In vivo evaluation of two tissue transglutaminase PET tracers in an orthotopic tumour xenograft model

Background The protein cross-linking enzyme tissue transglutaminase (TG2; EC 2.3.2.13) is associated with the pathogenesis of various diseases, including cancer. Recently, the synthesis and initial evaluation of two high-potential radiolabelled irreversible TG2 inhibitors were reported by us. In the present study, these two compounds were evaluated further in a breast cancer (MDA-MB-231) tumour xenograft model for imaging active tissue transglutaminase in vivo. Results The metabolic stability of [11C]1 and [18F]2 in SCID mice was comparable to the previously reported stability in Wistar rats. Quantitative real-time polymerase chain reaction analysis on MDA-MB-231 cells and isolated tumours showed a high level of TG2 expression with very low expression of other transglutaminases. PET imaging showed low tumour uptake of [11C]1 (approx. 0.5 percentage of the injected dose per gram (%ID/g) at 40–60 min p.i.) and with relatively fast washout. Tumour uptake for [18F]2 was steadily increasing over time (approx. 1.7 %ID/g at 40–60 min p.i.). Pretreatment of the animals with the TG2 inhibitor ERW1041E resulted in lower tumour activity concentrations, and this inhibitory effect was enhanced using unlabelled 2. Conclusions Whereas the TG2 targeting potential of [11C]1 in this model seems inadequate, targeting of TG2 using [18F]2 was achieved. As such, [18F]2 could be used in future studies to clarify the role of active tissue transglutaminase in disease.


Background
Transglutaminases comprise a family of enzymes responsible for the calcium-dependent intra-and intermolecular cross-linking of proteins between the side chains of glutamine and lysine residues, forming an epsilon-(gamma-glutaminyl)-lysine bond [1]. Tissue transglutaminase (TG2) is ubiquitously expressed and, under physiological conditions, plays a role in, e.g. apoptosis, cell differentiation and cell migration [2,3]. The cross-linking activity of this enzyme is tightly regulated by various mechanisms. First, TG2 exists in two distinct conformations, referred to as closed and open conformations, respectively [4,5]. Only in the open conformation, which is associated with high-calcium and low-guanosine diphosphate/guanosine triphosphate (GDP/ GTP) concentrations, the active site cysteine residue is exposed and transamidation can be expected [5]. In the closed conformation, two consecutive C-terminal β-barrels sterically limit transamidation activity [4]. Second, the redox state of TG2 determines its catalytic activity, since a Cys370-Cys371 disulphide bridge, despite locking TG2 in its open conformation, hampers transamidation [6]. Finally, cross-linking activity of TG2 is regulated by formation of ternary protein complexes on the cell surface with extracellular matrix proteins, such as fibronectin and membranebound integrins [7]. Clearly, this multitude of regulatory mechanisms poses a challenge for assessing TG2 crosslinking activity in vivo. Often changes in TG2 expression levels or immunohistochemical detection of epsilon-(gamma-glutaminyl)-lysine bonds are used as ex vivo biomarkers of TG2 activity. Alternatively, transglutaminase mediated incorporation of systemically administered biotin-labelled amine substrates can be detected immunohistochemically after sacrificing the test animal [8,9].
TG2 is strongly associated with the pathogenesis of cancer, celiac disease, and fibrotic and neurodegenerative diseases [10][11][12][13][14][15], in which its role is assumed to be related to its cross-linking activity. The fact that TG2 knock-out mice are phenotypically healthy in a stressfree environment has further boosted TG2 as a potential target for therapeutic intervention [16]. Nowadays, a wide array of TG2 inhibitors has been developed [17]. Nevertheless, further development of potent inhibitors towards clinical studies, for example by evaluation in animal models, has been limited. The availability of a validated TG2 PET tracer is likely to stimulate in vivo research of potent TG2 inhibitors, because it will allow monitoring of target engagement in vivo by novel TG2 inhibitors [18]. As a result, a deeper understanding of TG2 biology in various diseases might be obtained.
Recently, carbon-11 and fluorine-18 labelled small molecule TG2 PET tracers have been developed by our group [19,20]. [ 11 C]1 ( Fig. 1, IC 50 53 nM) was selected out of three carbon-11 labelled TG2 inhibitors based on its superior metabolic stability [19]. In addition, [ 18 F]2, a peptidic TG2 inhibitor (Fig. 1, IC 50 104 nM), was developed [20]. Despite being completely metabolised in vivo after just 15 min post injection, imaging with [ 18 F]2 was suggested as also the formed radiometabolite was previously shown to be an equipotent inhibitor of TG2 (IC 50 45 nM) [20]. Both selected compounds, [ 11 C]1 and [ 18 F]2, were able to discriminate between active and inactive tissue transglutaminase in vitro and demonstrated specific and selective binding to MDA-MB-231 tumour sections, as assessed by in vitro autoradiography experiments. However, in vitro autoradiography assays do not necessarily reflect in vivo biology. Therefore, the aim of the present study was to determine whether these new tracers are able to target TG2 also in vivo. To this end, compounds [ 18

