Altered Blood and Brain Expression of Inflammation and Redox Genes in Alzheimer’s Disease, Common to APPV717I × TAUP301L Mice and Patients

Despite intensive research, the pathophysiology of Alzheimer’s disease (AD) is still not fully understood, and currently there are no effective treatments. Therefore, there is an unmet need for reliable biomarkers and animal models of AD to develop innovative therapeutic strategies addressing early pathologic events such as neuroinflammation and redox disturbances. The study aims to identify inflammatory and redox dysregulations in the context of AD-specific neuronal cell death and DNA damage, using the APPV717I× TAUP301L (AT) mouse model of AD. The expression of 84 inflammatory and 84 redox genes in the hippocampus and peripheral blood of double transgenic AT mice was evaluated against age-matched controls. A distinctive gene expression profile in the hippocampus and the blood of AT mice was identified, addressing DNA damage, apoptosis and thrombosis, complemented by inflammatory factors and receptors, along with ROS producers and antioxidants. Gene expression dysregulations that are common to AT mice and AD patients guided the final selection of candidate biomarkers. The identified inflammation and redox genes, common to AD patients and AT mice, might be valuable candidate biomarkers for preclinical drug development that could be readily translated to clinical trials.


Introduction
Alzheimer's disease (AD) is a chronic and progressive neurodegenerative disorder, characterized by broad neurologic, cognitive, functional, behavioral and psychological impairments [1,2], that is considered the greatest challenge for health and social care in the 21st century, having a global impact on patients and their families, communities and society [3].
Genetic factors underlie AD pathophysiology, with gene variants of APP, PSEN1 and PSEN2 leading to early AD onset (EOAD) in familial forms of the disease, and the APOE ε4 allele being associated with increased disease severity in EOAD as well as in late-onset forms (LOAD) of familial and sporadic AD [4,5]. While genetics provides a solid background for disease onset and evolution, the dysregulation of gene circuits and Int. J. Mol. Sci. 2022, 23, 5799 2 of 18 functional impairment of critical proteins are intensively investigated as well to obtain new insights into AD's pathophysiology [6,7].
Nevertheless, despite decades of intensive research in the field, AD's pathophysiology is still not fully understood, and currently there are no effective treatments. For instance, clinical trials on drugs addressing amyloid plaques and TAU tangles, the typical pathologic features of AD, have failed to slowdown cognitive decline in patients with mild cognitive impairment or AD dementia [8][9][10]. Therefore, AD research has currently shifted to alternative pathological processes that may initiate and further sustain neuronal cell death and the accumulation of DNA damage in the hippocampus [11][12][13]. Evidence is continuously being produced that systemic and local inflammation and redox alterations precede overt AD symptoms by years, and may be considered at least risk factors, if not straightforward mechanisms in AD pathogenesis [14]. Thus, neuroinflammation mediated by resident cells (microglia and astrocytes) as well as by capillary endothelial cells and brain-infiltrating leukocytes, in conjunction with chronic alterations of redox-sensitive signaling pathways [15,16], were shown to greatly contribute to synaptic dysfunction, neuronal death and the inhibition of neurogenesis [17,18]. Upon DNA damage, microglia and astrocytes accumulate DNA fragments "leaking" to the cytoplasm where they trigger STING-dependent sterile inflammatory processes and neurotoxicity [19]. Moreover, oxidative DNA damage [20,21] in the AD brain, tightly connected with Aβ and TAU pathologies [12], seem to occur prior to disease onset [22,23].
A case-control clinical study recently published by us [24] evidenced in whole blood nine inflammation and seven redox genes that discriminate very well between mild AD patients and controls, resulting from the increased activity of NFκB, NRF2 and several zinc finger and helix-loop-helix transcription factors. Altogether, the results emphasize systemic dysfunctions that might be an echo of the pathological events occurring in the AD brain.
In addition to new inflammation and redox biomarkers, appropriate animal models of AD that recapitulate the combined human amyloid and TAU pathology and express inflammation-and redox-related biomarkers, are highly needed for the preclinical development of anti-inflammatory and antioxidant therapeutic strategies in AD [25].
In this context, we carried out a case-control study investigating through qRT-PCR the expression levels of 168 inflammation and redox genes in the hippocampus and peripheral whole blood of double transgenic (hAPP V717I and hTAU P301L ) mice against age-matched controls. These mice, which express APP and TAU variants of human genes in the brain [26], have been shown to recapitulate with aging the combined amyloid and TAU pathology of the human disease [26,27]. The study highlighted inflammation and redox disturbances in the context of neuronal cell death and DNA damage, and allowed the selection of particular genes that are commonly dysregulated in AD patients and the investigated mouse model. This gene panel may be further used for the preclinical development of innovative anti-inflammatory and antioxidant therapies in AD.

