Alleviation of Cognitive Impairment-like Behaviors, Neuroinflammation, Colitis, and Gut Dysbiosis in 5xFAD Transgenic and Aged Mice by Lactobacillus mucosae and Bifidobacterium longum

Neuropsychiatric disorders including Alzheimer’s disease (AD) may cause gut inflammation and dysbiosis. Gut inflammation-suppressing probiotics alleviate neuropsychiatric disorders. Herein, to understand whether anti-inflammatory probiotics Lactobacillus mucosae NK41 and Bifidobacterium longum NK46, which suppressed tumor necrosis factor (TNF)-α expression in lipopolysaccharide (LPS)-stimulated macrophages, could alleviate cognitive impairment, we first examined their effects on cognitive function, gut inflammation, and gut microbiota composition in 5xFAD-transgenic mice. Oral administration of NK41 or NK46 decreased cognitive impairment-like behaviors, hippocampal amyloid-β (Aβ), TNF-α and interleukin (IL)-1β expression, hippocampal NF-κB+Iba1+ cell population, and Aβ accumulation, while hippocampal brain-derived neurotropic factor (BDNF) and IL-10 expression and BDNF+NeuN+ cell population increased. They also decreased TNF-α and IL-1β expression and NF-κB+CD11c+ cell population in the colon. They also reduced fecal and blood LPS levels and gut Proteobacteria and Verrucomicrobia populations (including Akkkermansiaceae), which are positively associated with hippocampal TNF-α and fecal LPS levels and negatively correlated with hippocampal BDNF level. However, they increased Odoribactericeae, which positively correlated with BDNF expression level and TNF-α to IL-10 expression ratio. The combination of NK41 and NK46 (4:1, NKc), which potently inhibited TNF-α expression in LPS-stimulated macrophages, additively alleviated cognitive impairment-like behaviors in 5xFAD-transgenic or aged mice. NKc increased hippocampal BDNF+NeuN+ cell population and BDNF expression in 5xFAD-transgenic or aged mice, while hippocampal TNF-α and IL-1β expression decreased. NKc also decreased TNF-α and IL-1β expression in the colon and LPS levels in the blood and feces. These findings suggest that gut bacteria and its product LPS may be closely connected with occurrence of cognitive impairment and neuroinflammation and the combination of NK41 and NK46 can additively alleviate cognitive impairment and neuroinflammation by inducing NF-κB-suppressed BDNF expression and suppressing LPS-producing gut bacteria.


Introduction
Alzheimer's disease (AD) is a highly prevalent, neurodegenerative, and cognitive disorder in the elderly [1,2]. Aging is a main risk factor for the development of AD [3]. The pathological hallmark of AD patients is the accumulation of insoluble amyloid-β (Aβ) and hyperphosphorylated tau in the brain, which induce neuroinflammation and cortical and hippocampal cholinergic neurotransmission disruption [4,5]. Moreover, AD is occurred and deteriorated by microbial infection, gut microbiota dysbiosis, and their byproducts such as 1 × 10 9 CFU/mouse/day of NK46; and LpC, 1 × 10 9 CFU/mouse/day of NK41 and NK46 [4:1] mix [NKc], suspended in saline) were orally gavaged once a day for 5 days and their ameliorating effects on the cognitive decline were examined. NC group was treated with saline instead of test agents (Figure 1a).
First, we randomly divided mice in five groups (NC, Lp, LpL, LpB, and LpC group consisted of 6 mice. Four groups (Lp, LpL, LpB, and LpC) were intraperit injected with LPS (10 µg/kg/day) once a day for 5 days, as previously reported [22 the next day, test agents (Lp, vehicle; LpL, 1 × 10 9 CFU/mouse/day of NK41; LpB, CFU/mouse/day of NK46; and LpC, 1 × 10 9 CFU/mouse/day of NK41 and NK46 [4 [NKc], suspended in saline) were orally gavaged once a day for 5 days and their a rating effects on the cognitive decline were examined. NC group was treated with instead of test agents (Figure 1a). Second, we randomly divided 5xFAD-transgenic mice in four groups (Tg, Tg and TgC) and young healthy control mice in one group (Nc). Each group consist mice. From the next day, test agents (Tg, vehicle; Tg41, 1 × 10 9 CFU/mouse/day of Tg46, 1 × 10 9 CFU/mouse/day of NK46; and TgM, 1 × 10 9 CFU/mouse/day of NK pended in saline) were orally gavaged once a day (6 times per week) for 8 weeks an ameliorating effects on the cognitive decline were examined. Nc group was treate saline instead of test agents (Figure 2b).
Cognitive impairment-like behaviors were measured on the day after the final administration of test agents in the Y-maze, novel object recognition, and Barnes maze tasks. Mice were euthanized in a chamber filled with CO2, followed by cervical dislocation. Brains, colons, and bloods were collected for the determination of biomarkers.
For the immunofluorescence assay, mice were trans-cardiacally perfused with 4% paraformaldehyde. The hippocampus and colon tissues were collected, post-fixed with 4% paraformaldehyde, cryoprotected in 30% sucrose solution, frozen, and sectioned using a cryostat.