Cell culture
MDA-MB-231 human breast cancer cells were purchased from American Type Culture Collection (Rockville, MD, USA). Cells were cultured at 37°C, 5% CO 2 in Dulbecco's modified Eagle medium with 4.5 g/L glucose (Lonza, Basel, Switzerland) with HEPES supplemented with L-glutamine and fetal calf serum (5%).

Xenograft model
Severe combined immunodeficient (SCID) female mice (6-8 weeks, 20 to 25 g, Charles River, Wilmington, MA, USA) were housed in sterile cages under standard conditions (24°C, 60% relative humidity, 12-h light/dark cycles) and provided with water and food ad libitum. MDA-MB-231 cells (1 × 10 6 ) were injected orthotopically in the fat pad of the second thoracic mammary glands (bilateral) [21]. Tumour dimensions were measured using a Vernier calliper, and tumour volume was calculated using the formula (x 2 y)/2 (x and y being the width and length, respectively) for an ellipsoid. At 8 weeks after MDA-MB-231 cell injection, tumours reached the target size of 200 mm 3 . This study was performed according to national regulations and was approved by the Animal Experimentation Ethics Committee of the VU University Medical Center.

QPCR analysis
Total messenger ribonucleic acid (mRNA) was isolated from MDA-MB-231 tumour cells or tumour tissue using Trizol Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. RNA was reverse-transcribed into complementary deoxyribonucleic acid (cDNA) using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, Ca, USA) using 0.5 μg oligo-dT primers according to the manufacturer's instructions. For the subsequent quantitative real-time polymerase chain reaction (qPCR), the Power SYBR Green Master Mix (Applied Biosystems) was used. Primers were purchased from Eurogentec (Maastricht, Netherlands), and qPCR was performed in MicroAmp Optical 96-well Reaction Plates (Applied  Biosystems) on a StepOnePlus Real-Time PCR system (Applied Biosystems). The reaction mixture (20 μL) was composed of 1 × Power SYBR Green buffer (Applied Biosystems), 3.75 pmol of each primer (see Table 1 for primer details), and 12.5 ng cDNA. The thermal cycling conditions were an initial 10 min at 95°C followed by 50 cycles of 15 s at 95°C and 1 min at 60°C. The specificity of the reaction was checked by means of melt curve analysis. Relative expression levels of the target genes were determined by LinRegPCR software (version 2014.3; website: http://www.hfrc.nl) using the following equation N 0 = N q / (E^C q ) (N 0 = target quantity, N q = fluorescence threshold value, E = mean PCR efficiency per amplicon, C q = threshold cycle) [22], after which the value was normalised to the expression level of the reference gene glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) using the following formula (N o, gene of interest /N o, GAPDH ). Results are expressed as gene expression relative to GAPDH ± standard deviation (n = 4 for both tumour cells and tumour tissue measurements).

Chemical synthesis General
All reagents were obtained from commercial sources (Sigma Aldrich, St. Louis, USA). Solvents were obtained from Biosolve (Valkenswaard, the Netherlands) and used   13 C 77.16ppm). Electrospray ionisation-high resolution mass spectrometry (ESI-HRMS) was carried out using a Bruker microTOF-Q instrument in positive ion mode (capillary potential of 4500 V).
The synthesis of unlabelled 2 was performed as published elsewhere [20]. Synthesis of ERW1041E was performed according to published procedures (Scheme 3) [9]. Analytical characterizations were in accordance with reported values [9,23].

