Sulforaphane covalently interacts with the transglutaminase 2 cancer maintenance protein to alter its structure and suppress its activity

Abstract

Type 2 transglutaminase (TG2) functions as an important cancer cell survival protein in a range of cancers including epidermal squamous cell carcinoma. TG2 exists in open and closed conformations each of which has a distinct and mutually exclusive activity. The closed conformation has GTP-binding/GTPase activity while the open conformation functions as a transamidase to catalyze protein-protein crosslinking. GTP-binding/GTPase activity is required for TG2 maintenance of the aggressive cancer phenotype. Thus, identifying agents that convert TG2 from the closed to the open GTP-binding/GTPase inactive conformation is an important cancer prevention/treatment strategy. Sulforaphane (SFN) is an important diet-derived cancer prevention agent that is known to possess a reactive isothiocyanate group and has potent anti-cancer activity. Using a biotin tagged SFN analog (Biotin-ITC) and kinetic analysis we show that SFN covalently and irreversibly binds to recombinant TG2 to inhibit transamidase activity and shift TG2 to an open/extended conformation, leading to a partial inhibition of GTP binding. We also show that incubation of cancer cells or cancer cell extract with Biotin-ITC results in formation of a TG2/Biotin-ITC complex and that SFN treatment of cancer cells inhibits TG2 transamidase activity and shifts TG2 to an open/extended conformation. These findings identify TG2 as a direct SFN anticancer target in epidermal squamous cell carcinoma.

Introduction

Tumor cells survive by avoiding normal cell death processes¹. Cancer stem-like cells comprise an important subpopulation of tumor cells that display enhanced survival and tumor formation^2,3. Epidermal cancer stem-like cells (ECS cells) display enhanced invasion, migration and ability to form highly vascularized and rapidly growing tumors as compared to non-stem cancer cells^4–6. This finding has led to a search for targets to reduce ECS cell survival. An important goal is identification of proteins that are elevated in level or activity in cancer stem cells that may serve as therapy targets. Previous studies indicate that transglutaminase type 2 (TG2) is markedly elevated in cancer stem-like cells and it has a role in driving cancer progression^3,5–7. TG2 displays distinct enzymatic activities that are associated with TG2 conformation^8–10. Closed/folded conformation TG2 binds and hydrolyzes GTP as a G-protein signal transduction protein^11,12, while open/extended conformation TG2 catalyzes calcium-dependent formation of covalent crosslinks (transamidation) between the γ-carboxamide group of a peptide bound glutamine and primary amine substrates^8,9,13–15. Closed TG2 predominates in the intracellular environment where were GTP levels are high and calcium levels are low^8,11; however, increased intracellular calcium, which can occur, for example, during cell death/differentiation, shifts TG2 to an open conformation and activates transamidase activity^8,9,14–16. These enzymatic activities are mutually exclusive as crosslinking activity is allosterically activated by Ca²⁺ and inhibited by GTP, GDP, and GMP^11,17.

An important pro-cancer signaling role of closed conformation TG2 has been shown in a number of cancer models^3,5,18,19 and we have shown that the GTP-binding/G-protein signaling activity of closed conformation TG2 drives the ECS cell cancer phenotype^3,5. Thus, we have worked to identify agents that suppress TG2 function^5,20,21. Our recent study indicates that NC9, a specific TG2 inhibitor, inactivates TG2 by binding to the transamidase site to produce a conformation change that distorts the TG2 GTP binding site to reduce pro-cancer activity²¹. The net impact is to inhibit both TG2 GTP binding and transamidase activities.

In the present study, we examine the role of sulforaphane (1-isothiocyanato-4-(methylsulfinyl) butane, SFN) as an inhibitor of TG2 function. SFN, a natural isothiocyanate derived from cruciferous vegetables, is an important cancer prevention and treatment agent that reduces tumor formation but produces no observable side effects^22–25. Although a host of studies show that SFN suppresses the skin cancer phenotype^26–29, only limited information is available regarding cellular targets that are modified by SFN^30–32. We show that SFN covalently and irreversibly binds to TG2 to alter its conformation and inhibit its transamidase and GTP binding activities and that this is associated with a reduction in cancer phenotype.

Materials and Methods

Cells and Reagents

SCC-13³³ and HaCaT³⁴ are epidermis derived malignant and immortalized cell lines, respectively. Dulbecco’s modified Eagle’s medium (11960–077), sodium pyruvate (11360–070), L-glutamine (25030–164), and 0.25% trypsin-EDTA (25200–056) were from Gibco (Gaithersburg, MD). Spheroid growth medium is DMEM/F12 (1:1) (DMT-10–090-CV, Mediatech Inc, Manassas, VA) containing 2% B27 serum-free supplement (17504–044, Invitrogen, Frederick, MD) and 20 ng/ml EGF (E4269), 0.4% bovine serum albumin (B4287) and 4 μg/ml insulin (#19278) which were obtained from Sigma (St. Louis, MO)⁴. Heat-inactivated FCS (F4135, C7522), A23187 ionophore (C7522), allyl isothiocyanate (AITC, 377430), benzyl isothiocyanate (BITC, 252492) and butyl isothiocyanate (BUITC, 253790) were also obtained Sigma. Cell lysis buffer (9803) was from Cell Signaling Technology (Danvers, MA). Anti-TG2 (MAB3839) and streptavidin-agarose beads (16–126) were obtained from EMD Millipore (Billerica, MA). BD Biocoat cell inserts (353097) and Matrigel (354234) were from BD Biosciences (Franklin Lakes, NJ). Fluorescein cadaverine (FC) was from Life Technologies (Frederick, MD). Synthesis of NC9, a TG2 specific inhibitor, is described elsewhere³⁵. SFN (8044) was purchased from LKT (St. Paul, MN). Biotin-ITC was synthesized as described in Supplemental Materials. Ultra-low attachment Costar cluster dishes (#4371) were obtained from Corning (Tewksbury, MA). Measurement of significant difference was performed using the student’s t-test with a minimum of three independent repeated experiments.