Behavioral Tasks
A Y-maze task was carried out in a three-arm (120°) horizontal maze (length, 40 cm; width, 3 cm; and wall height, 12 cm) illuminated around 150 lux, as previously reported [23]. A mouse was placed on an arm and the sequence and number of arm entries were recorded for 8 min. A spontaneous alternation was defined as entries into all three arms on consecutive choices. The ratio (%) of actual to possible alternations was calculated.
A novel object recognition test was performed in an open box (length, 45 cm; width, 45 cm; and height, 45 cm) made with a black acrylic panel, which was illuminated around 30 lux [24]. In the first training experiment, a mouse was placed in the box with two identical objects (cylinder: diameter, 3.4 cm; and height, 8 cm) a in the box, and freely exposed to the objects for 10 min. The second experiment was performed for 10 min 24 h after the first experiment. A mouse was placed in a box with one old object used in the first experiment and one new object (cuboid: length, 5 cm; width, 3 cm; and height, 8 cm) and Anti-inflammatory effects of NK41, NK46, and NKc in LPS-stimulated macrophages and LPS-injected mice. Effects on TNF-α (a) and IL-10 expression (b) and their expression (TNF-α to IL-10) ratio (c) in LPS-stimulated macrophages. NK41, NK46, and NKc were treated with 1 × 10 4 or 1 × 10 6 CFU/mL). (d) Effects on spontaneous alternation in the Y-maze task. (e) Effects on the NF-κB + Iba1 + and BDNF + NeuN + cell populations in the hippocampal CA1 region. Test agents (LpL, 1 × 10 9 CFU/mice/day of NK41; LpB, 1 × 10 9 CFU/mice/day of NK46; LpC, 110 9 CFU/mice/day of NKc, NK41 and NK46 [4:1] mix) were orally administered. Normal control (Nc) and LPS-injected mice (Lp) were treated with vehicle instead of test agents. Data values were described as mean ± SD (n = 6). # p < 0.05 vs. Nc. * p < 0.05 vs. Lp.
Third, we randomly divided aged mice in two groups (Ag and AgC) and young mice in one group (Yg). Each group consisted of 6 mice. From the next day, test agents [Ag, vehicle; and AgC, 1 × 10 9 CFU/mouse/day of NK41 and NK46 (4:1) mix (NKc), suspended in saline] were orally gavaged once a day (6 times per week) for 8 weeks and examined their ameliorating effects on the cognitive decline. Yg group was treated with saline instead of test agents (Figure 1c).
Cognitive impairment-like behaviors were measured on the day after the final administration of test agents in the Y-maze, novel object recognition, and Barnes maze tasks. Mice were euthanized in a chamber filled with CO 2 , followed by cervical dislocation. Brains, colons, and bloods were collected for the determination of biomarkers.
For the immunofluorescence assay, mice were trans-cardiacally perfused with 4% paraformaldehyde. The hippocampus and colon tissues were collected, post-fixed with 4% paraformaldehyde, cryoprotected in 30% sucrose solution, frozen, and sectioned using a cryostat.