PET imaging
Dynamic PET imaging was performed using dedicated small animal NanoPET/CT and NanoPET/MR scanners (Mediso Ltd., Hungary, Budapest) [24,25] with identical PET components. Mice (n = 4 per group) were anaesthetized with 4 and 2% isoflurane in 1 L · min −1 oxygen for induction and maintenance, respectively. Mice were positioned on the scanner bed, and the respiratory rate was monitored for the duration of the scan, adjusting anaesthesia when required. Reconstruction was performed with a fully 3-dimensional (3D) reconstruction algorithm using four iterations and six subsets, resulting in an isotropic 0.4-mm voxel dimension. Images were analysed using the freely available AMIDE-software version 1.0.4 (retrieved from https:// sourceforge.net/projects/amide/files/amide/1.0.4). Regions of interest (ROIs) were drawn around the tumour tissue and leg muscle. Results are expressed as percentage injected dose per gram (%ID/g). Error bars indicate standard deviation. After PET scanning experiments, animals were sacrificed by cervical dislocation, tumours were isolated, and stored at − 80°C until further use.

Haematoxylin and eosin staining
MDA-MB-231 tumour sections (10 μm) were dried and fixed with acetone (100%) for 10 min and subsequently dried at rt. Sections were then rehydrated in Tris buffered saline (TBS; two times 5 min) and demiwater (5 min) and stained with Mayer's haematoxylin solution (3 min) followed by rinsing with tap water (5 min). The sections were stained with 1% eosin Y solution (10-30 s) followed by dehydrating by sequential dipping in ethanol (70, 90, 96, 100 and 100%) and xylene. Sections were then mounted with coverslips using Entellan. Microscopy images were obtained using a Leica DN5000B microscope (Leica Microsystems, IL, USA).

Immunohistochemical staining
Immunohistochemical staining of TG2 was performed as described previously with minor modifications [19]. Fresh frozen MDA-MB-231 tumour sections (10 μm) were dried and fixed with acetone (100%) for 10 min, dried at rt and subsequently rehydrated using TBS (three times 5 min). Endogenous peroxidase activity was blocked with 0.3% H 2 O 2 and 0.1% NaN 3 in TBS for 15 min and then washed with TBS (three times 5 min). After blocking with 3% bovine serum albumin (BSA) in TBS with 0.5% TritonX-100 (TBS-T) for 20 min, incubation with polyclonal goat anti-guinea pig TG2 antibody (Upstate, Merck Millipore, Billerica, MA, USA) in TBS-T with 3% BSA was performed overnight at 4°C (dilution 1:4000). A negative control experiment was performed by omitting the primary antibody (results not shown). Sections were then washed with TBS (three times 5 min) prior to incubation with biotinylated donkey anti-goat secondary antibody (Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA, dilution 1:400) for 2 h at rt. Excess antibody was removed by washing with TBS (three times 5 min), and sections were incubated for 1 h with horseradish peroxidase avidin-biotin complex. After sequential washing with TBS (two times 5 min) and Tris-HCl (5 min), peroxidasemediated [1,1′-biphenyl]-3,3′,4,4′-tetraamine (DAB) oxidation was used for visualising TG2. Colouring was monitored ad oculos. After colouring was satisfactory, sections were washed sequentially with Tris-HCl (two times 5 min) and water (5 min). Dehydration was carried out by sequential dipping in ethanol (70, 90, 96, 100 and 100%) and xylene, and sections were mounted with coverslips using Entellan. Microscopy images were obtained using a Leica DN5000B microscope (Leica Microsystems, IL, USA).