Results
The expression levels of 168 inflammation and redox genes were analyzed by qRT-PCR using pathway-focused arrays in the hippocampus and whole blood from AT mice developing amyloidopathy (APP V717I ) and tauopathy (TAU P301L ), in comparison with agematched WT controls. Gene expression levels were calculated as 2 −∆CT values that were further processed as fold change (FC) values. Results were presented as fold regulation (FR) values, which are equal to the FC value if the FC is higher than 1 (resulting in positive superunit values for over-expressed genes), and is calculated as the negative inverse of the FC when the FC is below 1 (resulting negative superunit values for under-expressed genes).

Gene Expression Changes in the Hippocampus of AT Mice
We analyzed first those genes that were differentially expressed with the FR threshold set at 1.5 (p < 0.05). Thirteen inflammatory genes were over-expressed in the hippocam-pus of AT mice as compared with WT controls, while four genes were down-regulated (Tnfaip3, Erg1, Csf2 and Zap70), as presented in Figure 1A. Some of these genes underlie basic pathological processes in the AD brain, such as apoptosis (Casp8, Tnfsf10, Birc3 and Tnfaip3) and DNA damage responses (Atr, Ercc6 and Rag2). Thrombotic processes were also emphasized at gene expression level (F2r and Ptgs1) in the hippocampus of AT mice. Other genes identified in this study address the immune response by encoding cytokines (Il1b, Il1a, Il19 and Lta), chemokines (Ccl5) and growth factors (Erg1 and Csf2), as well as inflammation-triggering receptors (Tlr6 and Cd40) or signaling adaptors (Zap70). Most of these genes are involved in NFkB signaling or are NFkB target genes (marked with * in Figure 1A), demonstrating that NFkB-mediated inflammation has a broad fingerprint in the hippocampus of AT mice.

Gene Expression Changes in the Hippocampus of AT Mice
We analyzed first those genes that were differentially expressed with the FR threshold set at 1.5 (p < 0.05). Thirteen inflammatory genes were over-expressed in the hippocampus of AT mice as compared with WT controls, while four genes were down-regulated (Tnfaip3, Erg1, Csf2 and Zap70), as presented in Figure 1a. Some of these genes underlie basic pathological processes in the AD brain, such as apoptosis (Casp8, Tnfsf10, Birc3 and Tnfaip3) and DNA damage responses (Atr, Ercc6 and Rag2). Thrombotic processes were also emphasized at gene expression level (F2r and Ptgs1) in the hippocampus of AT mice. Other genes identified in this study address the immune response by encoding cytokines (Il1b, Il1a, Il19 and Lta), chemokines (Ccl5) and growth factors (Erg1 and Csf2), as well as inflammation-triggering receptors (Tlr6 and Cd40) or signaling adaptors (Zap70). Most of these genes are involved in NFkB signaling or are NFkB target genes (marked with * in Figure 1a), demonstrating that NFkB-mediated inflammation has a broad fingerprint in the hippocampus of AT mice.

Casp8
Tnfsf10 * Birc3 Tnfaip3   Twenty-one transcripts of redox genes were found to be up-regulated in the hippocampus of AT mice compared to WT controls ( Figure 1A), and only one gene showed decreased expression (Nos2). Some of the identified genes participate in the production of reactive oxygen species (Nox1, Noxa1, Nox4, Cyba, Idh1, Fmo2, Mpo and Mb) or reactive nitrogen species (Nos2). In turn, other genes identified in this study encode antioxidants that are involved in glutathione biosynthesis (Gss) and metabolism (Gpx1, Gpx2, Gpx3, Gpx5 and Gpx6), or in the thioredoxin system (Txn1). Several peroxidases (Mpo, Lpo, Tpo and Epx) were found to be up-regulated as well. Part of the identified genes are redox-responsive genes (marked with # in Figure 1A), with their increased transcription consequently suggesting an enhanced oxidative activity in the hippocampus of AT mice.

