We previously investigated the brain eCB system in the TDP‐43 mouse model of FTD and, following the finding of decreased expression of the NAE‐degrading enzyme, FAAH, we treated these mice with a FAAH inhibitor, which improved many behavioral and histopathological alterations linked to FTD in this model [12], effects that we demonstrated in a follow‐up study to be mediated by the activation of CB1 and/or CB2 receptors [14]. In the current study, we focused on the small intestine of this TDP‐43 mouse model, since it has been reported that alterations of the gut‐brain axis, also through the loss of intestinal microbiome homeostasis (gut “dysbiosis”), are involved in several neurodegenerative and neuroinflammatory disorders [26].

First, we observed a reduction of the expression of Gpr55 and Dsg2 and an increase of Pparg expression in the duodenum. The two formers could contribute to dysregulating intestinal inflammation and integrity/permeability, and the latter may represent a negative feedback adaptation to these potential effects. Interestingly, gut microbiota have been shown to be causally associated with FTD [27], and we previously reported that GF mice exhibit reduced Gpr55 expression in small intestine segments [19]. GPR55 in mice has been shown to be implicated in GI motility and inflammation and suggested to be involved in mucosal secretion and homeostasis [28]. On these bases, it is possible that the potential implication of intestinal GPR55 in FTD is modulated by the gut microbiota. Future experiments are needed to address the role of reduced GPR55 signaling in TDP‐43 mouse small intestine. As for Dsg2, its decrease is consistent with the “leaky gut” hypothesis in the context of age‐related diseases such as dementia [29].

Gpr55 and Pparg expression changes were not sustained by any change in the known eCBome ligands of the receptors encoded by these genes and quantified here by HPLC‐MS/MS, that is, N‐palmitoyl‐ethanolamine (PEA) for GPR55, and AEA and DHEA for PPARγ. The duodenum of FTD mice also showed, however, reduced FA levels and increased FA‐derived oxylipins. Interestingly, the increased levels of all oxylipins were accompanied by the decrease of the levels of their respective precursors, that is, DGLA for 12‐HETrE and 15‐HETrE; AA for 8‐HETE, 11‐HETE, 12‐HETE, 12‐KETE, 15‐HETE, and 5,12‐DiHETE; EPA for 12‐HEPE and 15‐HEPE; and DHA for 14‐HDHA + 7‐HDHA and 17‐HDHA. This is consistent with the quantitative conversion of PUFAs into the corresponding oxylipins. Additionally, we also observed increased levels of both SPMs and pro‐inflammatory LTB4 in the duodenum, again suggestive of an altered inflammatory state in this section of the intestine in the TDP‐43 mice group. The reduced levels of LEA, which activate the anti‐inflammatory and intestinal permeability‐reducing PPARα, can also be interpreted in this sense.

No significant gene expression changes were observed in the jejunum, but as previously observed in the duodenum, the levels of some PUFAs and PUFA‐derived oxylipin were altered in TDP‐43 mice. In this case, changes in oxylipin levels were not necessarily accompanied by opposite changes in the corresponding PUFA precursors. Indeed, while 12‐HETE, 5‐HETE, 8‐HETE, 11‐HETE, 15‐HETE, and 5,12‐DiHETE were increased, AA was not significantly changed in the FTD group. The decrease of 9‐HODE was not associated with any changes in LA, and 12‐HETrE and 15‐HETrE increases were not associated with any change in DGLA. Similarly, the increases of 4‐HDHA, 14‐HDHA+7‐HDHA, 17‐HDHA, and 17‐oxo‐DHA were not associated with any change in DHA levels. This disparity between duodenum and jejunum may be explained by the differences in anatomy, function, pH, and reserves of precursor PUFAs between the two regions. On the other hand, 12‐KETE and 5,12‐DiHETE increases were associated with a significant increase of their direct biosynthetic precursor, 12‐HETE.

