, and P405—induced significant proteasome dependent TG2 degradation at 24 hours (p < 0.05), with stable effects at 48 hours. These compounds also potently inhibited cell adhesion and migration (p < 0.005), outside-in signaling, and blocked TG2 enzymatic activity (p < 0.001). An unbiased evaluation using reverse phase protein array of P374-treated cells revealed 136 differentially expressed proteins, including protein networks related to cell adhesion and involved in extracellular matrix (ECM) interactions. P374 and P405 reduced omental colonization in vivo and P374 inhibited intraperitoneal tumor dissemination and growth. Visium HD based spatial profiling of human ovarian tumors identified TG2 as a highly enriched protein at the tumors invasive edge and the interface with the ECM. Together, our findings put forward novel TG2-targeting PROTACs which effectively degrade TG2, impair its functions, and block in-vivo tumor dissemination. These results highlight the potential development of TG2 degraders towards therapeutic targeting in ovarian cancer.

Introduction

Tissue transglutaminase (TG2) is a multifunctional enzyme that plays critical roles in both physiological and pathological contexts (1–3). A member of the transglutaminase family, TG2 catalyzes calcium-dependent post-translational modifications through transamidation, resulting in the formation of protein crosslinks. In addition to its enzymatic activity, TG2 functions as a GTPase and engages in various non-enzymatic interactions and also serves as a scaffold in intracellular signaling pathways (4). Under normal physiological conditions, TG2 contributes to wound healing, extracellular matrix (ECM) remodeling, apoptosis, and cell adhesion (5). Its dysregulation has been implicated in the pathogenesis of several diseases, including cancer (6, 7). TG2 is frequently upregulated in malignancy and associated with poor clinical outcomes. Elevated TG2 expression has been reported in several cancer types, including breast, pancreatic, glioblastoma, and ovarian cancer (OC) (8–10). TG2 promotes tumor cell survival, epithelial-to-mesenchymal transition (EMT), chemoresistance, stemness and metastasis (11–15). It is not fully understood which functions of the protein drive the pro-tumorigenic effects. TG2’s interaction with integrins and fibronectin protmoting cellular adhesion and migration has been implicated in its pro-metastatic function (12, 16). The enzymatic activity has been linked to remodelling of the extracellular matrix (17) and cancer-stroma interactions (18). The GTPase function is involved in chemoresistance and cancer stemness phenotype (19, 20). Its multifunctionality and involvement in diverse signaling networks make TG2 a challenging therapeutic target. Traditional approaches to inhibiting TG2 have encountered significant limitations, largely due to its conformational flexibility and multifunctional nature. Small-molecule inhibitors target only specific domains of TG2 (e.g., enzymatic core, fibronectin binding domain) (21–24) and as a result, not all of the protein functions are blocked. Therefore, strategies capable of eliminating TG2 protein in its entirety are needed (25).

PROteolysis TArgeting Chimeras (PROTACs) represent an emerging modality in targeted protein degradation. Unlike conventional inhibitors, PROTACs are heterobifunctional molecules that tether a target protein to an E3 ubiquitin ligase, resulting in its ubiquitination and subsequent proteasomal degradation (26, 27). This catalytic mechanism allows for potent and selective elimination of disease-related proteins, including those traditionally considered “undruggable.” PROTACs offer several advantages, including increased specificity, prolonged effects, and the ability to disrupt both enzymatic and structural roles of their targets (28). Our group previously characterized the first PROTACs capable of degrading TG2 (29).

Building on those findings, we report here the design of a novel family of PROTACs targeting TG2 and their effects in ovarian cancer cell lines and mouse models. Our results show that TG2-targeting PROTACs effectively degrade the protein, impair its enzymatic activity, alter interactions with the ECM, resulting in inhibition of cell adhesion and migration, and ultimately suppress the tumorigenic potential in-vivo. These findings underscore the therapeutic potential of PROTACs as a strategy for targeting TG2 in ovarian cancer.

Results

Synthesis of new PROTACs targeting TG2:

After having identified effective TG2 degraders in our prior efforts (29), we continued to make further structural modifications to improve their protein degradation activity. To this end, we focused on exploring additional linker structures as well as alternate VHL ligand attachment vectors, while retaining the MT4-based ligand (30) on which our previous degraders were based.

We modified our previously-published degraders by replacing the polyethylene glycol (PEG) linker with an alkyl chain (Figure 1, 723 and 741). This change led to a steep drop in activity and the resulting compounds lost all TG2 degradation activity (Supplementary Figure 1, A and B). PROTACs bearing the VH101 phenol E3 ligase recruiter have demonstrated potency in other degradation applications. (31, 32). With MT4 established as an effective TG2 ligand and PEG chains suitable as linkers, we combined the VH101 phenol moiety with these features and prepared a series of PEG- and phenol-linked VHL PROTACs (Figure 1, 350-351). However, these compounds also demonstrated little-to-no TG2 degradation by western blot (Supplementary Figure 1, C and D). More recently, PROTACs employing a VH101 thiol E3 ligase recruiter were reported to have improved degradation activity (33, 34). We therefore incorporated the VH101 thiol ligand into the design of our TG2 PROTACs using MT4 as the protein of interest recruiter. Synthesis of the thiol ether PROTACs incorporating PEG and alkyl linkers was completed, along with alkyl-linked compounds that were connected to the MT4 ligand via an amide (Figure 1). Among these compounds, we identified 3 compounds, P374, P404 and P405, that demonstrated significant TG2 protein degradation activity (Figure 1, lower right side). Schemas for synthesis of this series of TG2 PROTACs are included in Supplementary Methods Schemes S1–S5.