Biological assays

Conditions for the spheroid formation, Matrigel invasion and migration assays were as previously described^4,36,37. For tumor xenograft studies, 100,000 spheroid-derived cells were resuspended in 200 μl of phosphate buffered saline containing 30% Matrigel. This amount was injected subcutaneously in each of two front flanks of five eight-week-old NOD-scid-IL2 receptor gamma chain knockout mice (NSG mice) per treatment group using a 26.5-gauge needle. NC9 was delivered by intraperitoneal injection of 20 mg/kg body weight three times/week (M/W/F)³⁶ and sulforaphane (SFN) was dissolved in water and 200 μl was delivered by oral gavage (10 μmoles per treatment) three times per week (M/W/F). These treatments began shortly after tumor cell injection. Tumor volume was measured at 4 weeks post-cancer cell injection as previously described⁴. All animal experiments were reviewed and approved by the University of Maryland-Baltimore Institutional Animal Care and Use Committee. Fluorescence lifetime imaging, in situ TG2 transamidase activity assay and GTP binding/GTP-agarose pull-down of TG2 from cell extracts were performed as previously described²¹. Native gel electrophoresis of recombinant hTG2 was performed as previously describe and the bands were visualized using Coomassie blue stain²¹.

Preparation of recombinant TG2

DNA encoding Type 2 transglutaminase (TG2) was codon optimized and engineered into pET28 expression vector having an His-tag at its N-terminus followed by an enterokinase cleavage site for purification purposes (TOPGene Technologies, Pointe-Claire, QC). Upon transformation of the TG2 expression plasmid into the E. coli BL21(DE3) strain, large scale protein production was performed using 5 L of LB media. Overexpression of His-tagged TG2 was induced with 0.1 mM IPTG when the spectrophotometer A600 reading was between 0.7 and 0.8 O.D. and the cells were allowed to grow for 18 h at 20 °C. Bacterial cells were centrifuged for 15 minutes, at 4,000 r.p.m., and the cell pellets were resuspended in lysis buffer (50 mM Na₂HPO₄ pH 7.5, 350 mM NaCl, 5 mM 2-mercaptoethanol, 0.5% v/v Triton-X100) such that 3 ml of lysis buffer was added per gram of wet pellet. Prior to sonication, the cell lysate was incubated at 4 C for 1 h with DNase, 10 mM CaCl₂ and 10 mM MgCl₂ to improve the DNase reaction efficiency under these conditions. Next, 3 cycles of sonication at 50% power amplitude were performed and followed by centrifugation at 15,000 r.p.m. for 45 min at 4 °C to remove cell debris. A clear supernatant was filtered and loaded onto a previously equilibrated HiPrep^™ 16/10 IMAC Sepharose 6 Fast Flow column (Cytiva, GE28-9365-52, Sigma-Aldrich, St. Louis, MO) with buffer A (50 mM Na₂HPO₄ pH 7.5, 350 mM NaCl, 5 mM BME) and B (buffer A plus 500 mM imidazole). His₆-(EK)-TG2 was eluted at 150 ± 20 mM imidazole with 5 ml fractions collected and analyzed by gel electrophoresis. Eluted fractions (molecular weight 77 kDa) were pooled and dialyzed for 5 h against buffer C (20 mM Tris pH 7.2, 50 mM NaCl, 1 mM DTT, 1 mM EDTA) and next loaded on a HiScreen^™ Capto Q ImpRes column (Cytiva, GE17-5470-15, Sigma-Aldrich, St. Louis, MO). Fractions corresponding to His₆-(EK)-TG2 were eluted with buffer D (buffer C plus 1 M NaCl), identified by gel electrophoresis, concentrated and injected onto a HiLoad^™ Superdex S200 PG size exclusion column (Cytiva, GE28-9893-35, Sigma-Aldrich, St. Louis, MO) previously equilibrated with 50 mM HEPES pH 7.0, 100 mM NaCl, 10% Glycerol, 1 mM EDTA, 5 mM DTT. Fractions containing the His₆-(EK)-TG2 were eluted from the S200 column and shown to be pure via SDS-PAGE (>99%). The identity and purity of final purified construct was confirmed by immunoblot and denaturing and native gel electrophoresis. The protein yield was typically >10 mg of purified protein per liter of bacterial cell culture, and its concentration was adjusted to 10 μM, aliquoted, and stored at −80 °C.