Behavioral Tasks
A Y-maze task was carried out in a three-arm (120 • ) horizontal maze (length, 40 cm; width, 3 cm; and wall height, 12 cm) illuminated around 150 lux, as previously reported [23]. A mouse was placed on an arm and the sequence and number of arm entries were recorded for 8 min. A spontaneous alternation was defined as entries into all three arms on consecutive choices. The ratio (%) of actual to possible alternations was calculated.
A novel object recognition test was performed in an open box (length, 45 cm; width, 45 cm; and height, 45 cm) made with a black acrylic panel, which was illuminated around  [24]. In the first training experiment, a mouse was placed in the box with two identical objects (cylinder: diameter, 3.4 cm; and height, 8 cm) a in the box, and freely exposed to the objects for 10 min. The second experiment was performed for 10 min 24 h after the first experiment. A mouse was placed in a box with one old object used in the first experiment and one new object (cuboid: length, 5 cm; width, 3 cm; and height, 8 cm) and recorded the number touching these objects for 10 min. Exploration (%) was indicated as the ratio percent of the sum of the new object-touching frequencies to the sum of all object-touching frequencies.
The Barnes maze task was performed in the maze consisted of a circular platform (diameter, 89 cm) with 20 holes (diameter, 5 cm) situated evenly around the perimeter and an escape box, which was located below the platform [25]. The room was illuminated around 250 lux. The training/acquisition phase was finished after a mouse entered the escape box. The test was maximally for 5 min. The mouse, which entered the box, was allowed to stay in the box for 30 s. If the mouse failed to enter the escape box within 5 min, it was led to the escape box. Mice were given two trials (for 5 min) each day for 5 consecutive days. The latency time to reach the escape hole was recorded.

Isolation and Culture of Macrophages
Macrophages were prepared, as reported previously [17]. Macrophage cells were then incubated with LPS (100 ng/mL, dissolved in saline) or vehicle in the absence or presence of probiotics (1 × 10 4 or 1 × 10 6 CFUs/mL) for 20 h. TNF-α and IL-10 levels were assayed using ELISA kits.

Immunoblotting and ELISA
Mouse brain and colon tissues were lysed with RIPA lysis buffer and centrifuged (10,000× g, 4 • C, 10 min). Cognition-related protein markers were measured by immunoblotting [26]. Levels of cytokines and myeloperoxidase in their supernatants were assayed by ELISA kits [26].

Immunohistochemical and Immunofluorescence Assay
For immunohistochemical assay, the sections of tissues (hippocampus) were washed with phosphate-buffered saline, blocked with serum, and immunostained with an antibody against Aβ42 (Invitrogen, Carbsband, CA, USA) to illuminate Aβ deposits, as previously reported [27]. For the immunofluorescence assay, the sections of tissues (hippocampus and colon) were washed with phosphate-buffered saline, blocked with serum, incubated with primary antibodies for Iba1  [25]. Cell nuclei were stained with DAPI. Immunostained sections were scanned with a confocal microscope.

Fecal Microbiota Analysis
Bacterial genomic DNA was isolated from the mouse fresh stool using a QIAamp DNA stool mini kit and amplified using barcoded primers (bacterial 16S rRNA V4 gene region) [19]. The amplicon sequencing was performed using Illumina iSeq 100 (San Diego, CA, USA). Sequenced reads were stored in the NCBI's short read archive (accession number, PRJNA872311, https://www.ncbi.nlm.nih.gov/bioproject/PRJNA872311/ accessed on 12 June 2023).

Assays of Fecal and Blood LPS Levels
The contents of LPS in the blood and feces were assayed by using the diazo-coupled limulus amoebocyte lysate (LAL) assays, as previously reported [28]. For the blood LPS assay, bloods collected by retroorbital bleeding into ethylenediaminetetraacetic acid-coated BD Microtainer ® tubes (Becton Dickinson, Franklin Lakes, NJ, USA) were centrifuged at 13,000× g for 15 min. The supernatant (5 µL) was diluted 1:10 in pyrogen-free water and inactivated for 10 min at 70 • C. For the fecal LPS assay, feces were placed in 10 mL of phosphate-buffered saline, sonicated for 15 min on ice, and centrifuged at 400× g for 10 min. Supernatants were filtrated through a 0.45 µm Millipore filter, re-filtrated through a 0.22 µm filter, and inactivated at 70 • C for 10 min. LPS contents of filtrates and supernatants (50 µL) were assayed using the LAL assay kit according to the manufacturer's protocol.

Whole Genome Analysis
The sequencing libraries were prepared according to the manufacturer's instructions of 20-kb Template Preparation Using BluePippin™ Size-Selection System using PacBio DNA Template Prep Kit 1.0 [24]. NK41 genome sequence (1 contigs) were obtained by using PacBio RSII platform and NK46 genome sequence (2 contig) was completely obtained by using PacBio Sequel platform.