Figure 1. Structures of previously-reported TG2 inhibitor MT4 and new PROTACs.

Schematical representation of MT4, compounds 7 and 11, along with new compounds P374, P404 and P405 described in this manuscript.

PROTACs P374, P404, and 405 significantly reduce TG2 protein levels in OC cells:

To evaluate the protein degrading activity of the new compounds P374, P404, and P405, we performed Western blot analysis in two TG2 expressing ovarian cancer cell lines, SKOV3 and OVCAR5. All three compounds reduced TG2 protein expression at 24 h, and this effect persisted up to 72 h in both SKOV3 (Figure 2A, B) and OVCAR5 cells (Figure 2C, D). No significant TG2 degradation was observed in SKOV3 cells at 6 h with any of the compounds (Supplementary Figure 2A). Compounds P404 and P405 showed modest TG2 degradation at 6 h at 10 μM concentration in OVCAR5 cells (Supplementary Figure 2B). In contrast, the other compounds failed to demonstrate TG2 degrading activity (Supplementary Figure 2C–F). Structures and NMR spectra of the synthesized compounds are included in Supplementary Figures 3–24.

Figure 2: Compounds P374, P404, and P405 significantly reduce TG2 protein levels in ovarian cancer cells.

(A, B) Western blot analysis of TG2 protein levels in SKOV3 cells treated with compounds P374, P404, and 405 at 0, 0.1, 1, and 10 μM for 24 hours (A) and 72 hours (B). (C, D) Western blot analysis of TG2 in OVCAR5 cells treated with compounds P374, P404, and P405 at 0, 0.1, 1, and 10 μM for 24 hours (C) and 72 hours (D). (E, F) Quantification of TG2 protein levels over time (0, 6, 24, and 72 hours) in SKOV3 (E) and OVCAR5 (F) cells treated with 10 μM of compounds P374, P404, and 405. n = 3 replicates. TG2 level normalized against GAPDH, considereing control as 100%. (G, H) Cell viability of SKOV3 (G) and OVCAR5 (H) cells treated with increasing concentrations (0, 0.1, 0.3, 1, 3, 10, and 30 μM) of compounds P374, P404, and 405. n = 3 replicates. All values represent means ± SD. ****p < 0.0001 by 2 way ANOVA with Tukey post-test.

TG2 degradation induced by the compounds was both time- and concentration-dependent. Maximal degradation reached approximately 80% and was observed at 24 hours, with TG2 levels remaining suppressed through 72 hours following treatment at 10 μM in both cell lines (Figure 2E, F). However, compound-induced TG2 degradation was not associated with a significant reduction in cell viability in either SKOV3 or OVCAR5 cells (Figure 2G, H), consistent with the known functions of the protein (13, 16, 21).

TG2 degradation induced by PROTACs is proteasome dependent and affects enzymatic activity of TG2:

PROTAC activity requires the recruitment of an E3 ubiquitin ligase complex. To determine whether the activity of the newly discovered compounds depends on proteasome-dependent degradation, we used MG132, a proteasome inhibitor. Co-treatment with MG132 prevented PROTAC-induced degradation of TG2 by all three compounds in SKOV3 cells (Figure 3A–C), confirming that POTAC induced TG2 degradation is proteasome-dependent.

Figure 3: PROTAC-induced degradation of TG2 is proteasome-dependent and reduces its enzymatic activity.

(A, B) Western blot analysis of TG2 levels in SKOV3 cells treated with 10 μM of P374 (A), or P404 and P405 (B) for 24 h, with or without the proteasome inhibitor MG132 (20 μM, 26 h). (C) Quantification of TG2 protein levels from (A) and (B). n = 3 replicates. (D, E) Western blot analysis of TG2 in SKOV3 (D) and OVCAR5 (E) cells treated with a negative control analog of P374 (non-VHL-binding) for 6 and 24 h at 0.1, 1, and 10 μM. Quantification of protein levels is shown in lower panels. (F) Enzymatic activity assay in SKOV3 cells treated with 10 μM of compounds P374, P404, and P405. Treatment with MT4 (20 μM), VHL ligand (10 μM), EDTA was used as negative controls. Protein lysates from TG2 knockdown cells (shTG2) was used as positive control. n = 3 replicates. All values represent means ± SD. *p < 0.05; ***p < 0.001; ****p < 0.0001 by 1 way ANOVA with Tukey post-test.

To further validate this dependency, we designed a negative control compound based on P374, similar to our previous work (29). This control compound is the enantiomer of P374, presenting inverted configuration of the stereocentre, rendering it unable to bind to VHL and, consequently, unable to induce proteasomal degradation. This structural modification prevented TG2 degradation in both SKOV3 and OVCAR5 cells (Figure 3D, E), providing additional confirmation that TG2 degradation is mediated via the proteasome.

Given that TG2 is a multifunctional protein, we next sought to determine whether the PROTACs affect TG2 enzymatic activity even though they do not directly bind the enzymatic site. TG2 transamidase activity was measured in protein lysates extracted from SKOV3 cells treated with 10 μM of compounds P374, P404, and 405 for 24 hours. MT4 (a TG2/fibronectin inhibitor (21) which does not affect TG2 expression levels) and VHL ligand 1 served as negative controls. Protein lysates from stably transduced cells with TG2 (13) or control targeting shRNA were used as positive controls. Compounds P374, P404, and 405 significantly reduced TG2 enzymatic activity (Figure 3F) to similar levels as TG2 targeting shRNA, whereas treatment with MT4 and VHL ligand 1 had no effect on transamidating activity. These results demonstrates that by reducing TG2 protein levels, the selected PROTACS impair all of its functions, including its catalytic function.