In the ileum, the increased expression of Cnr1 in TDP‐43 mice was associated with increased levels of its ligands, such as AEA and (with much lower affinity) DHEA. The increased expression of Gpr119 instead was accompanied by a reduction of its ligands, 2‐LG and 2‐OG. This is interesting because Gpr119 is known to play a role in the secretion of incretin hormones and incretins are currently under investigation to be potentially used as a new target in the treatment of neurodegenerative diseases [30]. The increased expression of Napepld was consistent with the trend towards increased levels of AEA, which could play a protective role in the ileum as it seems to do in the brain [12]. This adaptive response did not reach statistical significance as it may have been possibly mitigated by the concomitant increased expression of the AEA‐degrading enzyme, FAAH. It was recently shown that stabilizing NAPE‐PLD at the thiazide‐binding site is beneficial for curing hypertension [31]. Interestingly, hypertension is a risk factor for cognitive decline [32]. Thus, we can speculate that the potentially beneficial implication of intestinal NAPE‐PLD in FTD may be via its implication in hypertension. Finally, while most of the oxylipin changes observed in the ileum of the TDP‐43 group were not associated with changes in their PUFA precursor levels, the reduction of 13‐HODE‐G was accompanied by the significant reduction of its precursor, 2‐LG, in FTD mice.

In all the regions analyzed, oxylipin levels were altered in TDP‐43 mice, reflecting the oxidative state of the regions. This finding highlights the potential importance of PUFA‐derived oxidation products, whether enzymatically or nonenzymatically synthesized, in the pathophysiology of neurodegenerative disorders such as FTD. Both cerebral and peripheral oxidative stress have been associated with several brain conditions, such as stroke [33] and dementia [34], but little is known about oxylipin‐controlled free radical implication in FTD. However, oxylipins are known to be implicated in the modulation of neuroinflammation and to play a role in some neurodegenerative disorders, including dementias [35, 36]. Furthermore, it has been shown that several lipid metabolism abnormalities play a role in the pathophysiology of neurodegenerative diseases [37, 38].

Although underpowered due to the relatively low number of replicates, the analysis of the DNA of the intestinal content and feces suggested overall more microbiota changes in TDP‐43 mice in the lower part of the gastrointestinal tract. These preliminary results agree with the fact that microbial density within this tissue is lower in the proximal parts, and higher in its distal parts. All the alterations (except one in the ileum at the genus level) were observed in the caecum and feces. Gut microbiota, as well as some of their metabolites, have been demonstrated to be involved in several metabolic, psycho‐metabolic and mental disorders [17, 39, 40, 41]. As an important part of such metabolites, SCFAs have been extensively studied, and some of them are more characterized than others. These volatile lipophilic metabolites are known to be implicated in neurotransmission, neuroinflammation, neurometabolism and blood–brain barrier integrity [42]. Concerning their present quantification in the feces of FTD mice, they showed a statistically significant reduction of their levels in the case of butyric, isobutyric and acetic acid. Interestingly, these three SCFAs have been associated with good cognitive performance in mice [43] and humans [44], which suggests that the reduction of their levels in the TDP‐43‐FTD mouse model might be one of the causes of the lower cognitive performances of those mice, which is a feature also of FTD patients. Decreased SCFA levels have also been reported in other neurodegenerative disorders, such as Parkinson's disease [45] and Alzheimer's disease (AD) [46], and more generally they have also been associated with other mental disorders such as autism spectrum disorder (ASD), major depression, addiction‐related behaviors and schizophrenia [47]. As fecal SCFA levels represent only a portion of the total circulating SCFAs [47], their reduction in the feces of TDP‐43 mice might either be due to a reduction of their production by intestinal bacteria (due to a change in the diversity and composition of the gut microbiota) or to reduced excretion in the feces, with a bigger portion kept into the system. With this in mind, we thought it important to be careful with the interpretations of these data.