Gene Expression Changes in the Blood of AT Mice
A relatively different pattern of gene expression changes was registered in the blood of AT mice as compared to WT controls ( Figure 1B). We identified fifteen up-regulated and ten down-regulated genes related to the inflammatory NFkB pathway. These genes are involved in apoptosis (up-regulated Tnfrsf1a and Card10, and down-regulated Card11), DNA damage (down-regulated Eif2ak2) and thrombosis (up-regulated F2r), along with alterations in the metabolism of proteins (up-regulated Psmb5) and lipids (down-regulated Serpinb1b and Scd1). As in the hippocampus, we identified in blood the dysregulation of several genes involved in immune responses. These genes encode cytokines and their receptors (up-regulated Il1a and Il1b, and down-regulated Il1r1), and growth factors (up-regulated Csf1 and down-regulated Csf2). Moreover, we identified the increased expression of genes related to NFkB or to interfering signaling pathways, encoding receptors and ligands (up-regulated Tlr4, Tlr6, Ltbr and Cd27, and down-regulated Tlr1), ligands (up-regulated Tnfsf14), down-stream adaptor molecules(up-regulated Myd88 and down-regulated Ift172) and kinases (up-regulated Akt1), as well as transcription factors (up-regulated Rel, Atf1 and Stat1). The activation of the pro-inflammatory NFkB pathway is further sustained by the over-expression of NFkB target genes in the blood of AT mice (marked with * in Figure 1B), namely Il1b, Il1a, Csf1 and Csf2.
Five redox genes were found to be dysregulated in the blood of AT mice compared to WT controls ( Figure 1B). They are involved in superoxide production (up-regulated Ncf2) and antioxidant mechanisms (up-regulation of Gsr, accompanied by down-regulation of Gpx1, Cat and Prdx2), indicating redox disturbances in the blood of AT mice.

Analysis of Inflammation and Redox Gene Changes with Higher FR Values
Considering that genes with higher FR values are supposed to capture biologically meaningful information, we further selected those inflammation and redox genes exhibiting |FR| values > 1.8 and (p < 0.05) in the hippocampus and blood of AT mice as compared to WT controls (Figures 2 and 3, respectively). along with the NADPH-dependent dimethylaniline monooxygenase (Fmo2) and the myoglobin oxygen transporter (Mb). The transcript levels of some antioxidant genes were also found to be elevated (Figure 2g), comprising genes that encode glutathione peroxidases (Gpx1, 2, 3, 5 and 6) or thioredoxin (Txn1). In addition, three genes encoding peroxidases other than myeloperoxidase and glutathione peroxidases were found to be up-regulated as well (Figure 2h: Lpo, Tpo and Epx).
In the blood of AT mice, we identified 15 inflammation genes with |FR| values > 1.8 (p < 0.05), related to the NFkB signaling pathway (Figure 3a   Comparisons between mice groups were made using the Mann-Whitney U-test, and differences were considered significant for p < 0.05.

Figure 2.
Inflammation and redox-related genes that are differentially expressed in the hippocampus of AT mice (N = 10) vs. WT mice (N = 8), exhibiting |FR| values > 1.8. Data are presented as 2 −∆CT values, and lines represent the expression average ± standard error of the mean (SEM). Comparisons between mice groups were made using the Mann-Whitney U-test, and differences were considered significant for p < 0.05.  Comparisons between mice groups were made using the Mann-Whitney U-test, and differences were considered significant for p < 0.05.

Hippocampus-Blood Comparison
In the pool of the dysregulated genes described in Figures 2 and 3, exhibiting |FR| values > 1.8 (p < 0.05), five genes had expression changes both in the hippocampus and whole blood of AT mice as compared to WT controls ( Figure 4). Thus, the pro-inflammatory genes F2r, Tlr6 and Il1b were found to be up-regulated, while Csf2 was down-regulated. Accordingly, some inflammatory changes detected in the brain of AT mice seem to have an echo in peripheral blood. Meanwhile, the antioxidant Gpx1 gene presented opposite trends, being up-regulated in the hippocampus and down-regulated in blood, possibly due to oxidative status differences in the brain and blood of AT mice. Indeed, a distinctive and broad inflammatory NFkB fingerprint was evidenced in blood, whereas wide-ranging redox changes occurred specifically in the hippocampus of AT mice ( Figure  4), suggesting that distinctive pathological processes take place in each of these compartments.  Figure 2b). We also found 16 up-regulated redox genes fulfilling the same criteria in the hippocampus of AT mice (Figure 2f-h). Parts of these genes encode molecules involved in ROS production (Figure 2f), such as NADPH-oxidase components (Nox1, Noxa1, Nox4 and Cyba) and myeloperoxidase (Mpo), along with the NADPH-dependent dimethylaniline monooxygenase (Fmo2) and the myoglobin oxygen transporter (Mb). The transcript levels of some antioxidant genes were also found to be elevated (Figure 2g), comprising genes that encode glutathione peroxidases (Gpx1, 2, 3, 5 and 6) or thioredoxin (Txn1). In addition, three genes encoding peroxidases other than myeloperoxidase and glutathione peroxidases were found to be up-regulated as well (Figure 2h: Lpo, Tpo and Epx).
In the blood of AT mice, we identified 15 inflammation genes with |FR| values > 1.8 (p < 0.05), related to the NFkB signaling pathway (Figure 3a In addition, we found the up-regulated F2r gene is involved in thrombosis (Figure 3g), and the down-regulated Scd1 gene is related to monounsaturated fatty acid biosynthesis ( Figure 3h). As well as inflammatory genes, we found four up-regulated redox genes in the blood of AT mice that are involved in superoxide production (Figure 3i: Ncf2) or in antioxidant responses that address glutathione metabolism (Gsr and Gpx1) and catalase synthesis (Cat) (Figure 3j), with Gpx1 and Cat being involved in hydrogen peroxide detoxification.