New TG2 PROTACs impair cell migration, adhesion, and activation of the FAK/Src pathway:

After validating that these compounds effectively degrade TG2 in a proteasome-dependent manner and impair its transamidase activity, we next examined whether they also affect the functions of TG2 related to interaction with fibronectin and the extracellular matrix (ECM)(7, 21). To test this, we measured the effects of the new PROTACS on the migratory and adhesive properties of OC cells. Treatment of SKOV3 and OVCAR5 cells with 10 μM of compounds P374, P404, and 405 significantly reduced cell migration, as assessed by scratch assay at 18 hours (Figure 4A, B). Furthermore, treatment with the same compounds at the same concentration impaired the cells’ ability to adhere to a fibronectin-rich matrix (Figure 4C, D).

Figure 4: Compounds P374, P404, and P405 reduce cell migration, adhesion, and activation of the FAK/Src signaling pathway.

(A, B) Scratch wound migration assay in SKOV3 (A) and OVCAR5 (B) cells treated with 10 μM P374, P404, and P405. Quantification of wound closure at 18 h is shown alongside representative images. n = 3 replicates. (C, D) Fibronectin adhesion assay of SKOV3 (C) and OVCAR5 (D) cells treated with 10 μM P374, P404, and P405. n = 3 replicates. (E, F) Immunofluorescence analysis of SKOV3 cells seeded on fibronectin and treated with P374 at 1 and 10 μM. Cells were stained for TG2 and phosphorylated FAK^Tyr397 (E) or phosphorylated Src^Tyr416 (F). DAPI was used for nuclear staining. Quantification of fluorescence signal intensity is shown below each image. n = 15 cells per group. All values represent mean ± SD. ***p < 0.01; ****p < 0.0001, 1 way ANOVA with Tukey post-test.

TG2-mediated interaction between fibronectin and integrin β1 was shown to activate the FAK/Src signaling pathway (12). Activation of this pathway was assessed by immunofluorescence (IF) analysis of phosphorylated FAK (Tyr397) and Src (Tyr416). Treatment of SKOV3 cells with compound P374 at 1 μM and 10 μM significantly reduced the phosphorylation levels of both FAK and Src (Figure 4E, F), indicating that TG2 degradation induced by the compound disrupted downstream integrin signaling. These results suggest that the inhibition of cell migration and adhesion to ECM induced by the PROTACS are associated with inhibition of the FAK/Src pathway via TG2 degradation.

RPPA Analysis of PROTAC-Treated Cells Reveals Broad Proteomic Changes:

To gain deeper and unbiased insight into the cellular effects induced by PROTACs treatment, we performed Reverse Phase Protein Array (RPPA) analysis of SKOV3 cell lysates treated with control or with P374 for 24 hours. RPPA confirmed TG2 degradation at 24 hours (Figure 5A, blue arrow) and revealed significant alterations in the expression of other proteins. Notably, the expression levels of AKT, SOX2, MAPK2, and ATRX were reduced, among others (Figure 5A, red arrows).

Figure 5: RPPA analysis shows reduction of TG2-related proteins upon treatment with compound P374.

(A) Heatmap of proteins with significant changes between control and P374-treated SKOV3 cells as detected by RPPA. Blue arrow indicates TG2; red arrows highlight TG2-related proteins. n = 2. (B) Pathway analysis of downregulated proteins from RPPA data, performed using Metascape. (C) Pathway analysis of upregulated proteins identified in RPPA data, also performed using Metascape. (D) Effector analysis of downregulated proteins using Metascape. (E) Western blot analysis validating selected targets identified in the RPPA analysis, n=3.

Pathway enrichment analysis of differentially expressed proteins using Metascape (35) showed that treatment with P374 downregulated several pathways, including Leptin signaling, cancer-related pathways, positive regulation of cell adhesion, and the RAC1–PAK1–p38–MMP2 axis (Figure 5B). In contrast, pathways such as CAMKK2 signaling, negative regulation of catalytic activity, regulation of kinase activity, and regulation of catabolic processes were upregulated (Figure 5C). Effector analysis indicated that key regulators of the downregulated pathways included TWIST1, AR, HDAC1, and TP53 (Figure 5D).

To validate these findings, Western blot analysis of selected targets identified by RPPA was performed. Treatment with compound P374 reduced the protein expression levels of TG2, p-AKT, total AKT, SOX2, ATRX, and HER2/ErbB2 (Figure 5E), confirming the RPPA results. AKT and SOX2 are known TG2 targets (11, 36, 37), but other proteins identified as being downregulated after treatment with degrader P374 have not yet been described as being related to TG2. Taken together, these data suggest that PROTAC-mediated TG2 degradation induces significant changes in multiple signaling pathways, which are likely involved in the phenotype induced by the compound in OC cells and support further studies of these compounds in OC models.