Butyric acid is one of the most studied SCFAs. It is known to have several effects on the host, such as contributing to the maintenance of colon health, improving glucose homeostasis, maintaining gut integrity by regulating tight junction proteins, and reducing inflammation [48]. It may also have additional regulatory effects on the production of neurotrophic factors, such as BDNF [47], whose activity is known to be impaired in the case of TDP‐43 abnormal function [49]. Here, reduced butyric acid levels in TDP‐43 mice, however, were not correlated to decreased abundance of known butyrate‐producing bacteria (data not shown), such as Bacillus subtilis [50] and Eubacterium xylanophilum [51]. Nevertheless, butyric acid has been described to be negatively correlated with anxiety and depression, which may be relevant to some of the psychiatric symptoms observed in FTD patients as well as in the TDP‐43‐FTD mouse model. It is even hypothesized that depression might be a predictor of early‐onset FTD neurodegeneration and hence represent the first symptom of dementia [52]. Depression is the most prevalent noncognitive comorbidity that occurs along with cognitive deficits and is associated with neurodegenerative disorders and cognitive decline [53]. Even though the association between depression and FTD is not well established, it was documented that depressive symptoms are the most frequent neuropsychiatric comorbidity of AD, another neurodegenerative disease, for which 50% of the patients are concerned [54]. Anxiety, which is the most frequent comorbidity associated with depression, is also frequently reported in the case of FTD patients [55]. Interestingly, Kalkan et al. [18] found that butyric acid reduces eCBome signaling, which reinforces the relevance of studying the eCBome and SCFAs together in the context of FTD.

Several studies showed that isobutyric acid, like butyric and acetic acid, is dysregulated in brain disorders. Indeed, in a valproic acid‐induced ASD rat model, the levels of these SCFAs were reduced in the feces of ASD rats compared to control rats nonexposed to valproic acid. These changes were also accompanied by changes in gut microbiota composition and neurotransmitter levels in the brain [56]. Another rodent study showed that mice receiving an FMT with feces from exercised mice would manifest an increase in their learning and memory skills associated with an increase in their caecal levels of SCFAs, including isobutyric acid, butyric acid, and acetic acid [57]. Another study, this time in human plasma, associated the reduction of these SCFAs with a diabetic cognitive impairment when compared to type‐2‐diabetic patients without cognitive symptoms [44].

Acetic acid, like butyric acid, is known to regulate food intake by acting on the hypothalamus [17]. This is interesting because it has been described that FTD patients have a dysregulated food intake. Indeed, eating abnormalities are present in up to 60% of patients with FTD and are one of the six core symptoms required for the diagnosis of behavioral variant FTD (bvFTD). The changes may include the following parameters: swallowing, appetite, stereotypic eating behavior and table manners, food preference (including sweet preference), and other oral behaviors such as food cramming and increased smoking [58]. Furthermore, a case–control study highlighted that early‐onset dementia (EOD) can be influenced by environmental factors and lifestyles, including diet. Indeed, in this study it was shown that adherence to a specific dietary pattern (MIND, Mediterranean‐DASH Intervention for Neurodegenerative Delay) may decrease the risk of EOD [59]. Recently, a reduction in acetate levels in mice has been shown to be associated with obesity and elevated levels of the eCB, 2‐AG, in the hypothalamus [17], again highlighting the potential importance of the gut microbiome‐eCBome axis, also in the context of metabolic and eating disorders with a neuroinflammatory component [40].

As it is well known, the diet influences gut microbiota composition and diversity [60]. Therefore, it makes sense that FTD patients with disturbed eating habits end up with gut dysbiosis. However, in the context of our FTD model, TDP‐43 mice were all fed with the same diet and did not display eating alterations (data not shown). Nevertheless, their gut microbiota was altered compared to WT mice. Indeed, in 2020 [61] it was shown that the gut microbiota composition of a genetically modified mouse model of amyotrophic lateral sclerosis (ALS) and FTD (C9ORF72 mutation) plays an important role in the premature death of these mice and in the pathophysiology of FTD, by influencing the inflammatory response. When gut microbiota of FTD mice were targeted with broad spectrum antibiotics or fecal transplant, the inflammatory phenotypes of the mutant mice were attenuated. More generally and more recently, many studies demonstrated the implication of the gut microbiota in dementia disorders including FTD [27, 62]. Concerning the gut microbiota of our TDP‐43 mice, we preliminarily observed, in the feces, that Paraprevotella and Monoglobus were increased. Interestingly, it was previously shown that Paraprevotella was increased in mice with cognitive deficit in a model of AD [63]. Thus, species belonging to this genus may play a role in the pathophysiology of FTD. In contrast, a human study showed that Monoglobus abundance decreased in AD patients [64]. This may indicate either a species effect, or that this genus does not play a general key role in the pathophysiology of dementia.

