Motor injury is caused by TBK1 deficiency in the myeloid cell and rescued by PEI-manose-TBK1

Background: TBK1 haploinsufficiency has been shown to cause both amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD); however, the mechanism is unclear. Methods: A myeloid Tbk1 knockout mice (Tbk1-LKO mice) was established. Motor functional and pathological analyses were also performed. The p-TBK1 was tested by flow cytometry in the ALS animal model and patients. The inflammatory proteins and mRNA was analyzed by Western blot and RT-PCR. Results: We found that the latency to fall in seven-month-old Tbk1-LKO mice was significantly reduced on evaluation on two consecutive days. Overall, 25.6% of Tbk1-LKO mice presented paralysis symptoms and signs along with a loosened myelin sheath and axon degeneration at 14-16 months of age. Furthermore, Tbk1 deficiency in myeloid cells induced inflammatory cell infiltration and dysbacteriosis in the digestive tract. Additionally, p-Tbk1 content was reduced by 29.5% and 14.8% in monocytes of definite ALS and probable ALS patients and decreased by 27.6% and 45.5% in the monocytes and microglia of ALS animals, respectively. PEI-mannose-TBK1 or PEI-mannose-SaCas9-sgRNA for deleting mutant SOD1 in macrophages significantly delayed disease onset and prolonged survival in the ALS mouse model. Conclusions: These data suggest that inflammatory monocyte and macrophage infiltration and impaired innate immune defense are contributing factors to ALS and FTD. muciniphila; TNFR:tumor necrosis factor receptor.

The amplification conditions were as follows: 95°C for 10 min followed by 40 cycles of amplification at 95°C for 30 s, 60°C for 1 min, and 72°C for 30 s. The difference in CT (ΔCT) values was computed as the difference between the human SOD1 CT value and the mouse IL2 CT value (ΔCT=mIL2 CT-hSOD1 CT).
Total RNA was isolated from the tissues using an RNA rapid extraction kit (Generay Biotech Co., Ltd., GK3016) in accordance with the manufacturer's protocol. cDNA was prepared using AMV reverse transcriptase and random primers (Promega, A3500). Targeted gene primers and SYBR Green (Generay Biotech Co., Ltd., GK8020) were used for quantitative PCR on an M3005P (Agilent Technologies, Santa Clara, USA). After incubating the samples at 95°C for 10 min, 40 cycles of amplification at 95°C for 10 s, 60°C for 20 s, and 72°C for 15 s were performed. GAPDH was used as a reference gene. The synthesized primers used here are listed in Table s2.

Flow cytometry
Osmotic lysis of red blood cells was performed using 45 mL of RBLB (red blood cell lysis buffer: 50 mM NH4Cl, 70 mM NaHCO3, 0.1% EDTA, ddH2O) for 5 mL of whole blood for 10 min at RT. Leukocytes  Thoroughly   resuspended fixed/permeabilized cells were incubated with p-TBK1-PE (1:100, Cell Signaling, 13498) at 4°C for 30 min in the dark. After washing cells twice with 1×BD Perm/Wash TM buffer and resuspending in staining buffer, flow cytometric analysis was then performed.

Electron microscopy
Tissues were fixed in 4% glutaraldehyde and treated with 1% OsO 4 in 0.1 M PBS. The samples were dehydrated through graded acetone solutions and embedded in EPON 812. Ultrathin sections (70 nm) were placed on a copper grid and stained with uranyl acetate and lead citrate. The samples were observed by transmission electron microscopy (TEM, JEM-1230).

Bielschowsky staining
Staining of nerve fibers was carried out using a Hito Bielschowsky OptimStain Kit (HTKNS1126).
Briefly, the slides were incubated with Solution 1 for 15 min, followed by immersion in the Developer solution for 1-8 min until the tissues became golden-brown. After incubation with Solution 4, sections were rinsed with distilled water, dehydrated, cleared, and coverslipped.

Motor function test
A rotarod (Rota-Rod; Ugo Basile, Gemonio, Italy) was used to assess motor function in mice. With an arbitrary cut-off time of 180 s, three once-per-week trials were performed, and the longest time was recorded. The onset of disease was determined as a reduction in rotarod performance between weekly time points. The weight of the mice was measured every 7 days, starting at 60 days of age.
Accelerated rotarod testing. Mice were placed on the top of the revolving beam for 4 successive trials/day for 2 days, with 20-min intertrial intervals. The rod was accelerated gradually from 4 to 28 rpm over 2 min. Latencies before falling from the rod were recorded.
The footprints of the mice were recorded during continuous locomotion. The stride length was measured as the distance between footprints, and the average stride length was calculated from three stride lengths.
The weight of the mice was measured every 7 days, starting at 60 days of age.