Pharmacokinetic (PK) properties of new TG2 targeting PROTACS:

To enable in vivo studies, we evaluated PK properties of the new TG2 PROTACS (Table 1). Compounds P373, P404, and P405 were tested for stability in mouse liver microsomes. All three compounds were metabolized moderately quickly. Microsomal stability showed a T_1/2 ranging from 11 to 21 minutes for compounds P374, P404 and P405, and Cl_int ranging from 32 to 60 μl/min/mg. Their metabolism rate was similar to the parental inhibitor MT4 (21) and our recently-described TG2 PROTACs (compounds 7 and 11)(29), with P374 having a 1.5-2x longer half-life than P404 or P405. We also tested these three compounds for passive permeability in a PAMPA assay, which shows low but measurable permeability of all three new compounds. This compared favorably to the TG2 PROTAC compound 7 (29), which showed very low permeability, while being less permeable than the TG2 inhibitor MT4, as expected based on molecular weight. Both kinetic and thermodynamic solubility of these three compounds were modest, with value lower than the small TG2 inhibitor MT4 and the PEG-containing PROTACs compounds 7 and 11. Based on the lower in vitro clearance of compound P374 and the greater potency of P404 relative to P405, we next assessed the in vivo PK of compounds P374 and P404. Using a cassette format, each compound was administered to female C57BL/6 mice at doses of 2 mg/kg. At eight subsequent time points, plasma was extracted and drug levels were quantified by LC/MS/MS (Figure 6A). Each compound had a half-life between 4.2-5.4 hours. P404 had a higher C_max (870 nM versus 600 nM) and a greater exposure than P374.

Table 1.

illustrates in vitro DMPK values for new TG2 PROTACs P374, P404, and P405 showing mouse microsomal stability, PAMPA permeability, and solubility. Compound MT4 was reported previously and is the small molecule TG2 inhibitor control. Compounds 7 and 11 are TG2 PROTACs we recently reported (29).

	Microsomal Stability		PAMPA Permeability	Kinetic Solubility	Equilibrium Solubility
Compound ID	T1/2 (min)	Clint (μl/min/mg)	(10−6 cm*s−1)	(μM)	(μM)
MT4	15.2	46	2.60	54.96	0.76
Compound 7	12.4	56	0.00	1.19	0.24
Compound 11	18.4	38	0.90	9.95	0.17
P374	21.5	32	0.51	0.36	0.06
P404	11.6	60	0.63	0.63	0.01
P405	16.1	43	0.77	0.27	0.02

Figure 6: New compounds targeting TG2 reduce tumor colonization and progression in mouse models.

(A) Plasma PK of P374 and P404. Compounds were dosed in cassette format by intraperitoneal (IP) administration to female C57BL/6 mice at doses of 2 mg/kg each and plasma concentrations were measured at the indicated time points. (B, C) Omental colonization assay using luciferase-expressing OVCAR5 cells injected IP into nude mice and co-administered with either diluent vs. compound P374 (B) or P405 (C). Tumor colonization was assessed as luciferase signal intensity. Quantification of total tumor weights (D), tumor volumes (E), and tumor numbers (F) in IP OVCAR5 xenografts treated with diluent or P374 for 4 weeks. P374 or vehicle (control) were administered IP 55μg/Kg three times per week (n=5 mice per group). (G) TG2 expression assessed via IHC in xenografts from experiment in C-E (n= 5 tumors per group). Quantification of TG2 staining was performed using FIJI color deconvolution. (H) TG2 enzymatic activity was assessed in ascites fluid collected from the experiment in C-E (control n = 5; P374-treated n = 3). (I) TG2 enzymatic activity was assessed in malignant OC-associated ascites (n=5 specimens). All values represent means ± SD. *p < 0.05; ** < 0.01 by 2-tailed Student’s t-test.

New PROTACs Targeting TG2 Reduce Omental Colonization and Tumor Progression In-Vivo:

Based on the phenotypic changes observed in OC cells upon PROTAC-mediated TG2 degradation in-vitro and the knowledge that TG2 plays an important role in mediating OC peritoneal dissemination (15, 16, 38), we next investigated whether these compounds also exhibit anti-tumor effects in-vivo.

First, to assess omental colonization, OVCAR5-luciferase-expressing cells were injected intraperitoneally (IP) into nude mice and allowed to colonize the omentum for 3 days. Cells were co-injected with either DMSO (vehicle) or P374 or P405 at a concentration of 10 μM. Omental colonization was evaluated by excising the omentum and measuring luciferase activity 3 days post-injection following a previously described protocol (39). Both compounds P374 and P405 significantly reduced omental colonization compared to vehicle controls, as evidenced by lower luciferase signal intensity in TG2 targeting PROTAC injected animals (Figure 6B, C).

Second, to evaluate the effects of TG2 targeting PROTAC treatment in vivo, we performed an extended tumor progression study. OVCAR5-luciferase cells were injected IP into nude mice followed by treatment with either vehicle or P374 at 55 μg/kg IP 3 times per week for 4 weeks. Body weights were monitored throughout the experiment to assess toxicity and identify signs of disease progression (Supplementary Figure 25 A, B). No significant weight loss was induced by treatment with compound P374. At endpoint, mice treated with P374 exhibited significantly reduced tumor weights, tumor volumes, and tumor numbers (Figure 6D–F), indicating robust anti-tumor effects of the new PROTAC. No significant differences were observed in ascites volume between treatment and control groups (Supplementary Figure 25C).

To validate PROTAC induced TG2 degradation in vivo, TG2 expression was assessed in tumors and TG2 activity was measured in ascites. IHC analysis confirmed reduced TG2 expression in tumors from P374-treated versus control treated mice (Figure 6G, p< 0.05). Likewise, TG2 enzymatic activity in ascites fluid was significantly lower in mice treated with P374 compared to control treated mice (Figure 6H, p<0.05). Importantly, TG2 activity was measured in human OC associated ascites samples by using the same assay (Figure 6I, J), demonstrating that the range of TG2 transamidating activity in murine models is similar to the levels detected in human malignant ascites. Together, these results demonstrate that the novel TG2-targeting PROTACs described here are active in vivo, reducing tumor colonization and progression.