Biotin-ITC incubation with cell-free extract

Ten 100 mm dishes of 60 – 70% confluent cells were washed twice with cold PBS and dissolved in 150 – 200 μl of lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na₃VO₄, 1 mg/ml leupeptin) supplemented 1 mM PMSF and 5 μl of protease cocktail III (Sigma, 539134) per 1 ml of buffer. The cells were scraped and pooled in chilled tubes, sonicated 5x on a Fisher Scientific F60 Sonifier at setting 3 with bursts separated by 30 second intervals, and centrifuged at 10,000 RPM for 15 min at 4 C. An aliquot of total extract (supernatant = 1.77 μg protein/μl) was combined with 1 volume of 2 x Laemmli sample buffer containing reducing agent and boiled. For Biotin-ITC interaction assay, extract containing 500 μg of protein was incubated with 100 μM Biotin-ITC in a 400 μl volume for 2 h at 4 C. In some experiments SFN was added as a competitor. The sample is then passed through a 500 μl Zeba desalting column (7 kDa molecular weight cutoff, Pierce Inc., Rockford, IL, 89882) to remove unbound Biotin-ITC and the eluate was incubated with 400 μl of lysis buffer-equilibrated streptavidin beads for 60 to 90 min at 4 C. The beads (bound protein) were washed 5 times with lysis buffer. Equivalent amounts to unbound flow through and streptavidin-bound fractions were boiled in 2 x Laemmli sample buffer containing reducing agent for gel electrophoresis and immunoblot with anti-TG2.

Biotin-ITC labeling of TG2 in intact cells

Cancer cells (60% confluent cultures) were harvested with trypsin, resuspended in DMEM (no additives), passed through a 35 micron pore size cell strainer (BD Falcon, Bedford, MA, 352235) to remove cell clumps, and 1.5 million single cells were transferred to a sterile microfuge tube. The cells were washed with PBS and resuspended in 100 μl keratinocyte nucleofection solution (Lonza, VVPD-1002) containing 0 or 100 μM Biotin-ITC before transfer to a cuvette for electroporation using the T-018 setting. Pre-warmed cell culture medium (0.5 ml) was then added and the cells were transferred to a 100 mm tissue culture dish containing 10 ml of medium.

At 24 h post-electroporation, the cells were washed three times with PBS, resuspended in 150 – 200 μl per dish of lysis buffer, sonicated 5 times on setting 3 and centrifuged at 10,000 RPM at 4 C. Aliquots of total lysate were diluted with 2 x Laemmli sample buffer, boiled and stored at −20 C. To detect TG2/Biotin-ITC complex formation, the remaining sample was desalted on a Zeba column and equal cell equivalents of extract from each treatment were incubated with lysis buffer-equilibrated streptavidin beads for 1.5 h at 4 C with rotation. The beads (bound protein) were washed 5 times with lysis buffer. Cell equivalent amounts to unbound flow through and streptavidin-bound fractions were boiled in 2 x Laemmli sample buffer containing reducing agent for gel electrophoresis and immunoblot with anti-TG2.

Recombinant TG2 GTP-agarose binding assay

Recombinant TG2 protein was prepared as described³⁸ with the exception that neither GTP nor ATP were included in the preparation buffer. TG2 protein (1 μg) was incubated with SFN or Biotin-ITC at 4 C for 1 h in 50 μl of GTP binding buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 0.1 % Triton X-100, 2 mM PMSF and 50 μl of proteinase inhibitor cocktail (Sigma, 593134) per 10 ml buffer. A 5 μl aliquot of total sample was set aside for electrophoresis and the remainder was incubated with 100 μl of GTP-agarose beads, pre-equilibrated in GTP binding buffer, in a final volume of 500 μl for 30 min at 4 C. The beads were washed three times with GTP binding buffer and the initial flow through was again incubated with the beads for 30 min at 4 C and rewashed. Finally, the flow through was incubated overnight with the beads at 4 C and then extensively washed^21,39. The GTP-agarose bound protein was eluted by boiling in a final volume of 210 μl of Laemmli buffer and 25 μl (107 ng equivalents) of recombinant human TG2 was electrophoresed. The 5 μl of total sample was diluted to a final volume of 100 μl with Laemmli buffer and 10 μl (10 ng equivalents of recombinant TG2). The sample was then electrophoresed and the membrane blotted with anti-TG2.

SFN and Biotin-ITC inhibition kinetics

The transamidase activity of recombinant TG2, expressed⁴⁰ and purified from E. coli, was measured according to a previously published colorimetric activity assay⁴¹ using Cbz-Glu(γ-p-nitrophenyl ester)Gly (AL5). In brief, enzymatic inhibition assays were run under conditions established by Kitz and Wilson⁴² in the presence of 100 μM AL5. The reaction mixture included 50 mM MOPS buffer (pH 6.9), with 7.5 mM CaCl₂, 100 μM AL5, and 0 – 3000 μM concentrations of inhibitor in a final volume of 200 μl in a 96-well polystyrene microplate at 25 C. Stock concentrations of AL5, SFN and Biotin-ITC were prepared in DMSO, such that the final concentration of DMSO concentration did not exceed 5 % v/v in the final assay mixture. The enzymatic reaction was initiated by the addition of 5 mU/ml TG2, or water for the blank. Formation of the TG2-dependent product, p-nitrophenolate, was monitored at 405 nm via a BioTek Synergy 4 plate reader. Observed first-order rate constants of inactivation (k_obs) were determined via non-linear regression fitting to a mono-exponential model of absorption versus time, using GraphPad Prism 7 software. The acquired rate constants were then fitted by non-linear regression to a saturation kinetics model⁴³, correcting for competition with the assay substrate AL5 by dividing by α (1 + [S]/K_M),⁴⁴ to provide the k_inact and K_I inhibition parameters. Each experiment was completed in triplicate.