Microbiota analysis
Total DNA was extracted from the feces of mice. PCRs targeting the V3-V4 regions of the 16S rDNA genes (probes used: 319F, 5'-ACTCCTACGGGAGGCAGCAG-3' and 806R, 5'-GGACTACHVGGGTWTCTAAT-3) were performed, and sequencing reactions were performed using Illumina HiSeq sequencing technology for paired-end reads. Paired-end reads obtained were merged using FLASH v1.2.11 software, and reads with a length ≥ 400 bp were kept for the following analysis.
The merged sequences were processed with QIIME v1.8.0, and the sequences were binned into operational taxonomic units (OTUs) based on 97% identity. Alpha diversity, richness, Venn diagrams, heat maps, principal coordinate analysis, and linear discriminant analysis were further processed with a bioinformatic pipeline tool, BMK Cloud online (Biomarker Technologies Corporation, Beijing, China).

Statistical analysis
Data were analyzed using SPSS 13.0 software. Lifespan was compared between groups of mice by the Kaplan-Meier method. Comparisons among multiple groups were performed using one-way ANOVA followed by Student-Newman-Keuls or Dunn's T tests. All values are expressed as the mean ± S. E., and P < 0.05 was considered statistically significant.

Tbk1-LKO leads to motor injury
Recent works strongly support that TBK1 haploinsufficiency causes neurodegeneration, including ALS and FTD. It is well known that TBK1 plays a pivotal role in the innate immune response against virus infection. Therefore, we generated Tbk1 conditional KO mice by crossing Tbk1-flox mice [10] with LysMcre mice expressing Cre in the myeloid cells that induce the innate immune response. The resulting Tbk1 fl/fl LysMcre (hereafter called Tbk1-LKO) mice and Tbk1 fl/fl wild-type (WT) control mice were genotyped by PCR ( Figure 1A). Flow cytometry analyses readily detected that the rate of p-Tbk1 in the ly6c-positive monocytes of WT mice was 5.6%; however, that in the Tbk1-LKO mice was 0.4% ( Figure 1B). To assess the behavioral impact of Tbk1 deletion, locomotor function was evaluated by the rotarod test in Tbk1-LKO and age-matched WT mice. Seven-month-old Tbk1-LKO mice exhibited a significantly reduced latency to fall on evaluation on two consecutive days (WT on day one: 179.4±17.84 s, Tbk1-LKO on day one: 122.2 ±20.04 s; WT on day two: 222.6±11.86 s, Tbk1-LKO on day two: 165.6 ±24.92 s; Figure 1C). To determine whether the locomotor damage in the Tbk1-LKO mice deteriorated with increasing age, the common conditions of thirty-nine Tbk1-LKO mice were recorded until the cut-off age of 14-16 months (Figure s1A). Overall, 25.6% of Tbk1-LKO mice presented paralysis, abnormal footprints and stretch width, weight loss and shortened survival ( Figure   1D-I, Figure s1A). Furthermore, as shown in Figure 1J and Figure s1B, the gastrocnemius fiber area was reduced by 24% in the Tbk1-LKO mice compared with that in WT mice. We hypothesized that Tbk1 deletion could result in motor neuron degeneration and gliosis in the spinal cord. Surprisingly, we did not find degenerated motor neurons or activated microglia or astrocytes in the spinal cord activates p52. Nik-smi1, a noncanonical NF-κB inhibitor, restored cell activity by 11.1% compared with that Tnf-alpha and AML were administered (0.3±0.02). These data supported that Tbk1 inefficiency combined with Tnf-alpha coinduced neuron degeneration that was partially rescued by a noncanonical NF-κB inhibitor. Furthermore, AML activated human leukemic monocytes (THP-1) at doses of 1 μM and and induced the activation of noncanonical NF-κB pathways. Inhibition of Tbk1 activity by AML induced a two-fold increase in p52 generation in THP-1 cells (Figure s3D-F).