TG2 is an important target in OC metastasis:

We and others have shown that TG2 is upregulated in OC and mediates peritoneal metastasis (13, 14, 20, 38, 40). Quantification of TG2 transcript (TGM2) in a publicly available database shows upregulation of TGM2 in omental implants from stage III ovarian cancer compared to primary ovarian tumors (Supplementary Figure 26) further asserting the significance of this protein in cancer invasiveness and metastasis. To evaluate the spatial expression of TGM2, a TMA comprising 184 1-mm cores from 20 HGSOC patients was analyzed by Visium HD spatial profiling. While the minimal size of cores and variable content of epithelial cells in each core (Figure 7A) precluded meaningful comparison of TGM2 expression per patient and per anatomic tumor location, it was apparent that TGM2 expression was particularly enriched in small clusters of epithelial cells surrounded by ECM (Figure 7B). The invasive cells exhibited a mixture of ameboid and mesenchymal phenotypes marked by high nuclear atypia and pleomorphism (Figure 7C). Cell aggregates consisting of 1-20 invasive cells embedded in ECM were manually selected as an “invasive cluster” (Figure 7D) to identify genes differentially expressed in the “invasive cluster” compared to the rest of the FOV transcriptome.TGM2 was identified as one of the top 3 highly enriched transcripts in the “invasive cluster” (Figure 7E). These data, adding to already published reports (14, 38), support the significance of TG2 to the risk for OC dissemination and provide a robust rationale towards further development of TG2 targeting PROTACS as potential new therapeutic agents.

Figure 7: TG2 transcript is enriched in invasive HGSOC cells in surgical specimens.

(A) Representative Visium HD TMA FOV highlighting epithelial cells. Scale bar, 500 μm. (B) Top panel: TGM2 transcript expression overlay (blue, log₂ scale 0-3) in three representative cores. Middle panel: Corresponding H&E-stained cores. Bottom panels: Higher-magnification views of the boxed regions from the top panels, showing TGM2 transcript overlay on H&E images. Scale bars: top and middle panels, 500 μm; bottom panels, 50 μm. (C) Higher magnification H&E image showing invasive microclusters characterized by ameboid and mesenchymal morphologies, nuclear pleomorphism, and cytologic atypia. (D) Manual annotation of bins containing invasive cells (highlighted in pink) on the H&E image using the 10X Genomics Loupe Browser paint brush tool. Scale bar, 500 μm. (E) Differentially expressed genes enriched in the invasive cell cluster (pink bins in D) compared to all other bins within the FOV.

Conclusions and Discussion

Our findings describe a new family of TG2 PROTACS built on the small molecule MT4 scaffold which have prolonged degrader activity and promising in vitro and in vivo biological effects compared to the first generation of TG2 degraders we reported (29). For the next generation of degraders we preserved MT4, a small molecule that binds to the fibronectin binding domain of TG2 (21), as the recruiter for the protein of interest. We attached the TG2 and VHL ligands using a thiol ether along with PEG linkers leading to the design and synthesis of the family of compounds described here, P374, P404, and 405. The thiol ether connection in PROTACs is relatively uncommon and the work here demonstrates that this could be a viable strategy for designing PROTACs targeting other proteins. The new compounds had more prolonged degrading effects (24 hours) against TG2 and potent inhibitory effects against all the protein’s functions, including inhibitior of cell adhesion, migration, and transamidating activity.

TG2 upregulated in OC, as well as in other malignancies, has been linked to chemoresistance, cancer stemness, metastasis and cancer cell survival under stress (11, 15, 37, 41). A recent proteogenomic analysis of over 240 high grade serous ovarian tumors identified in an unbiased manner TG2 as the protein most upregulated in tumors resistant to chemotherapy (42), highlighting the high clinical relevance of this enzyme in OC. TG2 is inducible by TGF-β (11) and other pro-inflammatory cytokines and activates NF-KB and AKT survival pathways (37, 43), as well as β-catenin signaling (12, 44). Its involvement in metastasis in OC has been attributed to promotion of epithelial to mesenchymal transition (15) and facilitation of cell adhesion to the fibronectin matrix and activation of outside in signaling (16, 44). Here we re-affirm TG2’s association to metastasis by reporting increased TG2 expression in omental metastases compared to primary tumors and by showing that TG2 is among the highest upregulated transcripts in cancer cells at the invasive edge compared to the rest of the cells in the tumors. This unbiased discovery of TG2 in the most invasive cancer cells that drive metastasis provides further rationale towards finding effective blocking strategies.

However, the protein has remained an elusive target, due to its multiple functions linked to specific structural domains. For example, small molecules that inhibit the enzymatic function have no effects on the interaction between TG2 and fibronectin and integrin (45), and therefore do not block outside in signaling. However, some of the enzymatic inhibitors were shown to induce conformational changes in the protein structure resulting in concomitant inhibition of the GTPase function (24). In contrast, inhibitors of the TG2-fibronectin-integrin complex have no effect on the enzymatic and GTPase functions (21, 46). The neutralizing antibody zampilimab blocks TG2’s crosslinking activity in the stroma, but is not cell permeable and lacks intra-cellular activity (47). These challenges justify continued efforts towards developing cell permeable inhibitors capable of blocking all of TG2’s functions. Towards this end, we developed the first TG2 degraders which we showed can effectively block cellular adhesion and migration (29). However, these firstly characterized compounds had short activity in vitro, leading to continued efforts to improve biological potency and chemical properties.

