Lactobacillus gasseri NK109 and Its Supplement Alleviate Cognitive Impairment in Mice by Modulating NF-κB Activation, BDNF Expression, and Gut Microbiota Composition

De Man, Rogosa, and Sharpe (MRS) medium was purchased from BD (Franklin Lakes, NJ, USA). A soybean embryo ethanol extract powder (SE) was purchased from Mirae Biotech (Pocheonshi, Gyunggi-do, Republic of Korea).

NK109 was cultured in commercial media for probiotics, including MRS broth at 37 °C for 10 h, centrifuged, and freeze-dried [19]. NK109-contained supplement NS was prepared by mixing freeze-dried NK109 (0.5 × 10¹⁰ colony-forming unit [CFU]) with SE (10 mg), which was dry-heated at 60 °C for 30 min with shaking. NK109 and NS were suspended in saline for in vivo experiment.

Healthy volunteers (62–70 yrs old, 74–77 in CERAD-K, n = 3 [male, 1; female, 2]) and volunteers with MCI (62–70 yrs old, 52–63 in CERAD-K, n = 3 [male, 1; female, 2]) were recruited from Kyung Hee University Hospital (Seoul, Republic of Korea) (Supplementary Table S1). All volunteers enrolled for the stool collection are described in detail in Supplement File S1.

Ag C57BL/6 mice (male, 18-month-old), Tg C57BL/6 mice ((male, 4-month-old), adult C57BL/6 mice (male, 4-month-old), and young (Yg) C57BL/6 mice (male, 6-week-old) were purchased from RaonBio Co., Ltd. (Yongin-Shi, Republic of Korea). Mice were maintained in a controlled room and fed with water and food ad libitum. Animals were acclimatized for 7 days and used in the experiments. All animal experiments were ethically approved by the Committee for the Care and Use of Laboratory Animals in the University (IACUC, KHUASP(SE)-22129, 22236, and 22359) and were performed according to the Ethical Policies and Guidelines of Kyung Hee University for Laboratory Animals Care and Use.

First, mice were randomly divided in three groups of Ag mice (Ag, AgN, and AgS) and one group of Yg mice (Yg), consisting of six mice per group. Test agents (Yg, saline; Ag, saline; AgN, 1 × 10⁹ CFU NK109 per mouse; and AgS, NS [0.5 × 10⁹ CFU NK plus 1 mg SE] per mouse) were orally gavaged once a day (6 times per week) for 8 weeks.

Second, mice were randomly divided in three groups of Tg mice (Ag, AgN, and AgS) and one group of adult mice (Ad), consisting of seven mice per group. Test agents (Ad, saline; Tg, saline; TgN, 1 × 10⁹ CFU NK109 per mouse; and TgS, NS [0.5 × 10⁹ CFU NK plus 1 mg SE] per mouse) were orally gavaged once a day (6 times per week) for 8 weeks.

Third, mice were randomly divided in three groups of MCF-transplanted mice (Fm, FmN, and FmS), one group of healthy volunteer fecal microbiota (HF)-transplanted one group (Fh), and one group of Yg mice (Yg), consisting of seven mice per group. Cognition-impaired mice were prepared by oral gavage of MCF once a day for 5 days. Test agents (Yg, saline; Fh, saline; FmN, 1 × 10⁹ CFU NK109 per mouse; and FmS, NS [0.5 × 10⁹ CFU NK plus 1 mg SE] per mouse) were orally gavaged once a day for 5 days after the final transplantation of MCF or HF.

Cognitive function was measured in the Y-maze task (YMT), novel object recognition task (NORT), or Barnes maze task (BMT) the next day after the final gavage of test agents. Mice were euthanized in a CO₂ chamber, followed by cervical dislocation.

YMT, NORT, and BMT were performed in a three-arm horizontal maze (40 cm long and 3 cm wide with 12-cm-high walls), open field box (45 × 45 × 45 cm), and circular platform (89 cm, diameter) with 20 holes (5 cm, diameter)/one escape hole box, respectively, as previously reported [19,25]. The detailed protocols are described in Supplement.

The brain and colon tissues were lysed in radio immunoprecipitation assay lysis buffer (Pierce, Rockford, IL, USA) and centrifuged (10,000× g, 4 °C, 10 min). In the supernatant, BDNF, p16, p65, p-p65, CREB, p-CREB, Presenilin (Psen)-1, amyloid-β (Aβ), claudin-5, claudin-1, and β-action expression levels were assayed by immunoblotting [15]. IL-1β, IL-10, and TNF-α levels were determined using commercial ELISA kits (Ebioscience, Waltham, MA, USA) [15]. The detailed protocols are described in Supplement File S1.

Transcardially perfused brain and colon tissues were sectioned, as previously reported [23]. The sections were immunostained with primary antibodies against NeuN, BDNF, NF-κB, LPS, Iba1, and/or CD11c overnight and secondary antibodies conjugated with Alexa Fluor 594 (1:200, Invitrogen, Waltham, MA, USA) or Alexa Fluor 488 for 2 h. Immunostained sections were observed using a confocal microscope.