Tbk1-LKO induced intestinal tract inflammatory monocyte infiltration
Interestingly, we found that the intestinal tract was severely swollen in the Tbk1-LKO mice, especially in the small intestine (Figure 2A). Hematoxylin and eosin (HE) staining showed a large amount of inflammatory cells infiltrated in the mucous and submucous layers ( Figure 2B). Flow cytometry analysis showed that the abundance of CD11b-positive cells with high expression of Ly6c significantly increased in the small intestine, both of which are markers of inflammatory monocytes ( Figure 2C).
Given that axonal pathology is related to inflammatory cell infiltration, CD11b expression was evaluated in the cortex, spinal cord, muscle, spleen, liver and small intestine. The level of CD11b was significantly increased in the small intestine in Tbk1-LKO mice ( Figure 2D-E). The CD11b-positive cells aggregated on mainly the mucosa of the small intestine ( Figure 2F). Macrophages in the small intestine are derived from monocytes and differentiate into M1 macrophages that highly express CD68 and Tnf-alpha and into M2 macrophages marked by Arg1 and Ym1 expression. Real-time PCR revealed that the levels of Arg1 and Ym1 were significantly reduced and that the levels of Tnf-alpha and CD68 did not show an increased tendency in the small intestine and colon in the Tbk1-LKO mice compared with those in WT mice ( Figure 2G). We further performed immunostaining and immunoblotting to evaluate Arg1 expression in the small intestine. Arg1-positive cells were located mainly in the lamina propria of the small intestine. Consistent with the reduced level of Arg1 mRNA, Arg1 protein expression was also dramatically reduced in Tbk1-LKO mice (Figure s4A-C). Therefore, Tbk1 deletion in myeloid cells could induce inflammatory monocyte infiltration in the small intestine.
Next, enteric bacterial 16S sequencing was performed to determine whether Tbk1-LKO damaged the homeostasis of the microbiome in the gut. Indeed, the ratio of Firmicutes vs Bacteroidetes was reduced by 32% in the Tbk1-LKO mice compared with that in WT mice ( Figure 2H-I, p=0.05). Through KEGG prediction, dysbacteroisis could be related to immune disease and neurodegenerative disease ( Figure 2J).

The level of p-Tbk1 was reduced in inflammatory monocytes and microglia in ALS patients and animal models
Since Tbk1 deletion in myeloid cells led to nerve injury, the level of Tbk1 phosphorylation in leukocytes was tested by flow cytometry in ALS patients and an ALS animal model that overexpresses human mutant SOD1. Based on side and forward scatter patterns, we could determine three groups of cells, lymphocytes (purple), granulocytes (blue) and monocytes (green). In particular, the monocytes had a higher level of p-Tbk1 than the granulocytes and lymphocytes ( Figure 3A). Furthermore, the classical monocytes were gated by CD14 positive and CD16 negative signals, and we found that the mean fluorescent intensity (MFI) of p-Tbk1 was reduced by 29.5% and 14.8% in monocytes of definite ALS and probable ALS patients compared with that in monocytes of healthy subjects ( Figure 3B,   Figure s5A). In addition, 50% of ALS patients presented a lower level of p-TBK1 MFI than that in healthy subjects (Figure s5B). The level of p-Tbk1 was also analyzed by flow cytometry in the monocytes and microglia of ALS animals based on ly6c and cd11b expression, and p-Tbk1 was decreased by 27.6% and 45.5%, respectively ( Figure 3C-E). Subsequently, spinal cord sections were stained for CD11b and p-Tbk1 at different stages (60 days, 90 days, and endstage), which showed that p-Tbk1 was located in mainly amoeboid microglia characterized by a round cell body and thin filopodium-like processes, with little p-Tbk1 in ramified inactive microglia. With the progression of disease, the abundance of activated microglia significantly increased in the spinal cord and did not overlap with p-Tbk1 signal ( Figure 3F). Thus, the reduced level of p-Tbk1 in the monocytemacrophage system could contribute to the pathogenesis of ALS.
Recent studies demonstrated that human endogenous retrovirus (HER-V) activation might be one of the causes of ALS. Therefore, the mRNA of HERV-Kevn, HERV-Kpol and HERV-Kgag and the mRNA of IFN-beta, TNF-alpha and MAVS, which are related to activation of the TBK1-IRF3-IFN-beta pathway, were evaluated by real-time PCR. Figure s6A shows that the level of IFN-beta mRNA decreased significantly with the increase in TNF-alpha in the monocytes of ALS patients. However, we did not find that HERV-K was activated in the monocytes (Figure s6B). The rate of inflammatory monocytes was also significantly increased in the ALS patients (Figure s6C). The monocytes of ALS patients were further differentiated into macrophages by using M-CSF, which showed a shedding cell process and increasing phagocytosis rate (Figure s6D-E).