Histochemical staining
Histochemical staining of active TG2 was performed as described previously [19]. MDA-MB-231 tumour sections (10 μm) were pre-incubated in 100 mM Tris-HCl buffer (pH 7.4), 5 mM CaCl 2 , 1 mM dithiothreitol (DTT) rt (20 min). As negative control, the selective TG2 inhibitor Z006 was added to this solution at a final concentration of 100 μM (results not shown). Next, sections were incubated in 100 mM Tris-HCl buffer (pH 7.4), 5 mM CaCl 2 , 1 mM DTT and the TG2 amine donor substrate 5-(biotinamido)pentylamine (BAP; 0.05 mM) at 37°C for 30 min. After a short wash with TBS and water, the sections were dried at rt. Then, the sections were fixed with acetone for 10 min followed by washing with Tris buffered saline (TBS). Sections were blocked with 0.1% NaN 3 and 0.3% H 2 O 2 in TBS for 15 min, washed with TBS (three times 5 min) and incubated for 1 h with horseradish peroxidase avidin-biotin complex. After washing (two times 5 min TBS, then 5 min Tris-HCl), peroxidase was developed by addition of DAB and colouring was monitored ad oculos. Nuclear staining using haematoxylin was performed. Sections were washed with Tris-HCl and water, dehydrated by sequential dipping in ethanol (70, 90, 96, 100 and 100%) and xylene, and mounted with coverslips using Entellan. Microscopy images were obtained using a Leica DN5000B microscope (Leica Microsystems, IL, USA).

Autoradiography
Autoradiography was performed essentially as described previously [20]. MDA-MB-231 tumour sections (10 μm) were washed three times with 50 mM Tris-HCl buffer (pH 7.4) for 5 min. Sections were dried under a gentle air flow before incubation for 30 min with [ 18 F]2 (0.1 MBq · mL −1 ) in 5 mM Tris-HCl, pH 7.4, 5 mM CaCl 2 , 1 mM DTT. As a negative control inhibitor 1 was added to this incubation solution at 100 μM (results not shown). Washing was performed using 5 mM Tris-HCl (three times) followed by dipping in ice cold water. After drying in an air stream, tumour sections were exposed to a phosphorimaging screen (GE Healthcare, Buckinghamshire, UK) for 15 min and developed on a Typhoon FLA 7000 phosphor imager (GE Healthcare, Buckinghamshire, UK). Visualisation of binding was performed using Image-QuantTL v8.1.0.0 (GE Healthcare, Buckinghamshire, UK).

Statistical analysis
Where relevant, statistical analysis was performed using either a one-tailed paired Student's t test or a two-tailed paired Student's t test with a confidence interval of 95%.

mRNA expression of various human transglutaminases in MDA-MB-231 cells and tumour xenograft tissue
The expression of various tumour-related transglutaminases was determined by means of qPCR on both in vitro cultured MDA-MB-231 tumour cells and ex vivo MDA-MB-231 tumour material obtained after tumour inoculation in SCID mice. RNA expression levels of transglutaminase types 1-3 and 5 and blood coagulation factor XIII (TG1, TG2, TG3, TG5 and FXIII) were quantitatively determined relative to GAPDH RNA expression (Fig. 2). In MDA-MB-231 tumour cells as well as in tumour tissue, TG2 mRNA was most abundant. Low levels of TG1 mRNA were observed in both cells and tissue, whereas TG3 and TG5 mRNA were absent. FXIII mRNA was absent in cultured MDA-MB-231 tumour cells, whereas in tumour tissue low levels, relative to TG2 mRNA, were found.