Statistical Analysis
Data are indicated as mean ± standard deviation (SD) and analyzed by Graph-Pad Prism 9. The significance was analyzed by Kruskal-Wallis test with Dunn's post hoc test for non-parametric analysis (p < 0.05). The correlation between gut microbiota and spontaneous alternation, BDNF, TNF-α, or LPS level was analyzed using the Pearson correlation coefficient.
Next, to understand whether the gut microbiota endotoxin could be involved in the occurrence of systemic inflammation, we measured LPS levels in the blood and feces (Figure 5). The LPS level in the blood and feces was significantly higher in 5xFAD-transgenic mice than in healthy control mice. However, oral administration of NK41, NK46, or NKc significantly reduced LPS level in the blood and feces. (h) Effects on NF-κB + CD11c + cell population, assessed by immunofluorescence staining. Test agents (Tg, vehicle; TgL, 1 × 10 9 CFU/mouse/day of NK41; TgB, 1 × 10 9 CFU/mouse/day of NK46; TgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Normal control mice (Nc) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). # p < 0.05 vs. Nc. * p < 0.05 vs. Tg.
Next, to understand whether the gut microbiota endotoxin could be involved in the occurrence of systemic inflammation, we measured LPS levels in the blood and feces ( Figure 5). The LPS level in the blood and feces was significantly higher in 5xFAD-transgenic mice than in healthy control mice. However, oral administration of NK41, NK46, or NKc significantly reduced LPS level in the blood and feces. Figure 5. NKc decreased LPS level in the blood (a) and feces (b) of 5xFAD-transgenic mice (Tg). Test agents (Tg, vehicle; TgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Healthy control (NC) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). # p < 0.05 vs. Nc. * p < 0.05 vs. Tg.
We also investigated the effects of NK41, NK46, and NKc on the gut microbiota composition in 5xFAD-transgenic mice ( Figure 6 and Supplement Tables S1 and S2). The gut microbiota composition and β-diversity of 5xFAD-transgenic mice were significantly different to that of healthy control mice. However, α-diversity was not different. It was found that 5xFAD-transgenic mice exhibited the higher abundance of Proteobacteria, Verrucomicrobia, Cyanobacteria, and Deferibactereres (including Ruminococcaceae, Prevotellaceae, Helicobacteriaceae, and Scutterellaceae) populations and the lower abundance of Bacteroidetes (including Muribaculaceae, Lactobacillaceae, and Odoribacteriaceae) population compared to healthy control mice. However, oral administration of NK41, NK46, or NKc partially shifted the β-diversity in 5xFAD mice to that in healthy control mice, while the α-diversity was not affected. They increased the population of Bacteroidetes including Muribaculaceae and Odoribacteriaceae and decreased Proteobacteria and Verrucomicrobia populations including Ruminococcaceae, Erysipelotrichacea Prevotellaceae, Helicobacteriaceae, and Scutterellaceae. At the phylum level, Proteobacteria population was positively correlated with fecal LPS level, while the hippocampal BDNF expression level was negatively correlated. At the family level, Rhodospiraceae, Akkermaniaceae, and Scutterllaceae populations were negatively correlated with spontaneous alternation, while Clostridiaceae and Enterobacteriaceae population were positively correlated. Odoribacteriaceae and AC160630_f populations were positively correlated with hippocampal BDNF level, while Akkermansiaceae, Desulfovibrionaceae, and FR888536_f populations were negatively correlated. Blood LPS level was positively correlated with Akkermanisaceae, Bifidobacteriaceae, and Rhodospirillaceae populations, while Lachnospiraceae population was negatively correlated. Fecal LPS level was positively correlated with FR888536_f, Rhodospirillaceae, Akkermansiaceae, Deferribacteraceae populations. Hippocampal and colonic TNF-α to IL-10 expression ratios were positively correlated with Akkermansiaceae, Deferribacteriaceae, Desulfovibrionaceae, and Rhodospirillaceae populations, while Odoribacteriaceae population was negatively correlated. Test agents (Tg, vehicle; TgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Healthy control (NC) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). # p < 0.05 vs. Nc. * p < 0.05 vs. Tg.