Bacterial genomic DNAs were extracted from the fresh stools of volunteers using a QIAamp DNA stool mini kit (Qiagen, Hielden, Germany) and amplified using barcoded primers (bacterial 16S rRNA V4 gene region 16S rRNA genes) [26]. The amplicon sequencing was performed using Illumina iSeq 100. Sequenced reads were deposited in the short read archive of NCBI (accession number PRJNA 915132).

qPCR for NK109 was performed on the Rotor-Gene Q^® thermocycler using DNA polymerase and SYBR Green I (Takara Bio Inc., Kusatsu, Japan), as previously reported [27].

Experimental data are indicated as mean ± SD using GraphPad Prism 9 (GraphPad Software, Inc.). The significant difference was analyzed using one-way ANOVA followed by Duncan’s multiple range test (p < 0.05). The β-diversity (principal coordinate analysis [PCoA] plot) was indicated on the basis of generalized-unifrac as distance metric. The network of differentially enriched gut microbiota in spontaneous alteration in YMT or exploration in NORT scores was indicated on the basis of Pearson correlation coefficient (−1 ~ −0.22/0.22 ~ 1).

First, we examined whether the MCF microbiota transplantation could cause cognitive impairment in mice (Figure 1). Orally gavaged MCF microbiota impaired cognitive behaviors in the YMT, NORT, and BMT to 76.9% (F_4,30 = 8.21, p < 0.001), 29.8% (F_4,30 = 2.99, p < 0.001) and 374.0% (F_4,30 = 18.27, p < 0.001) of control mice, respectively, while orally gavaged HF microbiota were not affected (Figure 1A–C). However, the average of the arm entry numbers in the Y-maze task were not significantly different between all tested groups, suggesting that MCF treatment did not affect general locomotor activity, as described in the experiment of Sarter et al. [28]. To understand whether MCF could affect the expression of cognition-associated neuroinflammation makers TNF-α and IL-1β and cognitive function-inducing biomarkers BDNF [8,11], these biomarkers were assayed in the hippocampus using ELISA kits. MCF transplantation increased TNF-α and IL-1β expression in the hippocampus, while BDNF expression was decreased (Figure 1D–G). Next, we investigated the effects of NK109 and NS on MCF-impaired CI in mice (Figure 1A–C). Oral administration of NK109 and its supplement NS increased spontaneous alteration in the YMT to 94.3% and 94.9% of control mice, respectively, and exploration in the NORT to 47.7% and 50.0% of control mice, respectively, and decreased latency time in the BMT to 132.3% and 161.7% of control mice, respectively. Furthermore, their treatments decreased IL-1β and TNF-α expression, NF-κB⁺Iba1⁺ (activated microglial) cell numbers in the hippocampus, while BDNF, claudin-5, and IL-10 expression and BDNF⁺NeuN⁺ cell numbers were increased (Figure 1D–I). In the immunoblotting analysis, they also decreased p-p65/p65 expression in the hippocampus, while p-CREB/CREB and claudin-5 expression increased (Figure 1J).

MCF transplantation shortened colon length and increased IL-1β, TNF-α, and myeloperoxidase expression and NF-κB⁺CD11c⁺ cell numbers in the colon, while IL-10 and claudin-1 expression decreased (Figure 2A–F)). However, treatment with NK109 or its supplement NS down-regulated MCF-induced myeloperoxidase, TNF-α, and IL-1β expression and up-regulated claudin-1 and IL-10 expression.

We also investigated the effect of NK109 and its supplement NS on cognitive function in aged mice (Figure 3). Ag mice showed significant deficits in learning and memory function compared to Yg mice: they decreased spontaneous alteration in the YMT and exploration in the NORT to 54.5% (F_4,30 = 26.97, p < 0.001) and 58.0% (F_4,30 = 13.02, p < 0.001) of Yg mice, respectively, and increased latency time to 247.5% (F_4,30 = 25.32, p < 0.001) of Yg mice (Figure 3A–C). However, oral administration of NK109 and NS improved spontaneous alteration in Ag mice to 95.7% and 84.3% of Yg mice, respectively, exploration in Ag mice to 80.7% and 77.5% of Yg mice, respectively, and latency time in Ag mice to 180.1% and 127.7% of Yg mice, respectively. The cognition-improving potency between NK109 and its supplement NS was not significantly different. The expression of p16, TNF-α, and IL-1β expression and population of NF-κB⁺Iba1⁺ cell (activated microglia) population were higher in Ag mice than in Yg mice, while claudin-5 expression and BDNF⁺NeuN⁺ cell population increased (Figure 3D–I). In the immunoblotting analysis, they increased p-CREB/CREB and claudin-5 expression, while p-p65/p65 and p16 expression were decreased (Figure 3J).

In Ag mice, colon shortening and colonic myeloperoxidase, TNF-α, and IL-1β expression and NF-κB⁺CD11c⁺ cell numbers were increased more strongly than in young control mice, while IL-10 and claudin-1 expression decreased (Figure 4A–G). However, treatment with NK109 or NS down-regulated myeloperoxidase, TNF-α, and IL-1β expression and up-regulated IL-10 and claudin-1 expression in the colon of Ag mice.

