Tracking brain inflammation and oxidative stress over time in a mouse model of Alzheimer’s disease

Alzheimer disease (AD) is the most common cause of dementia and affects over 13 million people worldwide. This number is expected to increase to more than 100 million by 2050 [1]. There is no effective treatment for AD available on the market and, more worryingly, no reliable option for the early diagnosis of the disease. The recent approval of aducanumab by the FDA (Aduhelm) [2] has been an important milestone in disease intervention. However, the treatment was not as effective as it was initially expected. One of the responsible factors for this was hypothesized to be the disease stage of the patients enrolled in the clinical trials [3], and a review of the study suggests a lack of correlation between surrogate imaging biomarkers and clinical outcome [4]. These facts put a spotlight on both early diagnosis and the development of surrogate markers capable of evaluating and predicting disease progression.

Pathophysiologically, AD is characterized by the accumulation of amyloid-β (Aβ) aggregates in various conformations [5], the appearance of filamentous intraneuronal inclusions made of hyperphosphorylated Tau protein (p-Tau) [6], and synaptic dysfunction [7]. Some of these processes begin decades before the onset of clinical symptoms, which is why the most recent AT(N) scheme, published by the National Institute on Aging and Alzheimer’s Association, proposed establishing a definition of AD based on the assessment of disease biomarkers, namely Aβ plaque (A), fibrillary tau (T), and neurodegeneration or neuronal damage (N), and not on clinical symptoms [8]. Apart from characteristic pathology, AD patients also exhibit altered glucose metabolism [9], neuroinflammation [10], and oxidative stress [11]. In fact, oxidative stress and neuroinflammation were suggested to play the key role in the pathophysiology of neurodegeneration and promoted cognitive decline of dementia patients [12]. Furthermore, studies show that microglia and astroglia cells were activated in 5xFAD mice at the onset of Aβ plaque formation [13]. The role of oxidative stress and neuroinflammation in the onset of AD and its progression, and their association with Aβ plaques [14, 15], places them as a potential source of (imaging) biomarkers for AD.

In vivo, neuroinflammation and oxidative stress can be detected indirectly by assessing overexpression of the translocator protein (18 kDa) (TSPO) and the cystine/glutamate antiporter (x_c^–) system, respectively [16, 17]. A large pool of literature data indicates that TSPO is a promising target to monitor glial cell and infiltrated macrophage activation during the inflammation process [18–20]. Additionally, autoradiography using [³H]PBR28 (TSPO-specific radioligand) showed an increase in specific binding in 5xFAD compared to WT mice, confirming a strong relationship between neuroinflammation and upregulation of TSPO [21, 22]. On the other hand, overexpression of the x_c^– transporter leads to increased levels of antioxidant glutathione through an increased transport of cystine [23]. This makes this antiporter system one of the possible targets for assessment of oxidative stress by PET imaging, using radioactively labeled x_c^– substrates, such as glutamate analogue, (4S)-4-(3-[¹⁸F]fluoropropyl)-L-glutamate ([¹⁸F]FSPG) [24]. Although there has been some dispute over the ability of [¹⁸F]FSPG to cross intact blood-brain barrier (BBB), some studies indicate that limited transport into the brain could be possible [25, 26], and it has been successfully applied to investigate oxidative stress in cancer [27] and different neurological disorders, such as cerebral ischemia [20] and multiple sclerosis [28].

In this work, we report the longitudinal evaluation of [¹⁸F]DPA-714 and [¹⁸F]FSPG to assess the levels of neuroinflammation and oxidative stress, respectively, in the 5xFAD mouse model for AD by PET imaging in vivo. The results have been correlated to Aβ plaque burden, as determined by PET imaging using the validated radiotracer [¹⁸F]florbetaben and previously reported by our research group [29].

TSPO protein overexpression and cysteine/glutamate antiporter gene are known to be associated with AD, due to their involvement in neuroinflammation and formation of reactive oxygen and nitrogen species (RONS). For the purposes of this study, 5xFAD mice were chosen as they faithfully recapitulate AD-associated Aβ pathology, and the uptake of two radiotracers, [¹⁸F]DPA-714 and [¹⁸F]FSPG, was used as a surrogate for neuroinflammation (TSPO) and oxidative stress levels (system x_c^–), respectively. The results were compared to longitudinal PET imaging results with [¹⁸F]florbetaben obtained in the same mice as recently reported [29].

The use of a specific TSPO marker, [¹⁸F]DPA-714, was ruled appropriate to assess neuroinflammation in mice. IHC analysis showed low expression of TSPO in CB of 5xFAD mice (Figs. 4 and 5), suggesting that this brain region is an appropriate reference for the assessment of neuroinflammation through [¹⁸F]DPA-714-PET imaging. This is consistent with a previous study that selected CB as a reference region for [¹⁸F]DPA-714-PET imaging of a different animal model of AD [33]. Contrary to this previous study, our TACs in the whole brain did not show the presence of a plateau within the duration of the PET scans. This posed some doubts on the possibility of quantifying the results by determining the uptake ratio between the region of interest and CB. The results were therefore analyzed by Regional Logan Plot analysis using CB as the reference region and reported as DVRs. Contrary to the original Logan Plot analysis, Regional Logan Plot analysis does not require plasma sampling and allows the estimation of DVR (or non-displaceable binding potential, BP_ND; calculated as BP_ND = DVR − 1) from reversible ligand-receptor PET studies.