We also investigated the effects of NK41, NK46, and NKc on the gut microbiota composition in 5xFAD-transgenic mice ( Figure 6 and Supplement Tables S1 and S2). The gut microbiota composition and β-diversity of 5xFAD-transgenic mice were significantly different to that of healthy control mice. However, α-diversity was not different. It was found that 5xFAD-transgenic mice exhibited the higher abundance of Proteobacteria, Verrucomicrobia, Cyanobacteria, and Deferibactereres (including Ruminococcaceae, Prevotellaceae, Helicobacteriaceae, and Scutterellaceae) populations and the lower abundance of Bacteroidetes (including Muribaculaceae, Lactobacillaceae, and Odoribacteriaceae) population compared to healthy control mice. However, oral administration of NK41, NK46, or NKc partially shifted the β-diversity in 5xFAD mice to that in healthy control mice, while the α-diversity was not affected. They increased the population of Bacteroidetes including Muribaculaceae and Odoribacteriaceae and decreased Proteobacteria and Verrucomicrobia populations including Ruminococcaceae, Erysipelotrichacea Prevotellaceae, Helicobacteriaceae, and Scutterellaceae. At the phylum level, Proteobacteria population was positively correlated with fecal LPS level, while the hippocampal BDNF expression level was negatively correlated. At the family level, Rhodospiraceae, Akkermaniaceae, and Scutterllaceae populations were negatively correlated with spontaneous alternation, while Clostridiaceae and Enterobacteriaceae population were positively correlated. Odoribacteriaceae and AC160630_f populations were positively correlated with hippocampal BDNF level, while Akkermansiaceae, Desulfovibrionaceae, and FR888536_f populations were negatively correlated. Blood LPS level was positively correlated with Akkermanisaceae, Bifidobacteriaceae, and Rhodospirillaceae populations, while Lachnospiraceae population was negatively correlated. Fecal LPS level was positively correlated with FR888536_f, Rhodospirillaceae, Akkermansiaceae, Deferribacteraceae populations. Hippocampal and colonic TNF-α to IL-10 expression ratios were positively correlated with Akkermansiaceae, Deferribacteriaceae, Desulfovibrionaceae, and Rhodospirillaceae populations, while Odoribacteriaceae population was negatively correlated.  We also investigated the effects of NK41, NK46, and NKc on the gut microbiota composition in 5xFAD-transgenic mice ( Figure 6 and Supplement Tables S1 and S2). The gut microbiota composition and β-diversity of 5xFAD-transgenic mice were significantly different to that of healthy control mice. However, α-diversity was not different. It was found that 5xFAD-transgenic mice exhibited the higher abundance of Proteobacteria, Verrucomicrobia, Cyanobacteria, and Deferibactereres (including Ruminococcaceae, Prevotellaceae, Helicobacteriaceae, and Scutterellaceae) populations and the lower abundance of Bacteroidetes (including Muribaculaceae, Lactobacillaceae, and Odoribacteriaceae) population compared to healthy control mice. However, oral administration of NK41, NK46, or NKc partially shifted the β-diversity in 5xFAD mice to that in healthy control mice, while the α-diversity was not affected. They increased the population of Bacteroidetes including Muribaculaceae and Odoribacteriaceae and decreased Proteobacteria and Verrucomicrobia populations including Ruminococcaceae, Erysipelotrichacea Prevotellaceae, Helicobacteriaceae, and Scutterellaceae. At the phylum level, Proteobacteria population was positively correlated with fecal LPS level, while the hippocampal BDNF expression level was negatively correlated. At the family level, Rhodospiraceae, Akkermaniaceae, and Scutterllaceae populations were negatively correlated with spontaneous alternation, while Clostridiaceae and Enterobacteriaceae population were positively correlated. Odoribacteriaceae and AC160630_f populations were positively correlated with hippocampal BDNF level, while Akkermansiaceae, Desulfovibrionaceae, and FR888536_f populations were negatively correlated. Blood LPS level was positively correlated with Akkermanisaceae, Bifidobacteriaceae, and Rhodospirillaceae populations, while Lachnospiraceae population was negatively correlated. Fecal LPS level was positively correlated with FR888536_f, Rhodospirillaceae, Akkermansiaceae, Deferribacteraceae populations. Hippocampal and colonic TNF-α to IL-10 expression ratios were positively correlated with Akkermansiaceae, Deferribacteriaceae, Desulfovibrionaceae, and Rhodospirillaceae populations, while Odoribacteriaceae population was negatively correlated.  The relationship between gut microbiota and spontaneous alternation (g), hippocampal BDNF (h), hippocampal TNF-α to IL-10 expression ratio (i), colonic TNF-α to IL-10 expression ratio (j), blood LPS (bLPS, (k)), or fecal LPS (fLPS, (l)), assessed by Pearson network analysis. Test agents (Tg, vehicle; TgL, 1 × 10 9 CFU/mouse/day of NK41; TgB, 1 × 10 9 CFU/mouse/day of NK46; TgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Normal control mice (NC) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). * p < 0.05 vs. Tg.

