Synaptic density and neuronal metabolic function measured by PET in the unilateral 6-OHDA rat model of Parkinson’s disease

Parkinson’s disease (PD) is caused by progressive neurodegeneration and characterised by motor dysfunction. Neurodegeneration of dopaminergic neurons also causes aberrations within the cortico-striato-thalamo-cortical (CSTC) circuit, which has been hypothesised to lead to non-motor symptoms such as depression. Individuals with PD have both lower synaptic density and changes in neuronal metabolic function in the basal ganglia, as measured using [11C]UCB-J and [18F]FDG positron emission tomography (PET), respectively. However, the two radioligands have not been directly compared in the same PD subject or in neurodegeneration animal models. Here, we investigate [11C]UCB-J binding and [18F]FDG uptake in the CSTC circuit following a unilateral dopaminergic lesion in rats and compare it to sham lesioned rats. Rats received either a unilateral injection of 6-hydroxydopamine (6-OHDA) or saline in the medial forebrain bundle and rostral substantia nigra (n=4/group). After three weeks, all rats underwent two PET scans using [18F]FDG, followed by [11C]UCB-J on a separate day. [18F]FDG uptake and [11C]UCB-J binding were both lower in the ipsilateral striatal regions compared to the contralateral regions. Using [11C]UCB-J, we could detect an 8.7% decrease in the ipsilateral ventral midbrain, compared to a 2.9% decrease in ventral midbrain using [18F]FDG. Differential changes between hemispheres for [11C]UCB-J and [18F]FDG outcomes were also evident in the CSTC circuit’s cortical regions, especially in the orbitofrontal cortex and medial prefrontal cortex where higher synaptic density yet lower neuronal metabolic function was observed, following lesioning. In conclusion, [11C]UCB-J and [18F]FDG PET can detect divergent changes following a dopaminergic lesion in rats, especially in cortical regions that are not directly affected by the neurotoxin. These results suggest that combined [11C]UCB-J and [18F]FDG scans could yield a better picture of the heterogeneous cerebral changes in neurodegenerative disorders.


1
Introduction 53 Several techniques have been developed to identify disease-related neuronal patterns to aid early 54 detection and differential diagnoses of Parkinson's disease (PD). Examples of such methods are 55 positron emission tomography (PET) imaging to measure glucose metabolism (Loane and Politis,56 2011), dopamine synthesis, transporters, or receptors (Kerstens and Varrone, 2020). In PD, one 57 affected neuronal circuit is the cortico-striato-thalamo-cortical (CSTC) circuit (Vriend et al., 2014). 58 The CSTC circuit connects the cortex with the basal ganglia to control and coordinate goal-directed 59 behaviour. This circuit can be further divided into three loops: the motor, limbic, and associative 60 circuits (Groenewegen and Trimble, 2007; Vriend et al., 2014). The dopamine system innervates the 61 striatal regions of the CSTC circuits and is critical in modulating their output. A model of 6-62 hydroxydopamine (6-OHDA) induced dopaminergic lesion leads to modulation within the CSTC, 63 which will further help understand this circuit (Schwarting and Huston, 1996). 64 [ 11 C]UCB-J is a PET radioligand showing high affinity to synaptic vesicle glycoprotein 2A (SV2A) 65 . SV2A is ubiquitously expressed throughout the brain (Bajjalieh et al., 1994;66 Südhof, 2004) and is a suitable proxy for synaptic density (Finnema et al., 2016). Accordingly, 67 [  changes after a unilateral dopaminergic lesioning of the rat brain. As a control to assess differential 87 changes, we used both the contralateral hemisphere and compared the 6-OHDA model to a group of 88 sham-lesioned rats. Several studies have successfully detected changes in regional [ 18 F]FDG uptake 89 after a 6-OHDA lesion in both rats (Casteels et  The animals were acclimatised in the surgery room for at least 1 hour. Analgesia was provided with 106 carprofen (Rimadyl, Zoetis, NJ, USA) 5 mg/kg, subcutaneous (SC), 45 minutes before the surgery 107 and 24 hours and 48 hours postoperative. Before commencing the surgery, animals received 108 desmethylimipramine (25 mg/kg, intraperitoneal (IP)) mixed in physiological saline. 109 Desmethylimipramine protects the noradrenergic neurons from the neurotoxic effects (Esteban et al., 110 1999). Anaesthesia was induced with 3% isoflurane in oxygen and maintained through surgery with 111 1.2-1.8% isoflurane in oxygen. The rats were fixed on a stereotaxic apparatus (Kopf Instruments,  112 Tujunga, CA, USA) with the incisor bar set 3.3 mm below the level of the ear bars. An incision was 113 made on the scalp, and two bur-holes were drilled on one side of the skull using a dental micromotor 114 and round bur (0.5 mm). A 2 µg/µL solution of 6-OHDA (2,5-Dihydroxytyramine hydrobromide, 115 Sigma-Aldrich, Søborg, Denmark) in physiological saline containing 0.02% ascorbic acid or 116 physiological saline (containing 0.02% ascorbic acid) was drawn into a 10 µL syringe with a 33 g 117 needle (World Precision Instruments, Sarasota, FL, USA). 3 µL were infused into the medial  118 forebrain bundle (coordinates: AP= 4.8 mm, ML= 1.7 mm, DV= 8 mm) and 3 µL infused rostral to 119 substantia nigra (coordinates: AP= 3.6 mm, ML= 2 mm, DV= 8.3 mm) relative to the bregma to 120 ensure unilateral dopaminergic degeneration. The infusion was delivered at 151 nL/minutes driven by 121 an infusion pump (World Precision Instruments, Sarasota, FL, USA), followed by a 7 minute pause 122 prior to a slow withdrawal of the syringe needle. The incision was sutured back. After recovery from 123 anaesthesia, rats were returned to the recovery cage and housed alone for 48 hours and then housed in 124 pairs for recovery of 21 days to allow the development of the lesions. 125

