Introduction

Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder characterized by the loss of upper and lower motor neurons, leading to muscle weakness, disability, and death, with a median survival of 3 to 5 years [1]. ALS is a genetically and clinically heterogeneous disease in which the interaction between genetic background and environmental factors are thought to play a major role [1]. Familial ALS accounts for approximately 10% of cases, which results from genetic alterations in several genes including superoxide dismutase 1 (SOD1), TAR DNA-binding protein 43 (TDP-43), chromosome 9 open reading frame 72 (C9orf72), and fused in sarcoma (FUS) [2]. The remaining 90% of cases are sporadic, suggesting an important environmental component.

The intestinal microbiota encompasses trillions of organisms that inhabit the gut [3] and can play a role in neurologic diseases by modulating immune responses in the CNS, altering endocrine signaling along the hypothalamic pituitary axis, and by directly signaling through afferent nerves [4]. Studies have found alterations in the gut microbiota of patients with ALS. Brenner et al. studied 25 patients with ALS vs. 32 healthy controls [5] and found alterations in Ruminococcaceae. Blacher et al. analyzed a cohort of 37 patients with ALS vs. 29 healthy controls and found decreased abundance of microbial genes involved in nicotinamide and tryptophan metabolism [6]. Nicholson et al. sequenced the largest number of ALS subjects to date (n = 68) vs. healthy controls (n = 61) and found a decrease in the butyrate producing bacteria, Roseburia intestinalis and Eubacterium rectale [7]. The changes in butyrate producing bacteria are consistent with a case study in which 5 patients with ALS had low levels of other butyrate-producing bacteria [4) compared to unique genes (Supplemental Figs. 5 and 6). Antibiotics modulated 27 genes in the same direction in both WT and SOD1 mice (14 up and 13 down), including upregulation of the MGnD regulator Apoe and downregulation of ubiquitin binding protein (Ubc), involved in autophagy. Antibiotics modulated some genes in opposing directions in WT and SOD mice, including Tmem175, a lysosomal potassium channel which plays an important role in clearance of autophagosomes [56], and Gpr165, a homeostatic microglia gene. These findings indicate that the microbiota may have opposing effects on microglial function in healthy vs. diseased animals.

Discussion

Our study confirms and extends the finding that the gut microbiota plays a protective role in the SOD1 model of ALS and identifies a new mechanism related to microglia. The microbiota influence over microglia could be mediated by modulating peripheral immune cells that traffic to the CNS or the production of metabolites, which warrants further study. Microglia play an important role in the maintenance of brain homeostasis but lose this homeostatic function in ALS [20, 57]. We recently identified a neurodegenerative molecular signature in microglia from SOD1 mice which we termed MGnD, in contrast to a homeostatic microglial phenotype [16]. Studies have shown that the gut microbiota maintain microglial function in homeostasis and that antibiotics can decrease homeostatic microglia signatures [27]. Consistent with this, we found that antibiotics decreased homeostatic genes P2ry12, P2ry13, and Cst3 in SOD1 mice. However, we did not observe a decrease in homeostatic genes in antibiotic treated WT mice, as previously reported in Erny et al. [27]. A key difference is that we used a low-dose antibiotic treatment regimen that modulated the microbiota, whereas Erny et al. used a high-dose antibiotic treatment regimen that mimicked a germ-free state. Unique to our study, we found that antibiotics increased MGnD genes in SOD1 mice, including Apoe, Cst7, Lgals, Axl, and others, which is opposite to the effects observed in WT mice and models of AD and PD [11, 27, 28]. We found 109 genes modulated by antibiotics that enhanced SOD1 vs. WT changes in microglia, indicating that an antibiotic-induced dysbiosis amplifies a microglia MGnD phenotype. Furthermore, we found that changes in microglia preceded altered motor dysfunction in antibiotic-treated mice, suggesting that the microbiota slows disease progression by restraining a neurodegenerative microglial phenotype. It is possible that some of these genes, including granulin, reflect early markers of progression and may be a repair response to neurodegeneration. Grn is linked with disease progression in ALS as it is elevated in patients and mice that have progressed in their disease course but is not elevated at disease onset ALS patients compared to healthy controls or pre-symptomatic SOD1 mice [53]. While elevated Grn is associated with ALS, it is suggested to play a protective role by restraining microglia inflammation that leads to neurotoxicty [55].