Evaluation of our new and most promising degraders, compounds P374 and P404 in in vitro DMPK experiments showed moderate stability towards mouse liver microsomes, permeability, and solubility. The moderate permeability may be a factor in reducing their degradation potency and efficacy. Future studies to improve permeability may lead to more effective degraders. In vivo PK studies in mice showed reasonable half-lives and exposure, which led to their suitability for in vivo pharmacodynamic and efficacy studies. These in vivo parameters were well-correlated to their in vitro results, suggesting that for this series, in vitro DMPK testing could be an effective strategy for helping to generate new degraders with improved in vivo efficacy.

Importantly, we showed that P374 and P404 inhibited omental colonization and that compound P374 had potent in vivo activity blocking tumor growth and dissemination at relatively low doses. TG2 degrading activity in vivo was demonstrated by immunohistochemistry staining and measurement of transamidase activity in ascites specimens. These results provide confidence in the potential to effectively target TG2 in cancer models in vivo and to advance TG2 degraders towards clinical applications.

Experimental methods

Cell lines and reagents :

SKOV3 cells were purchased from ATCC and OVCAR5 cells were obtained from Dr. Marcus Peter at Northwestern University. Cells were used for experiments at low passage number and cultured at 37°C with 5% CO2 and 100% humidity. Cells were tested to be pathogen and mycoplasma free with regular in-house PCR-based testing (ATCC Universal Mycoplasma Detection Kit) and by Charles River Animal Diagnostic Services. Cell lines were authenticated by IDEXX BioAnalytics with short tandem repeat (STR) profiling. The anti-TG2 antibody (Cat#: ab2386) was purchased from Abcam. The GAPDH antibody (Cat#: sc-32233) was purchased from Santa Cruz Biotechnology). VHL-ligand 1 was procured from Cayman Chemicals (Cat#: 21591).

Western Blotting:

SKOV3 and OVCAR5 cells were seeded on 6 cm dishes and treated with different concentrations of compounds at different time points. Protein lysates were extracted using the RIPA buffer and quantified using the Bradford assay. After gel electrophoresis and transfer, PDVF membranes were blocked with TBS-Tween 5% bovine serum albumin (BSA) for 1 h and incubated with primary antibodies (1:1000 dilution, overnight at 4 °C). After incubation with the secondary antibody (anti-rabbit/mouse-horseradish peroxidase 1:2000 dilution) for 1 h at room temperature, the signal was developed using the SuperSignal West Pico PLUS Chemiluminescent Substrate (Cat#: 34580, Thermo Fisher Scientific) and detected using an ImageQuant LAS 4000 machine for chemiluminescent signal acquisition. To detect additional proteins, membranes were treated with Restore Western Blot Stripping Buffer (Cat#: 21059, Thermo Fisher Scientific), blocked, and then incubated with the primary antibody. After imaging, western blots were quantified using IMAGEJ, and statistical analysis was performed using GraphPad prism 8.

Scratch Migration assay:

Confluent 6-well plates were scratched using a sterile pipette tip, rinsed with 1x PBS to remove debris, and treated with experimental compounds at varying concentrations. Images of the wound area were captured at 0, 18, and 24 hours, and wound closure was quantified using ImageJ software. To assess cell migration, transwell assays were performed using 8 μm pore-size inserts (Cat#: PI8P01250, Merck-Millipore) SKOV3 cells were seeded in the upper chamber in serum-free media containing either the test compounds or vehicle control, while media supplemented with 10% serum was added to the lower chamber to act as a chemoattractant. After 18 hours, cells that had migrated through the membrane were fixed in pre-warmed 4% paraformaldehyde (PFA) for 20 minutes at room temperature and stained with crystal violet for 3 minutes. Membranes were then excised, mounted on cover slides, and imaged at 40x magnification for quantification.

Solid Phase Adhesion Assay:

Briefly, 6 well-dishes were coated with 10 μg/mL fibronectin diluted in phosphate-buffered saline (PBS) for 1 h. Cells (30,000 per well) were pretreated with compounds for 6 h were seeded and allowed to attach for 30 min. Unattached cells were removed by washing with PBS, and cells adhering to coated wells were quantified by using the CCK-8 assay (Cat# K1018, ApexBio), following the manufacturer’s protocol. Plates were imaged at 1 h and 2 h. Relative absorbance was measured at 496 nm and was used to estimate the relative number of cells attached to fibronectin-coated wells.

Cell viability Assay:

2,000 cells per well were seeded in 96-well plates and were treated with PROTAC compounds at different concentrations of 0.1, 0.3, 1, 3, 10, 30 μM for 24 h. Medium was refreshed after the 24 h treatment and cells were allowed to grow for an additional 48 h. Viability of the cells was determined using a CCK-8 kit (Cat# K1018, ApexBio) in accordance with the manufacturer’s protocol. Absorbance at 450 nm was measured with a microplate-reader (BioTek ELX800) to estimate cell viability.