Metabolite analysis of [ 11 C]1 and [ 18 F]2 in SCID mice
In SCID mice, compound [ 11 C]1 demonstrated moderate metabolism, with 24% intact tracer after 45 min (Table 2). Mainly polar metabolites were formed. [ 18 F]2 was fully metabolised to a single metabolite after 45 min. The metabolite was identified by LC-MS/MS analysis as the demethylated parent M1 (Fig. 3).  Fig. 4a, b, respectively. The [ 18 F]FDG scans revealed rather low uptake in the centres of the tumours, indicating mainly  Following i.v. administration in healthy animals, anaesthetised using isoflurane, blood plasma was obtained at 15 and 45 min. Blood plasma was separated into a polar and non-polar fraction using a SPE method. Non polar fractions were analysed on HPLC. Results are expressed as average percentage of total blood plasma activity ± standard deviation (n = at least three per data point). *Animals were sacrificed at 60 min post injection.
that the rims of the tumours were viable. Under baseline conditions, uptake of [ 11 C]1 in the tumour at baseline conditions showed a peak at 5-10 min after injection (Fig. 4a). Washout, however, was fast and activity concentrations were comparable to those in background tissue (i.e. the muscle, right femur). Furthermore, blocking of TG2 by pretreatment with ERW1041E did not result in lower tumour activity levels, but rather resulted in a counterintuitive significant increase in tumour activity, as well as in increased muscle activity concentration. PET scanning using compound [ 18 F]2 showed a time-dependent accumulation in tumour tissue up to 1.7 %ID/g at the 40-60 min time-frame, suggesting irreversible tumour targeting. Background values (the muscle, right femur) reached 0.8 %ID/g at this time-point, which is significantly lower (p = 0.0004) and did not display an increase in tissue activity over the scanning period (Fig. 4b). Pretreatment of the animals with the TG2 inhibitor ERW1041E resulted in a decrease in activity accumulation in the tumour tissue to 1.4 %ID/g, although this difference was not statistically significant (p = 0.06). A drastic and significant decrease in tumour activity concentration, however, was observed when unlabelled 2 was co-administered, to approximately 1.0 %ID/g (p = 0.007).
TG2 expression in MDA-MB-231 tumour tissue-immunohistochemistry and autoradiography on tumour tissue Following sacrifice of the animals, tumour sections were evaluated histochemically for TG2 expression and TG2 activity (representative sections of tumour tissues of three separate animals are shown in Fig. 5). In all tumours, haematoxylin/eosin staining (Fig. 5a) demonstrated areas with low nuclei concentrations in the core of the tumour, hinting at central necrosis. Immunohistochemical staining for TG2 expression (Fig. 5b) resulted in a distribution pattern highly resembling the haematoxylin/eosin staining, showing that TG2 is predominantly expressed in the viable part of the tumours. Transglutaminase-mediated BAP incorporation (Fig. 5c) and in vitro autoradiography employing [ 18 F]2 (Fig. 5d), both means of measuring the open (active) conformation of transglutaminase, demonstrated a similar distribution pattern as observed with the anti-TG2 antibody and the haematoxylin/eosin staining, i.e. incorporation of the substrate and inhibitor in the viable areas of tumour tissue.