Hippocampus-Blood Comparison
In the pool of the dysregulated genes described in Figures 2 and 3, exhibiting |FR| values > 1.8 (p < 0.05), five genes had expression changes both in the hippocampus and whole blood of AT mice as compared to WT controls ( Figure 4). Thus, the pro-inflammatory genes F2r, Tlr6 and Il1b were found to be up-regulated, while Csf2 was down-regulated. Accordingly, some inflammatory changes detected in the brain of AT mice seem to have an echo in peripheral blood. Meanwhile, the antioxidant Gpx1 gene presented opposite trends, being up-regulated in the hippocampus and down-regulated in blood, possibly due to oxidative status differences in the brain and blood of AT mice. Indeed, a distinctive and broad inflammatory NFkB fingerprint was evidenced in blood, whereas wide-ranging redox changes occurred specifically in the hippocampus of AT mice (Figure 4), suggesting that distinctive pathological processes take place in each of these compartments.

Correlation of Gene Expression Levels with Age
The analysis of inflammatory and redox gene dysregulation as a result of aging was analyzed in AT mice. A correlation study between age and the significantly dysregulated genes detected in the hippocampus (see Figure 2) and in the blood (see Figure 3) of AT mice and WT controls was performed. While no statistically relevant correlations were detected in the blood of AT mice (data not shown), eleven genes were significantly correlated with age (r > 0.800, p< 0.05) exclusively in the hippocampus of AT mice ( Figure 5). Four inflammation genes correlated with age in the AT group (Figure 5a), namely the antiinflammatory Il19 gene and the pro-inflammatory Lta gene, both encoding cytokines, along with the Atr gene involved in DNA repair and the Rag2 gene implicated in V(D)J gene recombination during B and T cell development. More correlations with age were registered in the case of hippocampal redox genes. These genes (Figure 5b) address ROS production, either superoxide (Noxa1 and Cyba) or hypochlorous acid (Mpo) generation, along with antioxidants such as glutathione peroxidases (Gpx5 and Gpx6), thioredoxin (Txn1) and thyroid peroxidase (Tpo). The lack of correlation in the control WT group in fact suggests that the associations observed in the hippocampus of AT mice reflected dis-

Correlation of Gene Expression Levels with Age
The analysis of inflammatory and redox gene dysregulation as a result of aging was analyzed in AT mice. A correlation study between age and the significantly dysregulated genes detected in the hippocampus (see Figure 2) and in the blood (see Figure 3) of AT mice and WT controls was performed. While no statistically relevant correlations were detected in the blood of AT mice (data not shown), eleven genes were significantly correlated with age (r > 0.800, p< 0.05) exclusively in the hippocampus of AT mice ( Figure 5). Four inflammation genes correlated with age in the AT group ( Figure 5A), namely the antiinflammatory Il19 gene and the pro-inflammatory Lta gene, both encoding cytokines, along with the Atr gene involved in DNA repair and the Rag2 gene implicated in V(D)J gene recombination during B and T cell development. More correlations with age were registered in the case of hippocampal redox genes. These genes ( Figure 5B) address ROS production, either superoxide (Noxa1 and Cyba) or hypochlorous acid (Mpo) generation, along with antioxidants such as glutathione peroxidases (Gpx5 and Gpx6), thioredoxin (Txn1) and thyroid peroxidase (Tpo). The lack of correlation in the control WT group in fact suggests that the associations observed in the hippocampus of AT mice reflected disease progression with age rather than biological aging. Thus, increased expression of the mentioned genes is expected to appear in more advanced forms of AD, as detected by us in mice older than 55 weeks.