Because the SOD1 model exhibits a progressive disease that requires more than 4 months of antibiotic intervention, we utilized two microbiota interventions that could lead to a chronic mild depletion (antibiotics) or augmentation (co-housing). In our study, we selected a low-dose antibiotic regimen shown to be well tolerated for long-term administration [10]. This leads to antibiotic-induced dysbiosis, rather than full microbiota depletion. We found that antibiotics initially depleted most populations of the endogenous microbiota, then led to an increase in antibiotic resistant organisms, which may be linked to the altered disease progression that we observe later in the disease course. Antibiotic treatment did not affect the time of disease onset but did affect disease progression, potentially suggesting that administration of antibiotics in the early-symptomatic phase could also have a similar effect of the disease course.

We identified several groups of bacteria depleted by antibiotics that may have beneficial roles in ALS. The Gram-negative anaerobe, Akkermansia was depleted at multiple time points. Akkermansia has recently been shown to ameliorate disease in antibiotic pre-treated SOD1 mice, which is linked to the production of nicotinomide [58]. Akkermansia may also have beneficial roles for other neurologic diseases, including multiple sclerosis [43], epilepsy [59], and Alzheimer’s disease [60]. Antibiotics also depleted several members of Clostridial clusters IV and XIVa, which are major butyrate producers in the gut. Two independent studies have found that butyrate producing bacteria were depleted in ALS [7, 15]. We did not observe motor dysfunction in antibiotic-treated non-transgenic WT mice. Thus, based on our experimental data, it appears that low-dose antibiotics worsen disease only in genetically susceptibly hosts.

It is possible that antibiotics could have an off-target effect and that antibiotic-mediated toxicity may be responsible for the changes, rather than antibiotic-induced alterations in the gut microbiota. A higher dose combination of antibiotics in the C9orf72 model of ALS initially reduced microglia infiltration into the CNS and ameliorated disease, but then led to off-target health consequences [29]. We did not observe a MGnD microglial phenotype in low-dose antibiotic-treated WT mice. Furthermore, in an animal model of AD, this low-dose combination of 8 antibiotics was well tolerated for several months and reversed a MGnD microglial phenotype [10, 28].

Several studies have found that co-housing a genetically susceptible mouse with a wildtype mouse can transfer disease phenotype, with either the pathogenic or protective trait transferred by co-housing [31,32,33,34]. For example, cohousing 5XFAD and WT mice led to cognitive impairment in the WT mice, which was associated with infiltration of Th1 cells in the brain and increased inflammatory cytokines [31]. In our co-housing experiment, we saw no effect on motor function, neurologic score, or survival in SOD1 or WT mice, suggesting that this trait is not transmissible via the microbiota alone. Cohousing may have less of an impact than antibiotics on the microbiome due to inherent colonization resistance. Furthermore, an important point for our study is that we also saw little difference between SOD1 mice and WT littermate microbiota. Other studies have found that differences in the gut microbiota in WT vs SOD1 mice are vivarium dependent [6], and vivarium-dependent changes in the gut microbiota in the C9orf72 ALS animal can determine disease susceptibility [29]. Thus, in a colony of SOD1 mice that have a microbiota distinct from WT mice, cohousing could alter potentially alter disease progression.

Conclusions

In summary, our study highlights the importance of disease-specific interactions between the microbiota and microglia. Furthermore, given the critical role of microglia in ALS, our finding that the microbiota restrains neurodegenerative microglia in SOD1 mice has important implications for the pathogenesis and treatment of subjects with ALS as detrimental effects of antibiotics have been observed in ALS patients. We identified two groups of bacteria that have been reported to have beneficial roles in ALS, including Akkermansia and butyrate-producing bacteria, and further work is needed to confirm their protective role. Finally, we were not able to confer disease protection to SOD1 mice or transmit disease to WT littermates via cohousing, highlighting the importance of the interaction of genetic and environmental risk in the SOD1 model of ALS.