**1. Introduction**

Glioblastoma multiforme (GBM) is the most common primary tumour of the central nervous system. Although the global incidence is rare with less than 10 per 100,000 people, the median survival rates for patients with GBM remain dramatically low despite complex surgical, pharmacological and radiation therapy approaches [1,2]. An important aspect contributing to this poor outcome is the genetic heterogeneity of GBM, which translates into heterogeneous expression patterns of potentially druggable targets [3]. Accordingly, the development of new targeted therapies as well as of biomarkers for predictions of treatment response would benefit from an improved understanding of how such spatiotemporal patterns evolve and change during pathogenesis [4–9]. Nuclear medicine imaging techniques offer a unique possibility to noninvasively assess the distribution and amount of certain biological targets and thus to contribute significantly to the drug-discovery process and later on to the evaluation of the treatment efficacy [10–12].

By application of suitable radiolabeled molecules, positron emission tomography (PET) in particular can assess such alterations with high sensitivity. Imaging agents for the investigation of the catabolic and anabolic metabolism can detect cancer-specific alterations in high-capacity processes such as glycolysis (by [18F]FDG), amino acid transport (by [11C]MET or [18F]FET), and membrane turnover (by [18F]FMC) [13,14]. They are currently utilized to improve the clinical management of brain cancer patients. Furthermore, the PET technology offers the principal possibility to investigate differences in the expression pattern and activity of diagnostically and therapeutically relevant proteins, such as receptors or enzymes, and to correlate them with tumour heterogeneity and aggressiveness. The current development of radiolabelled probes to image e.g., isocitrate dehydrogenase mutations (IDH1R132H) [15], or the glutamate carboxypeptidase II (prostate-specific membrane antigen, PSMA) [16], reflects the interest in preclinical and clinical research on detailed and targeted molecular characterisation of malignancies in the brain, which is a prerequisite to define the role of nuclear medicine imaging for the individualized treatment of patients with GBM [14,17].

Our research on the identification of new targets for brain cancer imaging focuses on the sigma-1 receptor (sig1R), an intracellular chaperone protein highly expressed in a variety of cancers including GBM [18,19]. Under physiological conditions, the sig1R is localized at the mitochondrion-associated endoplasmic reticulum membrane (MAM) and at the plasma membrane and is involved by interactions with other proteins in a number of pathways related to the metabolism and proliferation of cells. Accordingly, albeit at different levels, sig1R is expressed in all peripheral organs as well as in the central nervous system. Widely distributed in the brain [20], the sig1R is involved in memory, emotional, and sensory functions and changes in its expression are related to neurodegenerative diseases such as Huntington 's disease and Alzheimer's disease, as well as in stroke, depression and pain disorders [20,21]. Most likely, due to the translocation of the intracellular receptor from MAM to the plasma membrane and the cell nucleus, which is triggered under pathological conditions [22–25], sig1R is functionally involved into a variety of cellular pathways related to stress response and survival [25–27]. In addition, the expression of sig1R seems to be upregulated by cancer-specific mechanisms, as indicated by the high levels of sig1R protein discovered in many cancer cell lines [19,26–28]. The antiproliferative effect of pharmacological inhibition of sig1R by putative antagonistic ligands on cancer cell lines further substantiates the potential role of sig1R in cancer biology [27,29–33]. Sig1R ligands influence apoptosis, migration, and cell cycle progression pathways through their interaction with voltage-dependent K<sup>+</sup> channels, volume-regulated Cl<sup>−</sup> channels, or endoplasmic reticulum Ca2<sup>+</sup> release [22,31,32,34]. Altogether, the available data present strong evidence of an important role of sig1R in tumour biology, and for that reason, PET noninvasive molecular imaging of sig1R is assumed to improve our understanding of the role of this particular protein in tumour pathophysiology and to promote the development of a sig1R-targeted-therapies [35].

Already different radiotracers have been developed to investigate the expression of sig1R by PET such as [18F]FMSA4503 [36], [18F]SFE or [18F]FTC-146 [37,38]. However, only [11C]SA4503 and (S)-(−)-[18F]fluspidine were applied in research on the in vivo imaging of sig1R in brain tumours using heterotopic brain tumour models, as by the group of van Waarde, or orthotopic models, as by our group [39,40].

Encouraged by the approval of the in-house developed radiopharmaceutical (S)-(−)-[18F]fluspidine for clinical trials (EudraCT Numbers: 2014-005427-27, 2016-001757-41), the promising results of a collaborative pilot study in an orthotopic mouse model of GBM [40], and the establishment of orthotopic brain tumour models in our group, we decided to investigate further the relevance of sig1R in brain cancer biology and to evaluate the potential of (S)-(−)-[18F]fluspidine-PET to characterise brain tumours on a molecular level. We report herein on the assessment of GBM-specific expression of sig1R by a combination of in vitro and in vivo approaches using mice bearing intracranial U87-MG tumours as a preclinical orthotopic model of human GBM and the sig1R-specific radioligand (S)-(−)-[18F]fluspidine. Initially, we validated the in vivo selectivity of (S)-(−)-[18F]fluspidine for sig1R in a sig1R-knockout mouse model. Then, we validated by radioligand binding assays the suitability of the U87-MG cell line used for the orthotopic GBM model regarding the presence of the target, and confirmed by autoradiography the unimpaired overexpression under in vivo conditions. By immunohistochemistry as radioligand-independent method, we could confirm the expression and overexpression of sig1R protein in U87-MG cells in 2D culture as well as in the cellular environment of the mouse brain. Finally, we performed dynamic PET/MRI (magnetic resonance imaging) studies to assess the pharmacokinetics of (S)-(−)-[18F]fluspidine in the orthotopic U87-MG mouse model of human GBM, and report on the very first detection of sig1R protein in human GBM tissue by means of in vitro autoradiography, validating the relevance of this target.