Regarding the x_c^– antiporter system, our animal model exhibited low but sufficient [¹⁸F]FSPG uptake into the brain. Contrarily to the case of [¹⁸F]DPA-714, no evident reference region could be selected for [¹⁸F]FSPG. The option of using the blood input function for kinetic modeling was also unfeasible, because the heart and major vessels, which have been previously used to determine image-derived input functions [34], were out of the FOV of the PET camera. The quantification was therefore based on averaging images obtained at the last three frames, to determine uptake values in each region, as described before [20]. The lack of an appropriate reference region disabled correction for deviations in the injected activity. Furthermore, a source of radioactivity outside of the brain was observed, likely due to uptake by immune cells in the meninges/subarachnoid space or a meningeal lymphatic vessel, as previously suggested [28]. This could result in radioactive signal spill-over into CB. All these factors reflected in higher intrasubject variability, which ultimately resulted in high standard deviation values. Despite these limitations, some conclusions could be made based on the obtained results.

Control mice exhibited a small but insignificant increase in [¹⁸F]DPA-714 and [¹⁸F]FSPG uptake in the brain over time up to 8 months, which indicated that TSPO and RONS did not evolve as a result of aging in these animals. In contrast, the longitudinal uptake of both radiotracers in the brain of diseased mice indicated that neuroinflammation and oxidative stress followed the trend observed for Aβ deposition, as indicated in our previous [¹⁸F]florbetaben-PET study [29]. Accordingly, the initial imaging session showed only slightly, non-significantly higher uptake of [¹⁸F]DPA-714 and [¹⁸F]FSPG in 5xFAD compared to WT mice. The difference increased as the disease progressed, at 5 and 8 months of age. Although both radiotracers showed increased uptake in CTX, HIP, and THA of 5xFAD compared to WT (brain regions which show increased beta-amyloid plaque burden at age > 3 months in this animal model [32]), the differences were notably higher for [¹⁸F]DPA-714. These findings were confirmed by qualitative immunofluorescence analysis. The brains of 8-month-old mice from this study were initially chosen, because no significant changes in radiotracer uptake occurred at later time points of the PET study. Additionally, PET image analysis could not account for changes in tissue uniformity and shrinkage, which would disable comparison between in vivo and ex vivo techniques. Ex vivo analysis showed that TSPO and xCT concentration increased in CTX, HIP, and THA of 5xFAD mice in comparison to age-matched controls, without notable differences among these brain regions. Furthermore, highly myelinated brain regions exhibited higher concentrations of microglia/infiltrated macrophages overexpressing TSPO, which could be the reason behind the notable increase of TSPO overexpression in THA observed in PET. On the other hand, xCT overexpression was observed in non-glial cells throughout the brain, suggesting that the increase of neuronal oxidative stress is associated with the overall increase of Aβ rather than with regional pathology. This also explained the absence of the reference region within the brain for PET image analysis. Additionally, higher oxidative stress levels induced lipid peroxidation and the subsequent accumulation of the 4 HNE in AD brains compared with WT.

[¹⁸F]DPA-714 uptake levels in 5xFAD mice continued to slightly increase in HIP and THA after the 8-month time point, but the uptake in CTX between 8 and 12 months remained constant. This is surprising, because more neuroinflammation is expected as the disease progresses. While this could be a specific characteristic of the disease or this animal model, a contribution of other factors, such as brain tissue shrinkage in some brain regions, might also be the cause for this phenomenon. Indeed, MRI analysis at 12 months of age showed cortical tissue shrinkage, a common feature in late-stage AD. Furthermore, DTI revealed extensive damage on brain microstructure, likely indicating neuronal damage. Of note, the inaccessibility of the PET-MRI camera prevented correct delineation of VOIs; therefore, the same brain atlas was used for all the mice, regardless of age. Although the brain size was corrected by MRI-CT co-registration, the brain tissue shrinkage could not be accounted for and is a possible reason behind these discrepancies.

In the figure, relative average values (SUVr to the cerebellum for [¹⁸F]florbetaben, obtained from [29]; SUV for [¹⁸F]FSPG; DVR for [¹⁸F]DPA-714) obtained both in CTX and HIP for each group and age are plotted. In both brain subregions, the profiles corresponding to [¹⁸F]florbetaben and [¹⁸F]DPA-714 follow the same trend, with a progressive increase over time that tends to stabilize after 8 months. Contrarily, the relative increased uptake for [¹⁸F]FSPG in 5xFAD animals seems to show an earlier, more abrupt onset between 2 and 4 months of age, to stabilize afterwards. Interestingly, relative values for this radiotracer drop to ca. 1 at the age of 11 months, which means that at this age, both WT and AD animals show similar tracer uptake. Noteworthy, this abrupt decrease is not due to a decrease in the uptake observed in 5xFAD animals, but to an increase in [¹⁸F]FSPG uptake in aged WT animals. The causes behind such an increase remain unclear and require further investigation. Still, the different time profile obtained for this tracer suggests that it could complement neuroinflammation and β-amyloid imaging biomarkers for more accurate longitudinal investigation of the pathophysiology of the disease and monitoring of AD progression.

In conclusion, [¹⁸F]DPA-714 and [¹⁸F]FSPG show prospect to monitor AD progression. TSPO overexpression and increased oxidative stress accompany Aβ accumulation, thus becoming potential in vivo diagnostic/prognostic tools for Aβ-associated neurodegeneration. The results support the involvement of neuroinflammatory processes and oxidative stress in Aβ pathology, adding to the pool of knowledge on AD disease mechanism and encouraging the expansion of AD diagnosis beyond the traditional AT(N) scheme.