Comparison of Dysregulated Inflammation and Redox Genes in AT Mice vs. AD Patients
We further compared the expression of genes identified in the hippocampus of AT mice ( Figure 2), with |FR| > 2 and p < 0.05, with available microarray datasets of postmortem brain samples from AD patients (Supplementary Table S4). As shown in Table 1, several dysregulated genes in AD patients were also altered in the AT mice model, including Il19, F2r, Tlr6, Il1b, Lta, Csf2, Gpx3 and Nox1. This panel of genes might be meaningful for preclinical investigations, using AT mice as a reliable model to monitor inflammatory and redox changes that could have a rapid translation into clinical studies.

Comparison of Dysregulated Inflammation and Redox Genes in AT Mice vs. AD Patients
We further compared the expression of genes identified in the hippocampus of AT mice ( Figure 2), with |FR| > 2 and p < 0.05, with available microarray datasets of post-mortem brain samples from AD patients (Supplementary Table S4). As shown in Table 1, several dysregulated genes in AD patients were also altered in the AT mice model, including Il19, F2r, Tlr6, Il1b, Lta, Csf2, Gpx3 and Nox1. This panel of genes might be meaningful for preclinical investigations, using AT mice as a reliable model to monitor inflammatory and redox changes that could have a rapid translation into clinical studies.
Genes with modified expression in the blood of AT mice (Figure 3), with FR > 1.8 (p < 0.05), were compared with previous data obtained by us in the blood of mild AD patients [24]. Four inflammation genes and one redox gene were found to be over-expressed in the blood of both AT mice and mild AD patients ( Table 2). The gene panel comprises inflammation genes related to the NFkB signaling pathway, encoding receptors (Ltbr and F2r), cytokines (I11b) and growth factors (Csf1), along with the antioxidant Gsr gene. The transcript levels of Gsr, identified using PCR arrays in the blood of 9 AT mice vs. 7 WT controls, with a mean age of 54.6 ± 3.5 weeks (Figure 3), as well as in the blood of mild AD patients [24], were validated in independent mice groups of 21 AT vs. 12 WT controls, with a mean age of 48.5 ± 4.9 weeks. Moreover, younger animals (7 AT vs. 8 WT controls, with a mean age of 37.1 ± 0.8 weeks), supposed to have a milder form of disease [27], were also investigated. In addition to the antioxidant Gsr gene, we analyzed in parallel the Osgin1 gene as a marker of enhanced oxidative activity, considering that only a relatively limited fingerprint of redox alterations was detected in the blood of AT mice (Figure 4).
The antioxidant Gsr gene and the redox-sensitive Osgin1 gene were both up-regulated in the group of 49-week-old mice and in the group of 37-week-old mice (Table 3). Results indicate an enhanced oxidative activity in the blood of AT mice that persisted during disease evolution and induced enhanced Gsr expression for antioxidation. Considering that Gsr is a target of the cytoprotective NRF2 transcription factor [28], it appears that NRF2 might become activated in response to an enhanced oxidative activity in the blood leukocytes of AT mice and mild AD patients. Altogether, the comparison of gene expression in the blood of AT mice and mild AD patients emphasized glutathione reductase as an early blood redox biomarker that can be evidenced even in milder forms of the disease, both in patients and AT mice. Table 3. Expression changes in selected redox genes in the blood of AT mice with various ages and of mild AD patients [24]. Results are expressed as FR values. Comparisons between diseased individuals (patients or mice) and the corresponding age-matched controls were made using the Mann-Whitney U-test, and differences were considered significant for p < 0.05.