Results

TG2 inhibitor and SFN treatment suppress the epidermal cancer stem-like cell (ECS cell) phenotype

TG2 is an important epidermal cancer survival factor and cancer therapy target³ that drives an aggressive cancer phenotype including enhancement of spheroid formation, invasion, migration and tumor formation³. We proposed that SFN may inhibit TG2 as a mechanism to suppress the cancer cell phenotype and reduce tumor formation. As an initial step we compared the impact of treatment with NC9, a TG2 inhibitor^21,45,46, and SFN on the ECS cell phenotype. Fig. 1A/B/C/D shows that NC9 inhibits ECS cell spheroid formation, invasion, migration and tumor formation. Fig. 1E/F/G/H shows that SFN treatment produces a similar suppression of these endpoints. Thus, both agents efficiently suppress important cancer endpoints. The similarity of the response to NC9 and SFN treatment, may suggest that TG2 functions as a SFN target. We therefore performed studies to assess the impact of SFN treatment on TG2 structure and activity.

Fig. 1. TG2 inhibitor and SFN treatment suppress the ECS cell cancer phenotype.

A SCC-13 spheroid-derived ECS cells were seeded at 40,000 cells per well in ultra-low attachment plates in spheroid medium and treated with NC9 as indicated. After three days, spheroid number and size distribution, were analyzed and expressed as mean ± SEM, n = 4. The asterisks indicate a significant reduction, p < 0.001. B ECS cells were seeded into Biocoat inserts atop a Matrigel layer and treated with 0 or 50 μM NC9 and cell invasion was monitored at 18 h. The asterisks indicate a significant reduction, n = 3, p < 0.001. C ECS cells were plated at confluent density on conventional culture plates and then scratched to create uniform wounds. NC9 was added and wound width was monitored at the indicated times. D SCC-13 cells derived ECS cells (0.1 million) were injected into NSG mice in each front flank and then treated with 0 or 20 mg/kg NC9 (M/W/F) for 0 – 5 wks. Tumor formation was monitored using calipers and tumors were photographed on week five at the time of harvest. The values are mean ± SEM, n = 8, and the asterisks indicates a significant difference, p < 0.001. E/F/G/H Spheroid-derived ECS cells were assayed for spheroid formation, invasion and migration in the presence of 0 or 20 μM SFN. The values are mean ± SEM. The asterisks indicate a significant reduction, p < 0.002, n = 4 (spheroid formation/invasion) and p < 0.001, n = 8 (tumor growth).

SFN regulates TG2 conformation

We first assessed the impact of SFN treatment on TG2 intracellular conformation using FLIM (Fluorescence Lifetime Imaging Microscopy). Cells were electroporated with mCER-TG2-YFP where mCER and YFP are fused at the N- and C-terminus, respectively, of wild-type TG2 (Fig. 2A). FLIM provides a quantitative assessment of fluorescence lifetime by measuring energy transfer between the donor (mCER) and acceptor (YFP) probes^47–51. The donor probe is excited by laser illumination and energy transfer to the acceptor is monitored by measuring donor lifetime. The lifetime is shorter when TG2 is in the closed conformation since the mCER and YFP probes are closely juxtaposed which facilitates energy transfer (Fig. 2A).

Fig. 2. Impact of SFN treatment on TG2 structure and catalytic activity.