Reverse Phase Proteomic Analysis (RPPA):

SKOV3 cells were cultured in 15 cm dishes to confluency and treated for 24 h with either P374 or vehicle control. Cells were lifted from plates and centrifuged for 3 minutes at 1000 rpm to obtain cell pellets. Cell pellets were flash frozen and processed for protein extraction and RPPA Analysis at MD Anderson Cancer Center Proteomics Core Facility. Triplicate protein extracts per condition were diluted 2-fold and spotted on nitrocellulose coated slides. Proteins were detected by probing with specific antibodies and signals are obtained by tyramide dye deposition and DAB reaction. Digital images of slides were captured using Huron TissueScope scanner. Mean net intensities of spots were used for curve-fitting for each slide. Relative protein levels were determined by using RPPASPACE.

Immunofluorescence (IF):

SKOV3 cells were plated as monolayers on FN-coated 8 chamber cell culture slide (Cat# 229168, CELLTREAT). After treatment, cells were fixed in 4% paraformaldehyde for 10 min at room temperature, washed 3 X with PBS, permeabilized using 0.2% Triton X-100 in PBS (5 min) and blocked for 1 h with 3% goat serum in 1X PBS-0.1% Triton X-100. Subsequently, cells were incubated with primary antibodies (diluted 1:100) in blocking buffer overnight at 4 °C. After 3 X washing in 1X PBS-0.1% Triton X-100 cells were incubated for 1 h at room temperature with Alexa fluor-488-conjugated anti-rabbit or Alexa fluor-647-conjugated anti-mouse secondary antibodies (1:1000; Thermo Fisher Scientific Inc). Isotype-specific IgG was a negative control. Nuclei were visualized by 4’,6-diamidino-2-phenylindole (DAPI) staining (Vectashield, Vector Laboratories). The analysis was performed by using the multiphoton microscope STELLARIS 8 DIVE (Leica Microsystems Inc., Deerfield, IL, USA) at the Indiana Center for Biological Microscopy (ICBM) facility (Indianapolis, IN, USA). TG2, integrin β1, p-FAK^Tyr397, and p-Src^Tyr416 fluorescence was quantified in ImageJ as area of integrated intensity for each selected cell. Next, the corrected total cell fluorescence (CTCF) was calculated by using the following formula: CTCF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings). For the negative control experiments, only the anti-rabbit or -mouse IgG isotypes were incubated with the Alexa fluor-647-conjugated or Alexa fluor-488-conjugated anti-rabbit or anti-mouse secondary antibodies. Each image is representative of at least three replicates.

In-vivo omental colonization assay:

Experiments were conducted with 6-8 weeks old athymic female nude mice (Envigo) following a previously described protocol (39). In brief, 10 million luciferase expressing OVCAR5 cells pretreated for 24 h with 10 μM P374 or P404 or with diluent (control) were injected IP. Mice were euthanized after 3 days, and the omentum was collected and weighed. OC cells that had colonized the omentum were isolated from the harvested tissue by first mincing the tissue into small pieces (≤ 0.1 mm³) and then incubating it on a rotor for 1 h at 37 °C in 2 mL of digestion buffer containing 1 mg/mL collagenase type I (Cat#: 17100017, Thermo Fisher Scientific) dissolved in basal DMEM/F12 medium. Samples were centrifuged at room temperature for 5 minutes at 1000 RPM to isolate cancer cells. The supernatant was removed and pellets were used to measure luciferase levels/activity which was was quantified with a Luciferase Glow Assay Kit from Thermo Scientific (Cat#: 16176) along with signal enhancer to enhance luciferase reactions (Cat#: 16180, Thermo Fisher Scientific). Luminescence was measured at all wavelengths with a SpectraMax Gemini XS plate reader. Luciferase signal was normalized to the omentum weight for each animal.

Xenograft experiments:

Experiments were performed with 6-8 weeks old female athymic nude mice, purchased from Envigo. After 5 day acclimatization, mice were injected intraperitoneally (IP) with 2 million OVCAR5 cells pre- treated with 10 μM P374 or control for 24 h prior to injection. Mice were randomly assigned to IP treatment with P374 (55 μg/kg diluted in 100 μl three times per week) or control (100 μL PBS) three times a week for 4 weeks (n = 5 mice per group). Mice were monitored twice weekly and euthanized 28 days after injection of OVCAR5 cells. Tumor implants > 2 mm³ on mesentery, abdominal wall, and bowel surface were counted. Weights and dimensions of tumor implants larger than 3 mm³ were measured. Tumor volumes were calculated using the following formula: volume (mm³) = (length × width × height)/2. Ascites volumes were recorded. Experiments were approved by the Northwestern University Institutional Animal Care and Use Committee (IACUC, Protocol #IS00020868).

Transglutaminase 2 enzymatic activity:

Transamidase activity was quantified with a colorimetric assay from Novus Biologicals (Cat# NBP1-37008), following the manufacturer’s instructions. Briefly, samples (cell lysates or 50 μL ascites) were incubated for 30 minutes at 37 °C with 50 μL of assay mixture. Negative control was incubated with 50 mM EDTA under the same conditions. Wells were washed once with 1x Wash Buffer, once with 0.1M NAOH, twice with 1x Wash Buffer, and once with 0.5% BSA followed by a 15 min incubation at 37 °C of SAv-HRP (1:2000) solution. Wells were washed three times with 1x Wash Buffer. 100 μL of HRP substrate was incubated for 5 min and then the reaction was stopped using a blocking reagent. Absorbance at 450 nm was measured with a microplate-reader (BioTek ELX800).