PEI-mannose TBK1 extends survival of ALS transgenic mice
Considering that the level of p-Tbk1 was decreased in monocytes and microglia, the noncanonical NF-κB pathways could be activated in the ALS animal model. As shown in Figure 4A and Figure s7A, p52 was markedly increased in the ALS animal model, especially in microglia (Figure s7B). Therefore, we hypothesized that activated microglia, monocytes or macrophages expressing mutant SOD1 might be a potential therapeutic target. Moreover, we evaluated the distribution of PEI-mannose linked the p-EGFP-N1 plasmid targeting macrophages by intravenous injection. The rate of GFP-positive cells was 1.7%, 14.3% 15.8%, 26.3% and 18% among the CD206-positive cells in the blood, spinal cord, liver, spleen and small intestine, respectively ( Figure 4B). The SaCas9-sgRNA5 system, which was used in our previous study and effectively deleted mutant SOD1 in motor neurons, was delivered by PEImannose at the presymptom stage of SOD1G93A mice. Surprisingly, the onset of disease was significantly delayed for 26 days, and the footprint and rotarod performance was markedly ameliorated ( Figure 4D-G, video 1). However, the treatment did not significantly prolong the life span of SOD1G93A mice compared with that of the control mice ( Figure 4H

Discussion
FTD is a heterogeneous neurodegenerative disease that often coexists with ALS. Several genetic alterations, including mutations in the Tbk1 gene, are common to both disorders. Tbk1 loss-offunction mutations have been shown to cause FTD and familial ALS [5] , with reported frequencies of 1.1%, 3.4%, and 4.5% for these mutations in FTD, ALS, and FTD-ALS, respectively [19] . In this paper, we first reported that p-Tbk1 was reduced in monocytes and macrophages in sporadic ALS patients and in an SOD1 G93A animal model and that conditional deletion of Tbk1 in myeloid cells induced axonal nerve damage. Recent studies of Tbk1 conditional KO in B cells, T cells and dendritic cells have shown that Tbk1 regulates the homeostasis of immunity [13][14][15] . David R Beers and Stanley H Appel deeply reviewed the immune dysregulation in ALS and suggested that targeting the immune system could provide therapeutic strategies to ameliorate the pathogenesis of ALS [20] . ALS-related mutations (SOD1, TDP-43, C9, TBK1, OPTN, etc.) activated inflammatory responses and released inflammatory factors, including Tnf-alpha, Il-6, etc., eventually inducing motor neuron death [20][21] . Myeloid Tbk1 deletion induced inflammatory monocyte infiltration in the digestive tract and motor injury. PEImannose-TBK1 or PEI-mannose-AAV-SaCas9-sgRNA deleting mutant SOD1 in macrophages significantly delayed the disease onset and prolonged the survival of the SOD1 mutant animals. Thus, the inflammatory response caused by myeloid Tbk1 deficiency could contribute to the pathogenesis of ALS or FTD.
Several studies have supported that the gut microbiome could contribute to the pathogenesis of ALS and other neurodegenerative diseases. Eran Blacher et al demonstrated that Akkermansia muciniphila (AM) ameliorates whereas Ruminococcus torques and Parabacteroides distasonis exacerbates the symptoms of ALS [22] . The ALS mouse model presented a damaged tight junction structure and increased permeability [23] . In this paper, we found that Tbk1 deficiency in myeloid cells resulted in an intestinal microbiome shift, inflammatory monocyte infiltration and noncanonical NF-κB activation. The noncanonical NF-κB pathway selectively activates p100-sequestered NF-κB members, predominantly NF-κB2 p52 and RELB [24] . Shao-Cong Sun has deeply reviewed the role of noncanonical NF-κB in inflammation and immunity and elucidated that the noncanonical NF-κB pathway is slow and persistently activated by the ligands of a subset of tumor necrosis factor receptor (TNFR) superfamily members [25][26] . The slow progression with chronic inflammatory cell infiltration is the most prominent feature of ALS or FTD. Thus, enhancing TBK1 activity in macrophages could be a potential therapeutic target to control inflammatory cell infiltration and regulate the intestinal microorganism balance for ALS or FTD.

Conclusions
Our results suggest that inflammatory monocyte and macrophage infiltration and impaired innate immune defense are contributing factors to ALS and FTD and enhancing TBK1 activity could be a potential therapeutic target for patients related with ALS or FTD.     with PEI-mannose-TBK1, n=3, bar=50 mm, *P < 0.05, compared to WT control. Table s1 Common data for ALS and healthy subjects. Table s2 Primers used in this paper.
Video 1 The locomotor activity of mice treated with PEI-mannose-AAV-SaCas9-sgRNA targeting SOD1 was recorded at 120 days of age.
Video 2 The locomotor activity of mice treated with PEI-mannose-TBK1 was recorded at 120 days of age.