A Schematic model showing the structure of the closed and open TG2 structures. The mCER donor (N-terminal) and YFP acceptor (C-terminal) fluorescent probes are indicated. B SCC-13 cells were electroporated with 3 μg of mCER-TG2 or mCER-TG2-YFP and then incubated with 0 – 20 μM SFN for 48 h prior to determination of fluorescence lifetime. mCER donor probe intensity and lifetime fluorescent images (left) and lifetime distribution histograms (middle) are shown. The plot (right) shows the mean lifetime ± SEM. The double asterisks indicate a significant increase relative to control, n = 6, p < 0.001. C SFN treatment suppresses TG2 transamidase activity. SCC-13 cells were pre-loaded with fluorescein cadaverine (FC), treated with 20 μM SFN for 24 h and then challenged with calcium ionophore before visualization of intracellular TG2-dependent incorporation of FC into cellular structures. The images, which are representative of three separate experiments, show that the SFN treatment reduces TG2 transamidase activity as measured by reduced FC crosslinking. D SFN impact on GTP binding to endogenous TG2. SCC-13 cells were incubated for 24 h with 0 or 20 μM SFN and extracts were prepared and incubated with GTP-agarose beads. The beads were then washed, boiled in Laemmli buffer and cell equivalent amounts of total extract and pulldown samples were electrophoresed for immunoblot detection of TG2. Cell equivalent amounts of total extract and pulldown samples were electrophoresed in parallel to facilitate comparison of TG2 bound versus total. E Structure of SFN and Biotin-ITC. The isothiocyanate group is −N=C=S. F Biotin-ITC was incubated with SCC-13 cell total extract for 2 h at 4 C and the samples were passed over a Zeba column before incubation with streptavidin-agarose. Total Lysate (L) and streptavidin unbound flow-through (U) and bound (B) fractions were collected and electrophoresed for anti-TG2 immunoblot. G SCC-13 cells were electroporated in the presence of 0 or 100 μM Biotin-ITC, cultured for 24 h and extracts were prepared, passed over a Zeba column to remove free Biotin-ITC and incubated with streptavidin-agarose for 1.5 h at 4 C. Equal cell equivalents of unbound (U) flow-through and streptavidin bound (B) sample was electrophoresed and the blot was then incubated with anti-TG2. H Biotin-ITC (100 μM) was incubated with SCC-13 cell total extract for 2 h at 4 C in the presence or absence of SFN (20-fold molar excess) and the samples were passed over a Zeba column before incubation with streptavidin-agarose. Streptavidin unbound flow-through (U) and bound (B) fractions were collected and equal cell equivalents were electrophoresed for anti-TG2 immunoblot.

mCER-TG2-YFP expressing cells were treated with increasing concentrations of SFN and fluorescent lifetime was monitored. The fluorescent cell images in Fig. 2B (left) confirm mCER-TG2-YFP expression, and the blue and green lifetime plots (middle) show the lifetime distribution for control and 10 μM SFN-treated mCER-TG2-YFP expressing cells. The plots reveal that SFN treatment causes a shift in lifetime from 2.92 to 3.0 ns. The maximal possible lifetime determined using the mCER-TG2 construct (absence of YFP acceptor) is 3.20 ns (orange distribution). Although the numerical change (2.92 to 3.0) is small, this change indicates a substantial shift in distance between the probes which represents a marked change in protein conformation. The bar graph shown in Fig. 2B (right) quantifies the findings from multiple SFN treatment experiments and shows that the lifetime increases as a function of increasing SFN concentration. These findings suggest that SFN treatment shifts the conformational equilibrium to favor an open/extended conformation.

Impact of SFN on TG2 transamidase and GTP binding activity

The fact that SFN alters TG2 conformation predicts that function will also be altered²¹. We therefore measured the impact of SFN treatment on TG2 activity. To measure transamidase activity, cells were preloaded with a TG2 transamidase substrate, fluorescein cadaverine (FC)^52,53, and then treated with 0 or 20 μM SFN for 30 min. The cells were then incubated with calcium ionophore for 90 min. Ionophore treatment permits intracellular calcium accumulation which activates TG2 transamidase activity²¹. A reduction in TG2 transamidase activity results in reduced FC crosslinking as evidenced by the loss of cell-associated fluorescence⁵⁴. TG2 transamidase competent cells appear green due to crosslinking of FC to cellular proteins (Fig. 2C), and this incorporation is markedly less in SFN-treated cells, confirming that SFN inhibits TG2 transamidase activity.

We next measured the impact of SFN on TG2 GTP-binding. Cells were incubated with 0 or 20 μM SFN for 24 h and extracts were prepared, passed over a GTP-agarose column and the amount of GTP bound TG2 was detected by immunoblot^9,15. The gel lanes contain identical cell equivalents of extract to permit direct comparison of the amount of GTP bound versus total TG2. The experiment shows (Fig. 2D) that essentially all TG2 is GTP-bound in extract from untreated and SFN treated cells suggesting that SFN does not reduce TG2 GTP-binding activity.

SFN covalent interaction with TG2

SFN could act to indirectly influence TG2 structure and function by regulating cellular events that impact TG2 function or by direct binding to TG2. Based on the reactivity of its isothiocyanate moiety, we considered that SFN may directly interact with TG2. To assess this possibility, we synthesized a biotinylated analog of SFN, Biotin-ITC (biotin-cadaverine-isothiocyanate) (Fig. 2E), as a tool to detect Biotin-ITC/TG2 interaction. The synthesis of Biotin-ITC is described in the supplemental materials. We incubated Biotin-ITC with cell extract and then collected biotin-labeled proteins by incubating the extract with streptavidin-agarose beads. Fig. 2F reveals the presence of streptavidin-bound biotin-ITC-TG2 complex in Biotin-ITC treated extract, but an absence of this complex in untreated extract, suggesting that Biotin-ITC directly complexes with TG2. We next monitored intracellular interaction of Biotin-ITC with TG2. Biotin-ITC was delivered to cells by electroporation. Extracts were prepared and incubated with streptavidin-agarose, and the level of streptavidin-retained Biotin-ITC-TG2 complex was detected by immunoblot. Fig. 2G identifies the Biotin-ITC/TG2 complex in Biotin-ITC treated cells, and its absence in untreated cells. As expected, the level of Biotin-ITC-TG2 complex is low because Biotin-ITC uptake into cells is modest even when delivered by electroporation.

