Appropriate exercise level attenuates gut dysbiosis and valeric acid increase to improve neuroplasticity and cognitive function after surgery in mice

Abstract

Postoperative cognitive dysfunction (POCD) affects the outcome of millions of patients each year. Aging is a risk factor for POCD. Here, we showed that surgery induced learning and memory dysfunction in adult mice. Transplantation of feces from surgery mice but not from control mice led to learning and memory impairment in non-surgery mice. Low intensity exercise improved learning and memory in surgery mice. Exercise attenuated surgery-induced neuroinflammation and decrease of gut microbiota diversity. These exercise effects were present in non-exercise mice receiving feces from exercise mice. Exercise reduced valeric acid, a gut microbiota product, in the blood. Valeric acid worsened neuroinflammation, learning and memory in exercise mice with surgery. The downstream effects of exercise included attenuating growth factor decrease, maintaining astrocytes in the A2 phenotypical form possibly via decreasing C3 signaling and improving neuroplasticity. Similar to these results from adult mice, exercise attenuated learning and memory impairment in old mice with surgery. Old mice receiving feces from old exercise mice had better learning and memory than those receiving control old mouse feces. Surgery increased blood valeric acid. Valeric acid blocked exercise effects on learning and memory in old surgery mice. Exercise stabilized gut microbiota, reduced neuroinflammation, attenuated growth factor decrease and preserved neuroplasticity in old mice with surgery. These results provide direct evidence that gut microbiota alteration contributes to POCD development. Valeric acid is a mediator for this effect and a potential target for brain health. Low intensity exercise stabilizes gut microbiota in the presence of insult, such as surgery.

Fig. 1. Exercise improved learning and memory.

Nine-week old male mice with or without exercise for 4 weeks were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia (panels A–C). In another experiment, 9-week old mice were treated with antibiotics to eliminate their native gut microbiota and then transplanted with feces from exercise mice (Trans-Exe) or control mice (Trans-Control) 2 weeks before the surgery (panels D–F). A, D Training sessions of Barnes maze test. B, E Memory assessment of Barnes maze test. C, F Novel object recognition test. Results are mean ± S.D. with (panels E and F) or without (panels A and D) presentation of value of individual mouse or median ± interquartile range with presentation of value of individual mouse (panels B and C) (n = 17–20 for panels A–C, = 13–15 for panels D–F). Exe-I/Exercise-I low intensity exercise, Exe-m middle intensity exercise, Exe-h high intensity exercise, Sur surgery. *P < 0.05 compared the two curves.

Fig. 2. Exercise stabilized gut microbiota of mice with surgery.

Nine-week old male mice with or without exercise for 4 weeks were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia. In another experiment, 9-week old mice were treated with antibiotics to eliminate their native gut microbiota and then transplanted with feces from exercise mice (ET) or control mice (CT). Presentation of α diversity is in (panels A, C, E and G). Presentation of β diversity is in (panels B, D, F and H). n = 16 for (panels A and B, and = 8 for other panels). C-S: control mouse samples harvested before surgery, E-ES: exercise mouse samples harvested before surgery, S: surgery mouse samples harvested just before surgery, SP3: surgery mouse samples harvested on post-surgery day 3, SP7: surgery mouse samples harvested on post-surgery day 7, ES: exercise mouse samples harvested just before surgery, ESP3: exercise mouse samples harvested on post-surgery day 3, ESP7: exercise mouse samples harvested on post-surgery day 7, CT: samples harvested from mice transplanted with feces from control mice, ET: samples harvested from mice transplanted with feces from exercise mice.

Fig. 3. mRNA expression profile of mice.

Hippocampal samples were subjected to RNA-seq analysis. A, B Volcano plot. C Heatmap of mRNA abundance of genes whose expression was different among the three groups of animals. D–F Quantitative data of real-time PCR analysis. Results in (panels D–F) are mean ± S.D. with presentation of value of individual mouse (n = 5 for panels A–C, and = 3 for panels D–F). C control, S or Sur surgery, ES or Exe+Sur exercise plus surgery, Trans-Control: mice transplanted with feces from control mice, Trans-Exe: mice transplanted with feces from exercise mice, NS normal saline, Val valeric acid.

Fig. 4. Exercise attenuated surgery-induced immune and inflammatory responses.