Mice Blood Data
Human

Discussion
Clinical interventions in AD would greatly benefit from the detection of genetic blood biomarkers that would help in disease prognosis, monitoring and drug response. Accordingly, this study was aimed at identifying comparatively relevant biomarkers related to altered gene expression in the hippocampus and blood of the double transgenic AT mouse model. These mice express the mutated APP and TAU human genes in the brain, and closely recapitulate the human amyloid and TAU pathology [26], including cognitive deficits and the alteration in exploratory and anxiety-like behavior, fear learning and inflexibility in hippocampus-dependent learning [1,29].
The molecular fingerprint of inflammation and redox disturbances was characterized using pathway-focused PCR arrays in the context of disease-specific neuronal cell death and DNA damage. A broad molecular signature of NFkB-mediated inflammation and redox alterations was highlighted in the hippocampus and blood of AT mice, probably accounting for the pathologic features of AD in the investigated mouse model. As will be discussed below, several of the gene expression changes evidenced in AT mice were also detected in datasets on post-mortem brain samples from AD patients, further supporting their relevance in translational medicine.
In the hippocampus of AT mice, modified expression of genes involved in basic pathologic features of AD was registered, addressing DNA damage [30][31][32], TRAIL/caspase 8-mediated apoptosis [33] and increased risk of thrombosis [34]. Rescue mechanisms against apoptosis and DNA damage were highlighted by the over-expression of dedicated repair genes, representing an indirect proof of ongoing deleterious processes. Nevertheless, these mechanisms do not seem to be efficient enough for repairing cellular damage, as neurodegeneration was shown to evolve with age in AT mice [27].
Neuroinflammation is a central mechanism in AD, which exacerbates the amyloid and TAU pathology [14]. The transcription factor NFkB is considered a primary regulator of inflammatory responses in the AD brain, its activation being observed in microglia and astroglia surrounding Aβ plaques [35]. In this context, our results emphasized the up-regulation of the Tlr6and Lta genes, known to elicit NFkB-driven inflammation. In AD patients (Table 1), TLR6 and LTA were up-regulated in different brain regions, including the frontal cortex (TLR6 and LTA), gray matter and neocortex (LTA). NFkB activation and the enhanced production of pro-inflammatory factors were additionally demonstrated in the present study by the over-expression of various NFkB target genes. For instance, we found over-expression of genes encoding the pro-inflammatory cytokines IL-1α and IL-1β in the hippocampus of AT mice. IL-1β, a key regulator of neuroinflammation, has been reported to surround amyloid plaques in AD patients, and to be involved in excessive production and processing of the amyloid protein precursor [36,37]. IL-1β levels were found to be elevated in the brains of AD patients along with a six-fold increase in IL-1β immunoreactive microglia in the cerebral cortex [38]. However, the results obtained in various studies are contrasting [39][40][41]. For instance, the analysis of the GSE28146 dataset (Table 1) showed that IL1B mRNA levels were down-regulated in the grey matter of seven AD patients with a severe form of disease. Along with the pro-inflammatory Il1b gene, the up-regulation of the anti-inflammatory Il19 gene [42] was detected in the hippocampus of AT mice, as well as in the frontal cortex of AD patients ( Table 1). The observation that hippocampal Il19 gradually increased during AD progression in APP/PS1 Tg transgenic mice [43] sustains our findings.
The only markedly down-regulated inflammatory gene in the hippocampus of AT mice was Csf2 that encodes GM-CSF, an important neurotrophic factor of the central nervous system [44]. A recent study performed on mild to moderate AD patients showed that treatment with GM-CSF provided memory-enhancing benefits [45], and a clinical trial (NCT04902703) on AD patients for evaluating the safety and efficacy of GM-CSF (Sargramostim) is ongoing. In addition, a marked reduction in the GM-CSF receptor was detected in the hippocampus, suggesting a broader contribution of GM-CSF signaling to AD pathology [44].
As well as the dysregulation of innate immune responses, the impairment of adaptive immunity was attested in the hippocampus of AT mice, pointing towards an anergic state of T lymphocytes [46,47]. This could be a compensatory mechanism for controlling neuroinflammatory responses to Aβ aggregates and other toxic molecules in AD [48]. In turn, the increased levels of the RANTES chemokine reported in this study indicate the enhanced recruitment of T lymphocytes in the AD brain, which may sustain perivascular inflammation [49] and provide protection against thrombin toxicity [50].
A distinctive redox status was evidenced at the transcriptional level in the hippocampus of AT mice, characterized by the over-expression of genes involved in ROS production, mainly addressing superoxide generation by NOX1 [51]. NOX1 transcript levels were found to be up-regulated in the hippocampus of AD patients (Table 1), as well as in the frontal lobe of AD patients in the early stages of disease [52]. Moreover, a post-mortem study on the frontal and temporal cortex from mild cognitive impairment and AD patients in different stages showed elevated levels of various NOX components [53]. In addition to NOX enzymes, increased transcript levels of MPO were found in human AD brains, suggesting a potential contribution of hypochlorous acid to oxidative damage in AD [54].
The only redox gene found to be down-regulated in the hippocampus of AT mice was Nos2 which is involved in nitric oxide (NO) production [55]. Due to the duality of NO, data on its role in AD are contradictory [56]. It has been shown that genetic removal of Nos2 can promote TAU pathology [55], hence sustaining other studies that emphasize the protective role of NO in AD [55,57]. In turn, other studies suggest that NO is involved in nitrosative damage in AD [58,59] via the generation of the highly toxic peroxynitrite in the presence of superoxide [59].
The over-expression of several genes involved in ROS production was accompanied by elevated transcript levels of particular antioxidant genes involved in glutathione biosynthesis and metabolism or in the thioredoxin system, indicating that protective antioxidant mechanisms might be elicited in the AD brain in response to an enhanced oxidative activity. In line with our data, increased levels of the GPX3 transcript were found in the entorhinal cortex from 36 AD patients compared to 16 controls (Table 1). Moreover, thioredoxin up-regulation was shown to play a neuroprotective role in AD [60]. Nevertheless, the antioxidant defense appears to be inefficient as long as disease is persisting and progressing with age in AT mice. It has been suggested that the dysregulation of glutathione homeostasis may contribute to AD pathogenesis [61], and that glutathione levels assessed by proton magnetic resonance spectroscopy in specific brain regions could be clinically relevant in AD [62].
Some of the genes found to be dysregulated in the hippocampus of AT mice were found to be dependent on disease progression with age, therefore being suitable for evaluating the impact of experimental therapies on the disease course.
Blood biomarkers would be a valuable tool for early diagnosis and disease monitoring using minimally invasive methods, if AD has an echo in the blood [63]. We highlighted a common gene transcription pattern in the hippocampus and whole blood of AT mice, suggesting that some transcriptional changes in the brain are mirrored in peripheral blood where they can be dynamically monitored. Such a gene is Il1b whose product was also significantly elevated in the blood of AD patients [64]. In addition, our results indicate that there are important differences in the ongoing pathologic and repair processes occurring in the hippocampus and in the blood of AT mice, with redox disturbances being dominant in the hippocampus, while the inflammation fingerprint is dominant in the blood of AT mice.
By comparing the expression pattern of inflammation and redox genes in AT mice and AD patients, we selected a panel of common genes, specific either for brain or blood.
The hippocampus panel comprises the inflammatory genes F2r, Tlr6 and Lta, participating in the NFkB signaling pathway, along with the anti-inflammatory Il19 gene, complemented by the redox genes Nox1 and Gpx3 that are involved in superoxide production and hydrogen peroxide detoxification, respectively. Nonetheless, other genes found by us to have a significant expression change in AT mice might also be useful for preclinical investigations, considering that their products were shown in various human and mice studies to have a role in AD pathogenesis.
In blood, we identified a distinctive panel of inflammatory genes that encode receptors (Ltbr and F2r), cytokines (Il1b) and growth factors (Csf1), along with the Gsr redox gene. Gsr appears to be an early blood biomarker that can be evidenced even in milder forms of the disease, both in patients and AT mice. Of note is that F2r over-expression is common to the hippocampus and blood, indicating endothelial dysfunction and an increased risk of thrombosis in these compartments, both in mice and AD patients. Moreover, noncanonical NFkB activation through the lymphotoxin B receptor encoded by Ltbr was also demonstrated both in the hippocampus and blood of AT mice.
Altogether, the obtained results bring into focus panels of genes with a common trend of expression changes in AT mice and AD patients that may overcome the generally low animal-to-human translational success [65]. Accordingly, the double transgenic AT model, relevant for mild to moderate AD, might be useful for drug development by allowing a rapid and reliable transfer of preclinical results into clinical studies. Moreover, those common genes with similar expression changes in the blood and hippocampus of AT mice would be useful for dynamic therapy monitoring in blood samples in which the echo of the disease was evidenced at the level of particular inflammation and redox genes.