Although Biotin-ITC and SFN are similar in structure, particularly regarding the isothiocyanate moiety (−N=C=S), they are not identical (Fig. 2E). For this reason, we wished to provide evidence that SFN can compete with Biotin-ITC for TG2 binding. We incubated cell extract with Biotin-ITC in the presence and absence of excess SFN and collected Biotin-ITC labeled TG2 on streptavidin beads to measure Biotin-ITC labeled TG2 level. Fig. 2H shows that unbound TG2 is reduced and bound TG2 increased in Biotin-ITC treated extract. Moreover, incubation with a 20-fold molar excess of SFN markedly reduces Biotin-ITC interaction with TG2, suggesting that Biotin-ITC and SFN share a similar or identical site on TG2.

SFN regulation of TG2 in HaCaT cells

To assess if SFN regulates TG2 in another epidermis derived cell line, we examined the impact of SFN on TG2 function in HaCaT cells. The fluorescent cell images in Fig. 3A (left) confirm mCER-TG2-YFP expression, and the blue and green lifetime plots (middle) show the lifetime distribution following treatment with 0 or 5 μM SFN. SFN treatment shifts the lifetime from 2.93 to 2.99 ns. The maximal lifetime, determined using the mCER-TG2 construct, is 3.18 ns (orange distribution). The change in lifetime (2.93 to 2.99 ns) indicates a substantial shift in TG2 conformation. The bar graph (right) provide quantitative data showing that SFN shifts TG2 to favor an open/extended conformation.

Fig. 3. Impact of SFN treatment on TG2 function in HaCaT cells.

A HaCaT cells were electroporated with 3 μg of mCER-TG2 or mCER-TG2-YFP and then incubated with 0 or 20 μM SFN for 48 h prior to fluorescent lifetime determination. mCER donor probe intensity and lifetime images (left) and lifetime distribution histograms (middle) are shown. The plot (right) shows the mean lifetime ± SEM. The double asterisks indicate a significant increase, n = 6, p < 0.001 for mCER-TG2-YFP expressing cells. B SFN treatment suppresses TG2 transamidase activity. HaCaT cells were pre-loaded with fluorescein cadaverine (FC), treated with 20 μM SFN for 24 h and then challenged with calcium ionophore before visualization of intracellular TG2-dependent incorporation of FC into cellular structures. C SFN impact on GTP binding to endogenous TG2. HaCaT cells were incubated for 24 h with 0 or 20 μM SFN and extracts were prepared and incubated with GTP-agarose beads. The beads were then washed, boiled in Laemmli buffer and cell equivalent amounts of total extract and pulldown samples were electrophoresed to compared TG2 bound versus total. D Biotin-ITC (100 μM) was incubated with HaCaT cell total extract for 2 h at 4 C in the presence or absence of SFN (20-fold molar excess) and the samples were passed over a Zeba column before incubation with streptavidin-agarose. Streptavidin unbound flow-through (U) and bound (B) fractions were collected and equal cell equivalents were electrophoresed for anti-TG2 immunoblot. E HaCaT spheroid derived ECS cells were seeded into a Trans Well chamber atop a Matrigel layer and treated with 0 or 20 μM SFN and cell invasion was monitored at 18 h. The asterisks indicate a significant reduction, n = 3, p < 0.001. F ECS cells were plated at confluent density on conventional culture plates and then scratched to create uniform wounds. SFN was added and wound closure was monitored at the indicated times.

We next measured the impact of SFN treatment on TG2 transamidase and GTP-binding activities. The cells were loaded with fluorescent cadaverine (FC), treated with 0 or 20 μM SFN and challenged with calcium. Control cells appear green due to TG2 transamidase-dependent crosslinking of FC to intracellular proteins and this is markedly inhibited by SFN treatment (Fig. 3B) indicating that TG2 transamidase activity is suppressed. To measure the SFN impact on GTP binding, cells were incubated with 0 or 20 μM SFN for 24 h and extracts were prepared to monitor TG2 binding to GTP-agarose^9,15. Essentially all of the TG2 is GTP-bound in control HaCaT cells (Fig. 3C) and GTP binding is not reduced by SFN treatment. Fig. 3D shows that unbound TG2 is reduced and bound TG2 increased in Biotin-ITC treated HaCaT cell extract. In addition, incubation with a 20-fold molar excess of SFN markedly reduces Biotin-ITC interaction with TG2, suggesting that these ligands bind at a similar or identical site. To assess the impact of SFN on the HaCaT cell ECS cell phenotype, we monitored the impact of treatment on HaCaT-derived ECS cell invasion and migration. Fig. 3E/F shows that SFN markedly suppresses these endpoints.

Impact of SFN on the conformation of recombinant TG2

The experiments in Figs. 2B and 3A indicate that SFN treatment shifts TG2 from a closed/folded to an open/extended conformation. However, these experiments do not reveal if other cell proteins are required to produce this conformation change. To test this, we incubated recombinant TG2 with SFN, and other isothiocyanates, and measured TG2 conformation by native gel electrophoresis⁵⁵ where the closed/folded conformation TG2 migrates more rapidly than the open/extended conformation^{8–10,15,56,57}. Recombinant TG2 was incubated for 1 h with SFN and then supplemented with 1 mM MgCl₂ and 0 or 500 μM GTP for 1 h before electrophoresis. It has been reported that GTP treatment shifts TG2 to a closed/folded conformation²¹. Fig. 4A shows that purified recombinant TG2 migrates in the closed conformation in the absence or presence of GTP and that addition of 500 μM SFN partially shifts TG2 to the extended conformation. We further show that other isothiocyanates, including allyl isothiocyanate (AITC), benzyl isothiocyanate (BITC) and butyl isothiocyanate (BUITC), also partially shift TG2 to the extended conformation.