Effect of NKc on Gut Inflammation in Aged Mice
Aged mice exhibited increased inflammatory marker (TNF-α and myeloperoxidase) levels in aged mice. However, oral administration of NKc decreased TNF-α, IL-1β, and myeloperoxidase expression and NF-κB + CD11c + cell population and increased claudin-1 and IL-10 expression in the colon (Figure 8).
Next, we also measured LPS levels in the blood and feces (Figure 9). The LPS level in the blood and feces was significantly higher in aged mice than in young control mice. However, oral administration of NKc significantly reduced LPS level in the blood and feces. myeloperoxidase expression and NF-κB + CD11c + cell population and increased claudin-1 and IL-10 expression in the colon (Figure 8). , and TNF-α to IL-10 expression ratio (f). (g) Effects on NF-κB + CD11c + cell population assessed by immunofluorescence staining. Test agents (Ag, vehicle; AgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Young mice (Yg) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). # p < 0.05 vs. Yg. * p < 0.05 vs. Ag.
Next, we also measured LPS levels in the blood and feces ( Figure 9). The LPS level in the blood and feces was significantly higher in aged mice than in young control mice. However, oral administration of NKc significantly reduced LPS level in the blood and feces. Figure 9. NKc decreased LPS level in the blood (a) and feces (b) of aged mice (Ag). Test agents (Ag, vehicle; AgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Young mice (Yg) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). # p < 0.05 vs. Yg. * p < 0.05 vs. Ag. Figure 9. NKc decreased LPS level in the blood (a) and feces (b) of aged mice (Ag). Test agents (Ag, vehicle; AgC, 1 × 10 9 CFU/mouse/day of NKc) were orally administered. Young mice (Yg) were treated with vehicle instead of test agents. Data were described as mean ± SD (n = 6). # p < 0.05 vs. Yg. * p < 0.05 vs. Ag.