Most of the microbiota taxonomic differences between TDP‐43 and WT mice that were suggested by our preliminary DNA sequencing results were observed in the caecum, in agreement with this region being the gastrointestinal section with the most bacterial diversity and abundance. In the caecum, we observed an increase in the relative abundance of the genera Butyricicoccus, Oscillospira, and Candidatus Arthromitus. Interestingly, Butyricicoccus was shown to be increased in an AD model of rats [65], which is consistent with our finding. In addition, Oscillospira was positively correlated with aging‐related symptoms [66], which is also in agreement with our findings, as FTD, like other dementias, is an age‐related disorder (early‐onset forms do not represent the majority of the cases). A study from 2019 showed that Candidatus Arthromitus is a specific inducer of differentiation in T helper 17 cells, which secrete the inflammatory cytokine IL‐17 [67] that was observed to be increased in FTD patients [68], results that are consistent with our findings. In the caecum we also showed a reduction in the abundances of Paraprevotella (contrary to what was observed in the feces), Mucispirillum, and Akkermansia. Consistent with our findings, Mucispirillum is reduced in mice with disrupted cognition [69] and has been reported to be negatively correlated with cognitive abilities and positively correlated with IL‐6 (pro‐inflammatory cytokine) levels in the brain cortex of the SAMP8 mouse model of dementia [70]. Indeed, a study on human circulating biomarkers of FTD published in 2015 reported an increase in the levels of plasmatic IL‐6 in 230 patients with FTD, independently of the clinical and genetic subtypes [71]. Moreover, in AD‐prone mice, Mucispirillum was strongly associated with immune inflammation [72]. In contrast, other studies showed a negative correlation between Mucispirillum and cognitive functions in AD mice [73], as well as with an increase in Aβ1–42‐induced AD mice [74]. Mucispirillum was also highlighted as a potential biomarker for insecticide exposure‐induced depression [75]. On the other hand, Akkermansia was increased in an AD mouse model with improved cognition [76], suggesting that cognitive impairment is probably associated with a reduction of Akkermansia, in agreement with our present finding of reduced Akkermansia relative abundance in TDP‐43 compared to WT mice. In the ileum, only the genus DNF00809 showed a difference between the two genotypes, being increased in TDP‐43 mice. Unfortunately, this genus is not so well documented concerning its link with the brain and cognition. Only one study showed a link with aging (which could be extrapolated to reduced cognition), showing that this genus was abundant in young female rats compared to middle‐aged ones [77].

The above considerations on the chemotaxonomic differences in the gut bacteriomes of WT and TDP‐43 mice are somehow toned down by two important limitations of this study. First, and as mentioned above, we could only use a relatively low number of mice for each genotype (6 WT and 5 TDP‐43 mice), which certainly limited the statistical power of our analysis, although it did not prevent us from observing potential differences among intestinal sections from the same genotype, nor from finding some interesting differences in SCFA levels. Secondly, WT and TDP‐43 mice were not bred together, that is, they were not littermates, and, additionally, mice belonging to each genotype were not single‐housed. This may have generated artifactual differences between genotypes, such as those related to cage effects. However, despite these limitations, we believe that the observed differences reflect, at least to some significant extent, genotypic differences, in as much as they may explain some genotypic behavioral differences and are often in agreement with previous findings in other models of dementia and accompanying co‐morbidities. Further, they are sided by strong differences in eCBome signaling that, on the one hand, are instead unlikely to be due to cage effects, and, on the other hand, have been previously shown to often determine, or be determined by, gut microbiome alterations [19, 20, 40, 78].