Fig. 4. SFN impact on purified human recombinant TG2.

A Recombinant TG2 was treated with 0 or 500 μM of each isothiocyanate in the presence of 0 or 500 μM GTP for 2 h at 4 C and equivalent amounts of protein were electrophoresed on native acrylamide gel. The extended and closed confirmations are indicated. The isothiocyanate chemical structures are indicated. B Recombinant TG2 was incubated with 100 μM Biotin-ITC or SFN for 2 h at 4 C and total samples (T) or samples passed over GTP-agarose (B) containing equal TG2 equivalents were electrophoresed and incubated with anti-TG2. C Recombinant TG2 was incubated with the indicated level of SFN for 2 h at 4 C and total and GTP agarose bound TG2 was measured by immunoblot. D/E Recombinant TG2 was incubated with increasing concentrations of SFN or Biotin-ITC to generate the saturation binding (top panels) and saturation kinetic (bottom panels) plots. F Impact of SFN on the structure and activity of intracellular TG2. The figure shows the TG2 amino-terminal β-sandwich domain (yellow) which binds fibronectin and integrins, the catalytic core (red) that contains the transamidase catalytic triad and two carboxyl-terminal β-barrel domains (blue, green) one of which (blue) contains that GTP/GDP binding site. TG2 exists in cells (left panel) in a closed/folded conformation that binds GTP (GTP binding +) to drive the cancer phenotype (Pro-Cancer +) but has low transamidase activity (Transamidase −)⁷. Irreversible SFN binding to the TG2 transamidase site (middle panel) inhibits transamidase activity and shifts the conformational equilibrium to an open/extended conformation that reduces GTP binding to reduce TG2 stimulation of the cancer phenotype (GTP binding −, Transamidase −, Pro-Cancer −). We propose that the SFN-bound open/extended conformation may be reversed by intracellular GTP to reestablish the closed conformation in a fraction of the molecules (right panel) which would be transamidase negative but display some TG2 GTP binding activity (GTP binding +, Transamidase −, Pro-Cancer +).

We next treated recombinant TG2 with 0 or 100 μM SFN or Biotin-ITC for 2 h at 4 C and monitored for binding to GTP agarose^9,15. The gel lanes contain identical quantities of recombinant TG2. Fig. 4B shows that a fraction of TG2 is bound to GTP and that this amount is further reduced by treatment with 100 μM SFN or Biotin-ITC. We further characterized the impact of treating with increasing concentrations of SFN at 4 C for 2 h. Fig. 4C shows that concentrations of SFN ≥ 10 μM reduce TG2 GTP binding but that binding is not completely inhibited even at SFN concentrations of 1000 μM. These findings indicate that SFN can interact with purified TG2 to partially suppress GTP binding activity.

SFN and Biotin-ITC irreversibly inhibit TG2 acyl-transferase activity in vitro

To assess the kinetics of TG2 inactivation by SFN, we incubated recombinant TG2 with SFN or Biotin-ITC and monitored the impact on transamidase activity. The inhibition curves were fitted to a mono-exponential association model to derive first order rate constants for inhibition (k_obs). The rate constants were measured at different inhibitor concentrations and fitted to a hyperbolic saturation binding model to derive the inactivation rate constant (k_inact) and inhibition constant (K_I). Inhibition of transamidase activity was measured over a range of concentrations for each compound as shown in the top panels in Fig. 4D/E. Saturation kinetic analysis shown in the bottom panels in Fig. 4D/E provide k_inact = 0.265 ± 0.074 min⁻¹ and K_I = 366 ± 157 μM values for SFN and k_inact = 0.133 ± 0.041 min⁻¹ and K_I = 205 ± 117 μM for Biotin-ITC. The kinetics of inhibition indicate that SFN and Biotin-ITC irreversibly inhibit TG2 transamidase activity. Calculation of the k_inact/K_I ratio of 724 M⁻¹ min⁻¹ for SFN and 648 M⁻¹ min⁻¹ for Biotin-ITC indicate a low overall inhibition efficiency primarily due to the low affinity of these compounds to TG2. For comparison, NC9 has kinetic parameters of k_inact = 2.60 ± 0.17 min⁻¹, K_I = 33.9 ± 3.4 μM and a k_inact/K_I ratio of 76400 M⁻¹ min⁻¹. The two orders of magnitude lower inhibition efficiency of SFN and Biotin-ITC is due to the lower rate constants of inactivation and lower affinities for TG2 as compared to NC9.

The kinetic parameters measured for SFN and Biotin-ITC also permitted calculation of TG2 inactivation rate constants at various inhibitor concentrations. For treatment with 20 μM, the SFN k_obs value of 0.014 min⁻¹ and the Biotin-ITC K_obs value of 0.012 min⁻¹ correspond to half-lives of inhibition of 50 and 59 min, respectively indicating that the incubation times used in this paper are sufficient to optimally inactive TG2.