The Whole Genome Properties of NK41 and NK46
To understand the genetic properties of NK41 and NK46, we analyzed their whole genome sequences ( Figure 10). The genome of NK41 was 1,988,697 bp with a GC content of 46.7%. The total number of CDS was 1,843. The number of tRNA and rRNA genes were 98 and 21. The genome sequence of NK41 showed the phylogenetic similarity to Lactobacillus mucosae DSM13345 (96.72%), Lactobacillus buchneri subp. buchneri DSM20057 (66.7%), and Lactobacillus parakefiri JCM8573 (66.6%), using OrthoANI. The genome of NK46 was 2,513,457 bp with a GC content of 59.9%. The total number of CDS was 2,510. The number of tRNA and rRNA genes were 78 and 12. The genome sequence of NK46 showed the phylogenetic similarity to Bifidobacterium longum JCM1217 (98.4%), Bifidobacterium longum DSM20211 (96.2%), and Bifidobacterium longum ATCC15697 (94.6%), using OrthoANI.

The Whole Genome Properties of NK41 and NK46
To understand the genetic properties of NK41 and NK46, we analyzed their whole genome sequences ( Figure 10). The genome of NK41 was 1,988,697 bp with a GC content of 46.7%. The total number of CDS was 1,843. The number of tRNA and rRNA genes were 98 and 21. The genome sequence of NK41 showed the phylogenetic similarity to Lactobacillus mucosae DSM13345 (96.72%), Lactobacillus buchneri subp. buchneri DSM20057 (66.7%), and Lactobacillus parakefiri JCM8573 (66.6%), using OrthoANI. The genome of NK46 was 2,513,457 bp with a GC content of 59.9%. The total number of CDS was 2,510. The number of tRNA and rRNA genes were 78 and 12. The genome sequence of NK46 showed the phylogenetic similarity to Bifidobacterium longum JCM1217 (98.4%), Bifidobacterium longum DSM20211 (96.2%), and Bifidobacterium longum ATCC15697 (94.6%), using OrthoANI. The pseudochromosome drawn from 1 contig for NK41 and 2 contigs for NK46. The outermost circle means contig. The second inner circle is color-coded for the CDS information analyzed in the forward strand. The third inner circle is the CDS information analyzed in the reverse strand. The fourth circle from outside is tRNA (blue) and rRNA (red). The inner circle indicates GC skew metric information (green, higher than the average; red, lower than the average). The innermost circle is GC ratio metrics information (blue, higher values than average; yellow, lower values) GC skew and GC ratio metrics are displayed at 10 kb intervals. (b) Neighbor-joining tree based on the OrthoANI distance matrix (analyzed by UPGMA dendrogram, Newick format). (c) The pairwise ortholog matrix table (generated and colored according to the similarity between matching sequences). (a) The pseudochromosome drawn from 1 contig for NK41 and 2 contigs for NK46. The outermost circle means contig. The second inner circle is color-coded for the CDS information analyzed in the forward strand. The third inner circle is the CDS information analyzed in the reverse strand. The fourth circle from outside is tRNA (blue) and rRNA (red). The inner circle indicates GC skew metric information (green, higher than the average; red, lower than the average). The innermost circle is GC ratio metrics information (blue, higher values than average; yellow, lower values) GC skew and GC ratio metrics are displayed at 10 kb intervals. (b) Neighbor-joining tree based on the OrthoANI distance matrix (analyzed by UPGMA dendrogram, Newick format). (c) The pairwise ortholog matrix table (generated and colored according to the similarity between matching sequences).