Study design and confirmation of lesion: 126
Four rats were injected unilaterally with 6-OHDA, while another four were injected with 127 physiological saline and divided into two groups, i.e., dopamine lesioned and sham lesioned; Figure 1  128 shows the study's overall design. After the recovery period, the rats were subjected to two PET scans 129 [ least 2 hours before the scan. Anaesthesia was induced using 3% isoflurane in oxygen. All rats were 158 placed in a 2 x 2 custom made rat holder (illustration in Figure 1) Schiffer's atlas (Schiffer et al., 2006) were applied to the PET image in standard space. The regions 196 (depicted in Figure 3 and Supplementary Figure 4) For [ 11 C]UCB-J, time-activity curves (TACs) for all VOIs were extracted from the PET images. 210 Estimates for the total blood activity was acquired using a non-invasive image-derived input function 211 (IDIF) that was used for estimating a surrogate of V T. V T was determined in each VOI, using the 212 one-tissue compartment model (1TCM), which has previously been validated for 'seed' at the point of highest activity in the heart and producing a VOI which is about the size of the 217 rat's left ventricle (5-6 voxels). In order to fit the 1TCM to the TACs, the blood volume fraction (V B ) 218 was fixed at 5%. In addition to V T , the micro-parameters K 1 and k 2 were also extracted from the 219 kinetic modelling. These micro-parameters were checked for the difference due to the surgical 220 procedure or any other reason. 1TC model fit to a representative region, ipsilateral and contralateral 221 striatum, are shown in Supplementary Figure 5. All micro parameters (K1 and k2) for all regions are 222 recorded in Supplementary Table 2. In addition to kinetic modelling, TACs were converted into 223 SUVs. Ipsilateral and contralateral striatum and ventral midbrain (sham and dopamine lesioned) 224 TACs were averaged for visual representation. This was performed using GraphPad Prism 9 225 (GraphPad Software, San Diego, CA, USA). 226

Statistics: 227
Due to the limited sample size and the number of comparisons undertaken, the study is exploratory in 228 nature, meaning that caution should be taken around drawing strong confirmatory conclusions from 229 the data. As such, all p-values reported should be considered as a continuous assessment of indirect 230 evidence against the null hypothesis of no difference between groups or hemispheres, and binary 231 conclusions of "significant" or "not significant" within the Neyman-Pearson Null-hypothesis-232 significance-testing framework should be avoided. 233 The data were analysed using Jamovi (Version 1.6, The jamovi project (2021)  Results 253

Representative [ 11 C]UCB-J and [ 18 F]FDG PET images 260
Representative rats, compared to the contralateral side ( Figure 5 and Table 1). No substantial differences were found 283 between the ipsilateral and contralateral sides within the sham lesioned animals. 284

Changes in cortical regions between [ 11 C]UCB-J binding and [ 18 F]FDG uptake 304
A post hoc analysis of changes in the cortical regions and thalamus between the lesion and sham 305 group (Figure 7) showed an increase in This study explored regional differences in [ 11 C]UCB-J binding and [ 18 F]FDG uptake using a 312 unilateral 6-OHDA dopaminergic lesion in rats, a commonly used animal model for PD. We 313 observed differences in SV2A density and neuronal metabolic function between ipsilateral and 314 contralateral hemispheres, especially the basal ganglia, which are well known to be innervated by 315 dopaminergic terminals. This suggests a decline in dopaminergic neurons and synapses due to the 6-316 OHDA lesion, consistent with TH immunostaining (Figure 2 previously only been reported in the prefrontal cortex (Casteels et al., 2008), while other studies 343 show unaltered metabolism (Kurachi et al., 1995). The decrease in orbitofrontal and medial 344 prefrontal cortical metabolic function may be due to the disrupted dopaminergic innervation from the 345 substantia nigra to the orbitofrontal cortex (Murphy and Deutch, 2018). 346 The 1TCM and 2TCM both work favourably with [ 11 C]UCB-J using the heart as an IDIF (Bertoglio 377 et al., 2020; Glorie et al., 2020). The use of IDIF and whole-brain normalisation allows longitudinal 378 studies in rodents since blood sampling often is laborious and error-prone. Although most of these 379 studies are using mice, we assume it translates well to rats. 380 The small sample size is a limitation of our study, making it particularly hard to conclude that there 381 are no differences (type 2 error). For that reason, we took an exploratory approach without pre-382 registered predictions and without corrections for multiple testing. As such, the results should be seen 383 as preliminary, and we caution against confirmatory conclusions from the results and encourage 384 future replications using larger samples and a more limited selection of analyses. Further, the 385 contralateral hemisphere may not be an ideal control region because of the inter-hemisphere 386 anatomical connection of the basal ganglia through the pedunculopontine nucleus (Breit et al., 2008 resolution PET scanner allows for simultaneous scanning of up to four rats, which further allowed us 395 to perform four [ 11 C]scans with a single radiosynthesis. Although this saves resources and enables a 396 more direct comparison between rats, the resolution of the HRRT is lower than other available 397 single-subject small animal micro-PET systems. Hence, our ability to identify potentially apparent 398 biological differences in small regions is limited due to, e.g., partial volume effects. 399 Regardless, we found a pattern in the regional cortical synaptic density and neuronal metabolic 400 function, which could be clinically relevant, especially changes within the anterior cingulate cortex 401 and orbitofrontal cortex. We see