Discussion
Transglutaminase mRNA expression in both MDA-MB-231 tumour cells and xenograft tissue essentially was limited to TG2 mRNA and exceeded other transglutaminases by at least a factor ten.  [19,20]. Consequently, these findings imply that the MDA-MB-231 xenograft model is suitable for TG2 PET tracer evaluation. Previously, metabolic stability of [ 11 C]1 as well as [ 18 F]2 was determined ex vivo in blood plasma of healthy Wistar rats. This study, however, was performed in a xenograft model in SCID mice, and therefore, the metabolic stability of both radiotracers was determined again in SCID mice and results were compared with previously obtained rat data [19,20]. Metabolism of both [ 11   followed the same pattern in both species, although in mice the conversion was less rapid. This difference is likely a result of species-dependent esterase activity [26], and comparable differences between methylester hydrolysis in mice and rats have been reported elsewhere [27]. Importantly, the peptidic backbone and the diazoketone functionality in [ 18 F]2, responsible for selective and irreversible binding, respectively, appear metabolically stable.
To prevent the development of slow-growing and highly necrotic tumours, an orthotopic mouse model was used, in which tumour cells were inoculated in the 2nd thoracic mammary fat pad [21]. At this position minimal interference with surrounding organs such as liver and heart is expected, thus facilitating definition of ROIs and PET data analysis. The use of the structurally unrelated TG2 inhibitor ERW1041E served as a means of determining selectivity of binding of TG2 radiotracers in tumour tissue. ERW1041E was administered at 50 mg · kg −1 , as it has been shown that this dose results in inhibition of TG2 activity to background levels in a mouse model of pulmonary hypertension [9]. In addition, the maximum TG2 inhibitory effect was observed 30 min after ERW1041E administration [9].
Compound [ 11 C]1 showed poor tumour uptake and relatively fast washout from the tumour, which was unexpected for an irreversibly binding radiotracer. Therefore, it is believed that the signal is indicative of perfusion rather than TG2 targeting. As both compound 1 and ERW1041E inhibit TG2 by irreversibly binding to the active site cysteine residue, a reduction in tumour uptake would be expected upon blocking with ERW1041E. The fact that both tumour and muscle activity concentrations increased as a result of ERW1041E pretreatment suggests to us a decreased clearance of [ 11 C]1 or its derived radioactive metabolites from the blood, although no further experiments were performed to support this claim. Despite the promising results that were previously obtained by in vitro autoradiography, where high specific and selective binding of [ 11 C]1 to MDA-MD-231 tumour sections was observed [19], [ 11 C]1 is not effective as a TG2 PET tracer in vivo. It is unlikely that this apparent ineffectiveness is a result of in vivo metabolism of [ 11 C]1 over the time-course of these experiments, as metabolism is moderate. Potentially, large differences between in vitro TG2 inhibitory potency and inhibition in actual biological systems could explain the inability of this compound to image active TG2 in vivo. Different classes of acryl amide based TG2 inhibitors, although highly potent against isolated TG2, have shown a reduction in inhibitory potency in cell or cell lysate assays [28,29]. The inhibitory potency of compound 1 in similar cell or cell lysate assays is unknown as such experiments have not been described [30].
In contrast to [ 11 C]1, [ 18 F]2 displayed higher tumour uptake, which increased over time, indicating irreversible binding of [ 18 F]2 to the tumour tissue. Uptake decreased after pretreating animals with ERW1041E, although the difference was not statistically significant (p = 0.06). This seemingly partial inhibition of TG2 might be due to the limited TG2 inhibitory potency of ERW1041E (K i : 11 μM [23]) compared with compound 2, although previously it has been shown that this compound was able to at least partially inhibit intestinal TG2 in a mouse model [8] and also in a mouse model of pulmonary hypertension to baseline levels [9]. Alternatively, the chosen time-point for pretreatment with ERW1041E (set at 30 min prior to tracer administration) might not be optimal for full TG2 inhibition, although at this similar time-point a reduction of TG2 activity by a factor of four in a pulmonary hypertension model was observed [9]. The blocking effect when co-administering unlabelled 2 and [ 18 F]2 (p = 0.007), implies that [ 18 F]2 is useful for imaging of local TG2 activity in vivo by indicating specific binding of [ 18 F]2. However, to establish the selectivity of this uptake, more studies are required.
Analysis of tumour sections showed that TG2 was expressed mainly in the more viable rim of the tumours and less in the necrotic core of the tumours. This decreased tumour viability in the core of the tumours is in close accordance with [ 18 F]FDG findings, which generally displayed higher uptake in the outer tumour areas. Because %ID/g was determined by drawing an area over the full tumour volume, areas inside the tumour that contain no tissue transglutaminase potentially underestimate the actual potential of [ 18 F]2 towards TG2 targeting. As PET images using [ 18 F]2 depicted higher activity concentrations in the outer area of the tumour, for future research, it might prove beneficial to perform such an imaging study at an earlier time-point in tumour development, potentially limiting tumour necrosis and thus increasing signal to noise ratios.
Although previously relatively high tissue transglutaminase expression levels in MDA-MB-231 tumour cell lysates were observed [10], and both TG2 expression and activity were further confirmed in xenografted MDA-MB-231 tumour tissue by means of immunohistochemistry and in vitro autoradiography [19,20], it is unknown to what extent TG2 shows transamidation activity in a biological setting such as the present MDA-MB-231 tumour model. Obviously, only in the open TG2 conformation, active site-directed PET tracers can be successfully applied for TG2 imaging [5,18]. Therefore, it is expected that primarily extracellular TG2, which is more likely to be in an open conformation due to high extracellular calcium concentrations, will be accessible for such tracers [18]. Previous in vitro studies have demonstrated that TG2 is highly expressed on the plasma membrane of MDA-MB-231 tumour cells [31]. Furthermore, pharmacological inhibition of TG2 on MDA-MB-231 cells using cell-impermeable small molecule inhibitors or antibodies resulted in decreased cell migration [31,32] and decreased invasiveness [33], which are both hallmarks of cancer [34]. Based on these findings, it is expected that the accumulation of activity in the tumour using [ 18 F]2 is due to targeting of extracellular TG2. Evidence suggests that extracellular TG2 plays a role in tumour progression by means of its transamidation activity, and therefore, the MDA-MB-231 tumour model appears suitable for evaluation of TG2 PET tracers in vivo, which is supported by the imaging results using [ 18