Animal Model
As animal model of AD, we used a double transgenic mouse model with neuronal expression of human hAPP V717I and hTAU P301L proteins (hereinafter referred to as AT) generated in C57/BL6j background. These mice express both APP and TAU human genes, under the control of the mouse Thy1 gene promoter. The AT mice were obtained by crossing heterozygous APP V717I mice with homozygous TAU P301L mice, for more than eight generations, as described previously [26]. The characteristics of the APP V717I and TAU P301L transgenic mice have been previously described [27,29],recapitulating with aging a combined amyloid and TAU pathology. In this model, the amyloid pathology sets in at 10 to 12 months, and tauopathy is notable at approximately 13 months [27].
Hippocampus and whole blood from 10 transgenic AT mice and 8 age-matched wild type mice (WT), with a mean age of 54.4 ± 3.4 weeks were investigated using pathwayfocused PCR arrays. Additionally, the Gsr and Osgin1 genes were analyzed by qPCR single gene assay in two independent groups, the first one comprising 21 AT vs. 12 WT (mean age 48.5 ± 4.9 weeks) and the second one with 7 AT vs. 8 WT mice (mean age 37.1± 0.8 weeks).
Mice were group-housed in simple cages under standard conditions (normal 12 h light/dark cycle, constant temperature and humidity), with ad libitum access to food and water. All animal experiments were approved by the Ethics Committee of "Victor Babes" National Institute of Pathology, Bucharest, Romania, authorization no. 39

Blood and Hippocampi Collection
Blood was collected by retro-orbital puncture in PAXgene RNA stabilizer solution (Qiagen, Hilden, Germany). Prior to brain collection, mice were anesthetized with ketaminexylazine by intraperitoneal injection (100 mg/kg ketamine and 10 mg/kg xylazine), and were thereafter transcardially perfused with phosphate-buffered saline. Accordingly, most of the blood leukocytes in the brain microvessels were removed, and resident brain cells mostly contributed to the gene expression profile in hippocampus. Brains removed from the skull were snap-frozen immediately after dissection and were stored at −80 • C until use. Immediately before use, hippocampi were dissected as previously described [66], and were homogenized in TRIzol™ reagent using the TissueRuptor (Qiagen, Hilden, Germany).