Human specimens:

De-identified human OC associated malignant ascites was collected from consenting donors with newly diagnosed ovarian cancer under Northwestern University IRB protocol STU00202468. Tissue microarrays (TMAs) of high grade serous ovarian carcinoma (HGSOC)(48) were used for 10X Genomics Visium HD analysis under the University of California Los Angeles (UCLA) IRB#24-6294.

Immunohistochemistry assay (IHC):

Slides were deparaffinized using a graded alcohol series, beginning with two 30-minute washes in xylene, followed by sequential 5-minute washes in 100% ethanol (twice), 90%, 70%, and 50% ethanol, and finally in distilled water (ddH₂O). Antigen retrieval was performed using 0.01 M citrate buffer (pH 6.0) at 95 °C for 30 minutes. Endogenous peroxidase activity was quenched by incubating the samples in 3% hydrogen peroxide (H₂O₂) for 15 minutes. Slides were then blocked with normal goat serum (NGS) for 30 minutes at room temperature to reduce nonspecific binding and incubated overnight at 4 °C with the primary antibody. Detection was performed by using the Dako LSAB2 System (Cat. #K0675, Agilent) according to the manufacturer’s instructions. Slides were counterstained with hematoxylin and mounted using Faramount Aqueous Mounting Medium (Cat. #S302580-2, Agilent). Quantification IHC images was performed by using FIJI/Image J for color deconvolution.

Visium HD spatial transcriptome analysis:

10X Genomics Visium HD analysis was performed by the UCLA Technology Center for Genomics & Bioinformatics (TCGB). Twelve individual TMA formalin-fixed paraffin-embedded (FFPE) tissue sections were processed using the 10X Genomics Visium HD Spatial Gene Expression platform according to the manufacturer’s protocol. Following tissue fixation and deparaffinization, H&E staining and brightfield imaging were performed. After imaging, the slides were destained and de-crosslinked, then twelve 6.5x6.6 fields of view (FOVs) were subjected to hybridization with a whole transcriptome probe panel. This was followed by probe ligation, CytAssist-enabled probe release, extension, and elution. Final libraries were amplified by quantitative PCR (qPCR). Library quality and concentration were evaluated using an Agilent TapeStation and Qubit fluorometer prior to sequencing. The libraries were sequenced on the Illumina NovaSeq X+ platform (PE 2x50). Data de-multiplexing was performed using BCLConvert software v4.3.13 (Illumina). Manual alignment of each high-resolution (40x) H&E microscope image to the corresponding CytAssist slide image was performed in the Loupe Browser (10X Genomics). Sequenced reads in FASTQ format were aligned and analyzed via the SpaceRanger count pipeline v3.1.2 (10X Genomics) using the 10X refdata-gex-GRCh38-2020-A reference database and 10X Visium_Human_Transcriptome_Probe_Set_v2.0_GRCh38-2020-A probe set under default settings. Clustering and differential expression analyses were performed using Loupe Browser (10X Genomics, version 8.0.0) with 8 μM bins. Bins containing invasive cells (defined as regions with 1-20 epithelial cells embedded within the ECM) were annotated using the paint brush tool in the 10X Genomics Loupe Browser and grouped into a single category labeled “invasive cluster.” Differential gene expression analysis was then performed between the “invasive cluster” and the remaining FOV using the Loupe Browser with default parameters.

Metabolism and Pharmacokinetics.

Hepatic microsomal stability:

Microsome stability was evaluated by incubating 1 μM test compound with 1 mg/mL mouse liver microsomes in 100 mM potassium phosphate buffer, pH 7.4. The reaction was initiated by adding NADPH (1 mM final concentration). Aliquots were removed at 0, 5, 10, 20, 40, and 60 minutes and added to acetonitrile (5x v:v) to stop the reaction and precipitate the protein. At the end of the assay, the samples were centrifuged through a Millipore Multiscreen Solvinter 0.45 micron low binding PTFE hydrophilic filter plate and analyzed by LC-MS/MS. Data was log transformed and represented as half-life and intrinsic clearance.

Parallel Artificial Membrane Permeability Assay (PAMPA):

Passive permeability was measured using the Corning Gentest Pre-coated PAMPA Plate System (Cat. #353015). 0.02% BSA in PBS was used in the donor side and 0.4% BSA in PBS was used in the top receiver plate. Compounds were added as 5 μM DMSO solutions to the bottom donor side. After incubation at 37 °C for 5 h, samples from the receiver side were removed and quantified by LC/MS/MS.

Solubility:

To measure kinetic solubility, 1 μL of a 10 mM DMSO stock solution was added into 99 μL PBS pH 7.4. This was maintained at room temp for 24 h, filtered, and analyzed by LC/MS/MS. To measure equilibrium solubility, approximately 1-2 mg of dry compound was added to a 8 mL test tube containing 1 mL PBS pH 7.4. The mixture was inverted for 24 h, filtered, and the filtrate measured by LC/MS/MS.

Statistical Analysis:

Data are reported as means ± standard deviation (SD). Statistical significance was determined using a two-tailed Student’s t-test (Prism 8, GraphPad Software) with significance levels denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. A p-value less than 0.005 was considered statistically significant. The number of biological replicates (n) for each experiment is provided in the corresponding figure captions.

Highlights:

• We describe the synthesis of a novel family of VHL based PROTACs targeting TG2.

• These PROTACS block both the fibronectin binding and transamidase functions of TG2.

• The TG2 targeting PROTACs have potent in vivo activity, reducing tumor progression.

• Spatial transcriptomics shows TG2 enriched in cancer cells at the invasive edge.

Data Availability:

All data will be made available upon request.