### **2. Results**

#### *2.1. Expression of sig1R in U87-MG Cells*

#### 2.1.1. Expression of sig1R in U87-MG Cells in 2D Cell Culture

We initially evaluated the expression of sig1R in U87-MG cells, a human primary glioblastoma cell line, grown in 2D cell culture by radioligand binding assays and immunohistochemistry to determine their suitability for the intended orthotopic mouse model of GBM. By a single saturation assay using (+)-[3H]pentazocine, an established sig1R-specific radioligand, a *B*max value of 129 fmol/mg protein and a *K*<sup>D</sup> value of 2.4 nM was determined. Specific binding of (S)-(−)-fluspidine towards (+)-[3H]pentazocine-labeled binding sites in U87-MG cells has been proven by displacement studies, and the affinity of (S)-(−)-[18F]fluspidine to sig1R has been determined with a *<sup>K</sup>*<sup>D</sup> of 16.7 nM.

To verify the identity of the specific binding site of the two radioligands by an independent method, we further performed immunohistochemistry using a sig1R-specific antibody. The thereby determined cytoplasmic staining of protein in isolated U87-MG cells, which corresponds with the labelling of sig1R in the positive control HEK-293 cells overexpressing human sig1R, is demonstrated in Figure 1A, B, respectively.

**Figure 1.** Immunofluorescent staining of sigma-1 receptors (sig1R). Representative image of the sig1R staining (**A**) in U87-MG cells grew in vitro, (**B**) in HEK-293 cells overexpressing human sigma-1 receptor (hsig1R) grew in vitro and (**C**) in a cryosection of U87-MG tumour cells orthotopically implanted in a mouse brain (scale bar: 25 μm, x40, green channel: sig1R staining, blue channel: nucleus staining).

#### 2.1.2. Expression of sig1R in U87-MG Cells Grown In Vivo

Because transplantation of human cancer cells into mice might be associated with an altered expression profile due to the significant change in the microenvironment, we subsequently investigated orthotopically implanted U87-MG tumour cells by immunohistochemistry and radioligand binding studies.

The strong immunofluorescence signal determined in tumour cells in cryosections obtained from mouse brain at day 27 after intracerebral transplantation of U87-MG cells indicates the persistently high expression of sig1R in the orthotopic GBM (Figure 1C). Besides, the fluorescent staining is informative in that not all cells in the field of view express sig1R, indicating a heterogeneous expression profile within the tumour bulk (Figure 1C). Thus, the conservation of sig1R expression irrespective of the environment (culture medium or brain dynamic environment) was confirmed (Figure 1).

Complementary autoradiography performed with the sig1R-specific PET tracer (S)-(−)-[18F]fluspidine confirmed the expression of sig1R in orthotopically growing U87-MG cells. The autoradiographic images presented in Figure 2 indicate a high density of binding sites in the tumour region (Figure 2B). The radioactive signal is nearly completely abolished by co-administration of the sig1R-specific ligand SA4503, a selective agonist commonly used as competitive agent (IC50 = 17.4 nM [41]) (Figure 2C). Interestingly, the macroscopic distribution pattern of (S)-(−)-[18F]fluspidine in the orthotopic tumour reflects a heterogeneous accumulation of activity within the tumour, similar to what was observed by immunohistochemistry on cellular level. By saturation studies, performed as homogenous radioligand displacement experiments by co-incubation of (S)-(−)-[18F]fluspidine with different concentrations of (S)-(−)-fluspidine, we determined the kinetic binding parameters of (S)-(−)-[18F]fluspidine in the tumour (T) and an internal reference region, the contralateral striatum (CL). In both regions, (S)-(−)-[18F]fluspidine bound specifically and with comparable affinities of *K*D, <sup>T</sup> = 17.5 ± 1.3 nM and *K*D, CL = 17.0 ± 4.8 nM. However, the sig1R density was ~1.7 times higher in the tumour area compared to the CL area, as reflected by values of *B*max, T = 704 ± 16 fmol/mg protein vs. *B*max, CL = 414 ± 36 fmol/mg protein.

**Figure 2.** In vitro autoradiography of the mouse brain bearing an orthotopic U87-MG xenograft. Representative autoradiographic images of the coronal plane of mouse brain slices: (**A**) Hematoxylin-eosin staining; (**B**) in vitro distribution of activity after incubation with 0.1 MBq/mL (S)-(−)-[18F]fluspidine, (**C**) co-incubation with 10 <sup>μ</sup>M SA4503 to determine the nonspecific binding and (**D**) with 10 nM of (S)-(−)-fluspidine as competing agent. Cx: cortex; CL: contralateral striatum; Th: thalamus; Hy: hypothalamus; T: tumour. Width of a mouse brain ~1 cm.