Discussion

TG2 is a key controller of cell function that is present in the cell cytosol, nucleus, at the plasma membrane and in the extracellular environment⁵⁸. TG2 includes an integrin- and fibronectin-binding N-terminal β-sandwich, a catalytic core encoding the Cys277/His335/Asp358 transamidase catalytic triad, and two C-terminal β-barrel domains, one of which (β-barrel1) encodes a guanine nucleotide binding site that is important in pro-cancer signaling⁵⁹ (Fig. 4F). TG2 exists in a closed/folded GTP/GDP binding (signaling) conformation in the guanine nucleotide-rich intracellular compartment^56,58,60, but shifts to an open/extended transamidase conformation in response to events that elevate intracellular calcium¹⁶. These conformation-dependent activities are mutually exclusive^{8,9,11,14,15,17}. A variety of small molecular inhibitors have been described that target TG2^{45,57,61–64}. Many of these are irreversible inhibitors designed to covalently interact with Cys277 in the catalytic triad of the transamidase site⁴⁵. Truant and associates used a novel fluorescence method to show that NC9³⁵, an irreversible inhibitor of TG2 transamidase activity^35,45, converts intracellular TG2 from a closed/folded to an open/extended conformation⁵⁶. We confirmed these findings using a host of TG2 inhibitors that bind to the transamidase active site cysteine (Cys277) to inhibit transamidase activity^21,35. Moreover, binding of these inhibitors promotes a conformational change that inhibits GTP binding activity by altering the structure of the TG2 GTP binding site²¹. Thus, although these agents specifically interact with the transamidase catalytic center, they also disorganize/inactivate the TG2 GTP binding site and this reduction in GTP binding activity leads to reduced ECS cell survival^21,35.

SFN is a promising anti-cancer isothiocyanate derived from broccoli, cabbage, cauliflower and other cruciferous vegetables⁶⁵ that is known to suppress cancer stem-like cell survival in multiple systems⁶⁶. Isothiocyanates are characterized by the presence of a highly reactive isothiocyanate group (−N=C=S) which reacts with sulfur, nitrogen, and oxygen-based nucleophiles and targets cysteines in proteins⁶⁷. However, it is an open question whether agents like SFN can regulate TG2 structure and activity as a mechanism to suppress the cancer phenotype. We therefore explored the possibility that SFN may regulate TG2 structure and activity.

To determine if SFN directly binds to TG2, we measured Biotin-ITC/TG2 complex formation in cells and cell extracts^68,69. Intact cells and cell extracts were incubated with Biotin-ITC and complexes were collected by streptavidin affinity chromatography. The ability of SFN to compete for Biotin-ITC binding suggests that both compounds bind to a common site on TG2 and validates the use of Biotin-ITC as a probe. In addition, kinetic analysis of SFN and Biotin-ITC binding to recombinant TG2 demonstrates low affinity direct and irreversible TG2 binding for each agent. We further examined the impact other related isothiocyanates and observed that they also partially shifted TG2 from a closed/folded to an open/extended conformation, as measured using native gel electrophoresis, suggesting that these compounds also trap TG2 in its open conformation. Each of these compounds have a highly reactive isothiocyanate group (Fig. 4A) which we propose is the group that reacts with the TG2 transamidase catalytic triad at Cys277 to inhibit transamidase activity and promote the shift in structure^21,35,56.

Previous studies show that NC9, a highly specific TG2 inhibitor^7,21 covalently and irreversibly binds to Cys277 in the transamidase catalytic triad to inhibit TG2 transamidase activity and this binding shifts TG2 from a closed/folded to an open/extended conformation which disorders the GTP binding domain to inhibit TG2 GTP binding^7,21,35. Based on these findings we anticipated that the SFN associated conformation change would lead to a reduction in TG2 GTP binding activity. Indeed, we show that treatment with SFN (or Biotin-ITC) inhibits recombinant TG2 transamidase activity, shifts TG2 to an open/extended conformation and partially reduces TG2 GTP binding activity. In contrast, studies in cell culture show that SFN binding reduces TG2 transamidase activity and shifts TG2 to an open/extended conformation but does not reduce TG2 GTP binding activity (Fig. 2D, Fig. 3C). We speculate that because SFN is a smaller ligand than NC9 and has a lower affinity for TG2 that it may not adequately fill the transamidase site in a way that is sufficient to quantitatively maintain TG2 in an extended conformation required to adequately disorder the GTP binding domain and attenuate GTP binding activity (Fig. 4F). Moreover, the high level of GTP present in the intracellular environment may antagonize the action of covalently bound SFN by forcing a fraction of the SFN/TG2 complex back to a GTP binding conformation (Fig. 4F), a process that may also be facilitated by other intracellular proteins. Thus, these interesting studies unequivocally demonstrate that SFN targets TG2 in cancer cells where it irreversibly binds to TG2 to alter structure and suppress transamidase activity. Although SFN binding to recombinant TG2 is associated with reduced GTP binding, additional studies will be necessary to completely understand the effects of SFN binding on TG2 GTP binding activity in intact cells.