Discussion
Aging-associated low-grade chronic inflammation as well as stress-induced gut inflammation and dysbiosis contribute to the development of AD [29,30]. In particular, Bifidobacteria and Lactobacilli populations were lower in the feces of elderly and aged mice than in young individuals and mice, respectively [31][32][33]. However, elderly and aged mice exhibited the higher Enterobacteriaceae population and bacterial LPS levels, which are closely connected with psychiatric disorders such as AD [32,34]. The gut microbiota of 5xFAD-transgenic mice, the well-known AD mouse model, also produce LPS more strongly than those of normal control mice [9]. In the present study, we also found that the LPS levels of 5xFAD-transgenic or aged mouse feces were higher than those of normal control mice. TNF-α and IL-1β expression and NF-κB-positive cell population were significantly higher in the hippocampus and colon of 5xFAD-transgenic or aged mice, while BDNF expression and BDNF-positive neural cell population were lower. In addition, Kim et al. reported that excessive expression of gut bacterial LPS caused gastrointestinal inflammation via the TLR4-linked NF-κB signaling [35]. Jang et al. reported that intraperitoneal injection of LPS caused cognitive impairment with neuroinflammation through the suppression of NF-κB activation-mediated BDNF expression [26]. Ma et al. reported that chronic exposure to gut microbiota dysbiosis-induced bacterial LPS could trigger the occurrence of neurodegenerative disorders such as AD [36]. Therefore, gut microbiota LPS is a risk factor for AD. Improving gut microbiota dysbiosis and suppressing gut bacterial LPS production may be beneficial for the therapy of cognitive decline in patients with AD and elderly.
In the present study, we found that NK41 and NK46 inhibited LPS-induced TNF-α expression in macrophages and NF-κB-positive cell population in mice. They alleviated LPS-induced cognitive impairment in the Y-maze test. They also suppressed TNF-α and IL-1β expression and NF-κB-positive cell population in the hippocampus and colon of 5xFAD-transgenic mice. They also suppressed Aβ and its related BACE expression and Aβ accumulation in the brain of 5xFAD-transgenic mice. However, they increased BDNF expression and BDBF + NeuN + cell population in the hippocampus. Decourt et al. reported that TNF-α induced Aβ accumulation in the brain rodents [37]. Kim et al. reported that NK41 suppressed TNF-α expression and increased BDNF expression in the hippocampus of mice with Escherichia coli K1-induced cognitive function [19]. Lee et al. reported that NK46 suppressed TNF-α expression and increased BDNF expression in the hippocampus of 5xFAD-transgenic mice [9]. These results suggest that anti-inflammatory probiotics NK41 and NK46 may decrease TNF-α and Aβ expression through the suppression of NF-κB activation, resulting in the increase in BDNF expression and BDNF-positive cell population in the brain. We also found that NK41 suppressed TNF-α, IL-1β, and p16 expression and increased BDNF expression and BDNF-positive cell population in hippocampus of aged mice (Supplement Figure S1). Lee et al. reported that NK46 suppressed TNF-α, IL-1β, and p16 expression in the hippocampus of aged mice. These results suggest that NK41 and NK46 can suppress TNF-α and p16 expression through the inhibition of NF-κB activation, resulting in the increase in BDNF expression and BDNF-positive cell population in the brain. We also found that NK41 and NK46 alleviated cognitive impairment-like behaviors in 5xFAD-transgenic or aged mice. The NF-κB activation in immune cells stimulates inflammatory response in the brain and suppresses BDNF expression in neuron cells of brain [38,39]. BDNF induces long-term potentiation in neuron cells, leading to the increase in the cognitive function [40,41]. These results suggest that NK41 and NK46 may increase cognitive decline through the induction of NF-κB-suppressed BDNF expression.
NKc, the combination of NK41 with NK46, also suppressed TNF-α, IL-1β, and Aβ expression, Aβ accumulation, and NF-κB-positive cell population in the hippocampus of 5xFAD-trangenic mice. NKc also suppressed TNF-α, IL-1β, and p16 expression and NF-κBpositive cell population in the hippocampus of aged mice. NKc also suppressed TNF-α and IL-1β expression and NF-κB-positive cell population in the colon of 5xFAD-transgenic or aged mice. Although the effect of NKc was higher than that of NK41 or NK46 alone, it was not different between them. These results suggest that the combination of N41 with NK46 may additively alleviate cognitive impairment-like symptoms.
We also found that 5xFAD-transgenic mice exhibited a higher abundance of gut Proteobacteria and Verrucomicrobiota populations compared to control mice. Fecal and blood LPS levels were higher in 5XFAD-transgenic or aged mice than in control mice. Lee et al. reported that Proteobacteria population and LPS levels were significantly higher in aged mice than in young mice [9]. Yun et al. also reported that the gut microbiota of 5xFADtransgenic mice strongly increased Proteobacteria population and LPS production [42]. However, we found that oral administration of NK41, NK46, and NKc reduced LPS levels in the blood and feces of 5xFAD-transgenic or aged mice. They reduced the population of Proteobacteria and Verrucomicrobia, which were positively correlated with the fecal or blood LPS level and negatively correlated with hippocampal BDNF expression level, in the gut microbiota of 5xFAD-transgenic mice. They also suppressed Akkermansiaceae, Sutterellaceae, and Desulfovibrionaceae populations, which were positively correlated with the fecal and blood LPS levels and hippocampal and fecal TNF-α to IL-10 expression ratios and negatively correlated with BDNF expression and spontaneous alternation. However, NK41, NK46, and NKc increased Odoribacteriaceae population, which was negatively correlated with fecal and blood LPS levels and hippocampal and fecal TNF-α to IL-10 expression ratios and positively correlated with BDNF expression and spontaneous alternation. These results suggest that NK41, NK46, and their mix NKc can alleviate systemic inflammation including neuroinflammation and colitis with gut dysbiosis by suppressing gut microbiota LPS production through the regulation of gut microbiota, resulting in the alleviation of cognitive impairment. Moreover, altough the genome sequences of NK41 and NK46 showed the phylogenetic similarity to Lactobacillus mucosae DSM13345 (96.7%) and Bifidobacterium longum JCM1217 (98.4%), respectively, their genome sequences were unique. Their cognitive-impairment-ameliorating effects may be due to their unique characteristics.

Conclusions
Probiotics NK41 and NK46 alleviated cognitive decline and neuroinflammation in 5xFAD-transgenic, aged, LPS-stimulated mice. NK41 and NK46 induced BDNF and NM-DAR expression in the brain. They also alleviated colitis and gut microbiota dysbiosis. The combination of NK41 with NK46 additively alleviated cognitive decline, neuroinflammation, and colitis by up-regulation of BDNF signaling and down-regulation of NF-κB signaling and gut microbiota dysbiosis.

Data Availability Statement:
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.