Validation of [18F]LW223 for TSPO PET imaging in mice and assessment of cellular TSPO expression during neuroinflammation

Abstract

Neuroinflammation is associated with a number of brain diseases, including dementias, stroke, multiple sclerosis, Parkinson’s disease, depression and many more, making it a common feature of cerebral pathology. Therefore, the use of imaging techniques that are capable of detecting neuroinflammation, such as Positron Emission Tomography (PET), become valuable and versatile tools for diagnosis and monitoring of progression of these diseases. Among the best-known biomarkers for neuroinflammation in PET research is the 18kDa translocator protein (TSPO). During neuroinflammation, TSPO expression in the brain is typically upregulated, and the increase in TSPO can be detected using TSPO-specific PET radiotracers. This PhD project focused on the validation of a novel TSPO PET radiotracer, [18F]LW223, in mice and its application to image sex-dependent TSPO expression in the healthy brain, as well as in neuroinflammation resulting from bacterial endotoxin-induced systemic inflammation (lipopolysaccharide (LPS) challenge) and vascular dementia (bilateral carotid artery stenosis, or BCAS). In addition, this project aimed to investigate cellular TSPO distribution in health and under inflammatory conditions. It was hypothesised, that 1) the binding of [¹⁸F]LW223 in vivo mirrors TSPO distribution in the mouse brain, while having low non-displaceable binding, high target occupancy and affinity for the target; 2) regional [¹⁸F]LW223 binding in the murine brain differs between healthy males and females; 3) TSPO expression in the brain of mice post-systemic LPS challenge and with BCAS can be detected using [¹⁸F]LW223; and 4) in healthy brain, TSPO is mostly expressed by endothelial cells, while under neuroinflammatory conditions, its expression shifts primarily to microglia.

The validation of [18F]LW223 PET imaging for preclinical use was performed in healthy young C57Bl/6J mice. Mice co-injected with [18F]LW223 and varying concentrations of unlabelled LW223 were used to estimate volume of non-displaceable binding, target occupancy and affinity of this PET radiotracer. In addition to dynamic PET imaging over 2 hours, arterial input functions (AIFs) were generated using an automated blood sampling system. In order to determine parent fraction (i.e. unmetabolised fraction) of the radiotracer in mouse blood and organs, a separate cohort of mice were injected with [18F]LW223, the tissues were collected at 2, 5, 10, 30, 60 and 120 minutes post-injection, processed and analysed using radio-high performance liquid chromatography (HPLC). To find the most suitable quantification method for [18F]LW223 PET data, multiple kinetic modelling methods with AIF, image-derived input functions (IDIFs) and population-derived input functions with metabolite correction were investigated, as well as simplified quantification methods, such as standard uptake values (SUV) and apparent volume of distribution (VTapp).

In order to assess differences between male and female TSPO expression, male and female C57Bl/6J mice were subjected to 2 hours of dynamic PET imaging immediately post-[18F]LW223 injection. Volume of interest-based SUV analysis and voxel-based parametric mapping were used to determine sex-dependent TSPO expression in mouse brain. The in vivo results were compared with in vitro [18F]LW223 autoradiography and TSPO immunohistochemistry. Additionally, the distribution of TSPO in glial and endothelial cells of male and female mouse brain was assessed by double immunofluorescence staining using antibodies against TSPO and astrocytes (GFAP), microglia/macrophages (Iba1), oligodendrocytes (Olig2) and endothelial cells (CD31).

After validation of [18F]LW223 in healthy animals, the radiotracer was used to image neuroinflammation resulting from acute (LPS) and chronic (BCAS) harmful stimuli. In the LPS model, mice were injected (i.p.) with vehicle, 0.5 mg/kg or 1 mg/kg of LPS from E. coli, and 24 hours later underwent [18F]LW223 PET imaging. In the LPS study, brain tissue was harvested following PET imaging session and divided into two hemispheres – one was lysed and used to measure levels of IL-1β, IL-6, TNF-α and TSPO using ELISA kits, and another was cryosectioned for in vitro [18F]LW223 autoradiography and ex vivo validation of PET results, as well as for double immunofluorescence staining with antibodies against TSPO and GFAP, Iba1, Olig2 and CD31 to assess cellular dynamics of TSPO expression under inflammatory conditions. The BCAS mixed coil model was generated by Prof Horsburgh’s group. Immediately post BCAS surgery, mice were placed on either control diet or diet containing colony stimulating factor 1 receptor (CSF1R) inhibitor GW2580 for 5 weeks, followed by PET imaging with [18F]LW223. Additionally, in order to explore translational potential of [18F]LW223, in vitro autoradiography was performed using human brain sections of healthy controls and small vessel disease (SVD) patients.

The validation study showed that binding of [18F]LW223 matched the known distribution of TSPO in the mouse body, had low volume non-displaceable binding in the brain (VND) (<10% of total binding), high target occupancy (~80-90%) and high affinity for TSPO (~2 nM). The radiometabolite studies showed high parent fraction of [18F]LW223 in mouse blood and organs up to 2 hours post radiotracer injection (~80-95%). It was also concluded that the 2-tissue compartment model with AIF provided the best fit for [18F]LW223 PET data, although its correlation with simplified outcome measures, such SUV90-120min or VTapp, allowed for just as accurate data quantification in healthy animals, but using less complicated experimental and analytical approaches.

Male and female PET data analysis revealed a significantly higher [18F]LW223 uptake in several male brain regions and heart tissue when compared to females. Autoradiography data positively correlated with both in vivo data and TSPO immunohistochemistry results, but failed to detect any sex-dependent regional differences. Cellular TSPO distribution determined by immunohistochemistry revealed that the highest proportion of TSPO was expressed within endothelial cells and highlighted a significantly higher astrocytic TSPO expression in male hippocampus.

The data obtained using mice post-systemic 24 hour LPS challenge showed upregulation of pro-inflammatory cytokines and increase in K1, VT and VTapp values. However, ex vivo validation techniques did not detect upregulation in TSPO under these experimental conditions, which correlated with k3/k4 values. In chronic neuroinflammatory model of BCAS, VT, k3/k4 and VTapp of [18F]LW223 were significantly higher, whereas treatment with GW2580 reduced these values to the levels similar to control animals. This suggested that [18F]LW223 could be applied to monitor efficacy of anti-inflammatory treatment strategies in the brain. Moreover, using human brain samples it was demonstrated that [18F]LW223, but not [3H]PK11195, was capable of detecting increased TSPO expression in small vessel disease in a sex-dependent manner, encouraging to purse translation of this novel TSPO PET radiotracer.

In summary, [18F]LW223 has the required properties to become a successful PET radiotracer. It has low VND, as well as high affinity, target occupancy and parent fraction allowing for simplified, but accurate, quantification of TSPO expression in mice. This allowed for detection of subtle differences in target expression, hence enabling to study sex-dependent TSPO expression in the brain in vivo. Data presented here shows that [18F]LW223 PET can assist with detection of acute and chronic neuroinflammation processes in vivo and non-invasively as well as support development of anti-inflammatory therapies. Cellular analysis of TSPO show significant differences in cell type and brain regions in health and during disease. To conclude, this PhD project presents original data on a promising TSPO PET radiotracer with translational potential and provides new insights into TSPO biology.