Gene Expression Analysis
Pathway-focused qPCR arrays. RNA isolation from hippocampi was performed using the miRNeasy Tissue/Cells Advanced Micro Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol, and RNA isolation from whole blood was performed using the modified PAXgene method [67]. RNA was quantified using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). Reverse transcription of 400 ng total RNA was performed for each array experiment using the RT2 First Strand Kit (Qiagen, Hilden, Germany). The expression of 84 key genes involved in redox responses and of 84 genes related to inflammatory processes (Supplementary Table S1) was assessed by qPCR on 7500 Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) and was evaluated with RT 2 Profiler™ PCR Array Mouse Oxidative Stress (PAMM-065ZA, Qiagen, Hilden, Germany) and RT 2 Profiler PCR Array Mouse NFkB Signaling Pathway (PAMM-025ZA, Qiagen, Hilden, Germany). The geometric mean of five housekeeping genes (Actb, B2m, Gapdh, Gusb and Hsp90ab1) was used to normalize the expression level of each transcript in hippocampus, while in blood four housekeeping genes (Actb, B2m, Gapdh and Gusb) were selected. The stability of the reference genes was established using the RefFinder algorithm (http://leonxie.esy.es/RefFinder/, accessed on 15 March 2022) [68]. The gene expression levels were calculated as 2 −∆CT values, and are reported in Supplementary  Table S2. The fold change (FC) in gene expression was calculated as mean2 −∆CT values in the interest mice group divided by the mean 2 −∆CT values in the control group. The obtained results were presented as fold regulation (FR) values: if FC values were higher than 1, FR was equal to the FC value, and if FC values were lower than 1, FR was calculated as the negative inverse of FC.
Single gene qPCR. RNA isolation from whole blood was performed using the modified PAXgene method [67]. Reverse transcription of 500 ng total RNA was performed using the 90 High-Capacity cDNA Reverse Transcription Kit (Thermo Scientific, Waltham, MA, USA) according to the manufacturer's protocol. The expression levels of Gsr and Osgin1 were assessed by qPCR on 7500 Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). The primers are annotated in Supplementary Table S3. The gene expression levels were normalized against the geometric mean of two reference genes, Gapdh and Tbp, using the primers annotated in Supplementary

GEO Data Mining
The following gene expression datasets, reporting the gene expression levels in different brain regions from AD patients and controls were retrieved from the Gene Expression Omnibus database (GEO): GSE1297 (hippocampus), GSE48350 (hippocampus), GSE185909 (frontal cortex), GSE28146 (grey matter), GSE37264 (neocortex) and GSE118553 (entorhinal cortex). The GEO2R analysis tool was used to screen the differentially expressed genes between AD patients and controls. The adjusted p-value was calculated using the Benjamini and Hochberg procedure (the false discovery rate method).

Statistical Analysis
Statistical analysis was performed using the Statistical Package for Social Science (SPSS, version 17.0) and GraphPad Prism 8.4.3. Results are presented as mean ± standard error of the mean (SEM) or standard deviation (SD). The differences in gene expression between AT mice and the corresponding age-matched controls were evaluated using the non-parametric Mann-Whitney U-test. Only changes in gene expression with 1.50 > FR > 1.5 and p < 0.05 were considered significant. Correlations between continuous variables were performed using the Pearson test. Correlations with p < 0.05 and r < −0.8 or r > 0.8 were considered significant.

Conclusions
Altogether, the investigated double transgenic AT mouse model, which recapitulates human amyloid and TAU pathology in terms of neurologic, cognitive, behavioral and psychologic deficits, exhibits specific inflammation and redox gene expression disturbances in the brain and peripheral blood that partly overlap with those detected in AD patients. The identified genes might be valuable candidate biomarkers for preclinical drug development, which could be readily translated into clinical trials. In this context, we appreciate that AT mice represent an operative mouse model for testing new anti-inflammatory and antioxidant therapies in AD by following-up the expression changes in the identified genes.

Conflicts of Interest:
The authors declare no conflict of interest.