We compared the fecal microbiota composition of aged mice with that of Yg mice (Figure 5, Supplementary Tables S2–S4). Ag mice exhibited a higher abundance of Proteobacteria and Actinobacteria populations compared to those in Yg mice (Figure 5A,B). However, the β-diversity between them was significantly different, while the α-diversity was not (Figure 5C,D). However, oral administration of NK109, but not NS, increased α-diversity in Ag mice (Figure 5C). Treatment with NS increased α-diversity in Yg mice. Nevertheless, treatment with NK109 or NS partially shifted β-diversity in Ag mice to that in Yg mice (Figure 5D). At the phylum level, treatment with NK109 or NS increased Tenericutes, Cyanobacteria, and Deferribacteres populations in Ag mice, while Actinobacteria population decreased in NK109-treated mice alone (Figure 5A, Supplementary Tables S3–S5). At the family level, treatment with NK109 or NS decreased Lactobacillaceae and Bifidobacteriaceae populations, while Rikenellaceae, Prevotellaceae, and Deferribacteraceae populations increased. In particular, treatment with NK109 alone decreased Helibacteriaceae and Odorobacteriaceae populations, while Porphyromonacaceae, FR888536_f populations increased (Figure 5B). In the network analysis, Bifidobacteriaceae, Lactbacillaceae, Bacteroidaceae, and Erysipelotrichaceae populations were negatively associated with spontaneous alteration in the YMT or exploration in the NORT, while Lachnospiraceae and Rikenellaceae populations were positively associated (Figure 5E, Supplementary Tables S6 and S7). Moreover, NK109 or NK treatment increased NK109 population in the feces of Ag mice (Figure 5E).

We also investigated the effects of NK109 and NS on CI in Tg mice (Figure 6). Tg mice exhibited significant deficits in learning and memory function compared with control mice: they decreased spontaneous alteration and exploration to 61.5% (F_3,20 = 18.49, p < 0.001) and 38.5% (F_3,20 = 35.47, p < 0.001) of control mice, respectively, and increased latency time to 486.7% (F_3,20 = 15.50, p < 0.001) of control mice (Figure 6A–C). However, oral administration of NK109 and NS improved spontaneous alteration in Tg mice to 80.5% and 89.4% of control mice, respectively, exploration in Tg mice to 58.0% and 62.0% of control mice, respectively, and latency time in Tg mice to 230.1% and 173.0% of control mice, respectively. The cognition-improving potency between NK109 and NS was not significantly different. Tg mice exhibited increased p-p65/p65, Psen-1, Aβ, TNF-α, and IL-1β expression and NF-κB⁺Iba1⁺ cell (activated microglia) population compared with those in the control, while p-CREB/CREB, claudin-5, and IL-10 expression and BDNF⁺NeuN⁺ cell population increased (Figure 6D–J). In the immunoblotting analysis, they increased p-CREB/CREB and claudin-5 expression, while Aβ, Psen-1, and p-p65/p65 expression were decreased (Figure 6K).

In Tg mice, colon shortening and colonic myeloperoxidase, TNF-α, and IL-1β expression and NF-κB+CD11c+ cell number increased more strongly than in control mice, while IL-10 and claudin-1 expression decreased (Figure 7A–F). However, oral administration of NK109 or NS decreased myeloperoxidase, TNF-α, and IL-1β expression and NF-κB-postive cell populations in the colon of Tg mice, while claudin-1 and IL-10 expression increased.

Next, we tested the effects of NK109 and NS on the gut microbiota composition in Tg mice (Figure 8, supplement Tables S7–S9). Tg mice exhibited a higher abundance of Proteobacteria and Tenericutes populations than the control mice (Figure 8A,B). Moreover, the β-diversity of Tg mouse fecal microbiota was different to those of the control mice, while the α-diversity was not different (Figure 8C,D). However, oral administration of NK109 or NS weakly shifted the β-diversity in Tg mice to that in control mice, while the α-diversity was not affected (Figure 8C,D). At the phylum level, treatment with NK109 or NS decreased Proteobacteria and Tenericutes populations in Tg mice, while Actinobacteria populations increased (Figure 8A, Supplementary Tables S8–S10). At the family level, NK109 or NS treatment increased Muribaculaceae and Rhodospiraceae populations, while Helicobacteriaceae, Porphyromonadaceae, and Mycoplasmataceae populations decreased (Figure 8B). In the network analysis, Muribaculaceae, Sutterellacease, and Coriobacteriaceae populations were positively associated with spontaneous alteration in the YMT or exploration in the NORT, while Gemella_f, Saccharimonas_f, and Christensenllaceae populations were negatively associated (Figure 8E, Supplementary Tables S11 and S12). Moreover, NK109 or NK treatment increased NK109 populations in the feces of Ag mice (Figure 8F).

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Microbiota data are deposited in NCBI (accession number PRJNA 915132).