Hippocampus was harvested at various times after surgery for immunostaining or ELISA. A Representative Iba-1 and C3ar immunostaining images of hippocampus harvested 48 h after surgery. B Quantification of Iba-1 and C3ar immunostaining of hippocampus harvested 48 h after surgery. C Representative C3 immunostaining images of hippocampus harvested 48 h after surgery. D Quantification of C3 by ELISA. E IL-1β quantified by ELISA. F IL-6 quantified by ELISA. Results in (panels B, E and F) are mean ± S.D. with presentation of value of individual mouse and result in (panels D) is median ± interquartile range with presentation of value of individual mouse (n = 6 for panel B,  = 14 for panel D,  = 9 for panels E and F). Exe exercise, Sur surgery, Post post-surgery.

Fig. 5. Exercise attenuated surgery-induced decrease of brain cell genesis and dendritic arborization in young adult mice.

Brain was harvested 19 days after surgery for immunostaining or Golgi staining. A Representative GFAP and BrdU immunostaining images of hippocampus. B Quantification of GFAP and BrdU positively stained cells in the hippocampus. C Representative Golgi staining images of hippocampus. D Quantification of intersections among the dendritic branches and spine density in the hippocampus. Results in (panels B and D) are mean ± S.D. with presentation of value of individual mouse (n = 6 for panel B,  = 8 for panel D). Exe exercise, Sur surgery.

Fig. 6. Exercise attenuated surgery-induced changes in short chain fatty acids (SCFAs).

Blood was harvested either 4 weeks after the onset of exercise protocol (for panels A, C, E and F) or 7 days after surgery (panels B and D) from 13 and 14-week old (young mice) or 19-month old mice (old mice). A Blood SCFAs just before the surgery in young mice. B Blood SCFAs 7 days after the surgery in young mice. C Blood SCFAs just before the surgery in old mice. D Blood SCFAs 7 days after the surgery in old mice. E Correlation presentation between gut bacteria and blood SCFA concentrations in young adult mice. F Correlation presentation between gut bacteria and fecal SCFA concentrations in young adult mice. Results in (panels A–D) are median ± interquartile range (for propionic acid, butyric acid, valeric acid and hexanic acid) with presentation of value of individual mouse or mean ± S.D. (for other SCFAs) with presentation of value of individual mouse (n = 7 for all panels). Exe exercise, Sur surgery. In (panels E and F), *P < 0.05 for the correlation, red: positive correlation, blue: negative correlation.

Fig. 7. Role of valeric acid and C3 signaling in exercise attenuation on surgery-induced learning and memory impairment.

Nine-week old male mice with or without exercise for 4 weeks in the presence or absence of intraperitoneal injection of valeric acid (one injection per week) were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia (panels A–C). In another experiment, 9-week old mice with or without exercise for 4 weeks were subjected to surgery and received intracerebroventricular injection of a C3 agonist or SB290157, a C3ar antagonist, at 0, 24, 48 and 72 h after surgery (panels D–F). A, D Training sessions of Barnes maze test. B, E Memory assessment of Barnes maze test. C, F Novel object recognition test. Results are mean ± S.D. in (panels A and D) and median ± interquartile range with presentation of value of individual mouse in panels B, C, E and F (n = 11–13 for panels A–C,  = 13–15 for panels D–F). Exe exercise, Sur surgery, NS normal saline, DMSO dimethylsulfoxide, Val valeric acid, Anta-C3ar C3ar antagonist, Agon-C3ar C3ar agonist. *P < 0.05 compared the two curves.

Fig. 8. Exercise via regulating gut microbiota attenuated surgery-induced GDNF decrease and GDNF participated in regulation of immune and inflammatory responses after surgery.

Nine-week old male mice with or without exercise for 4 weeks in the presence or absence of intraperitoneal injection of valeric acid (one injection per week) were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia (panels A and D). In second experiment, 9-week old mice were treated with antibiotics to eliminate their native gut microbiota and then transplanted with feces from control mice (Trans-Control) before the surgery (panel B). In third experiment, 9-week old mice were treated with antibiotics to eliminate their native gut microbiota and then transplanted with feces of exercise mice (Trans-Exe) or control mice (Trans-Control) before the surgery (panel C). In fourth experiment, 9-week old mice received intraperitoneal injection of valeric acid or normal saline (panels E and F). In fifth experiments, 9-week old mice were subjected to surgery and received intracerebroventricular injection of GDNF immediately after surgery (panels G–I). A–D GDNF concentrations in the hippocampus. E Valeric acid in the cerebral cortex. F, G C3 concentrations in the hippocampus. H IL−6 concentrations in the hippocampus. i IL-1β concentrations in the hippocampus. Results in (panels A, B, D and G) are median ± interquartile range with presentation of value of individual mouse and results in other panels are mean ± S.D. with presentation of value of individual mouse (n = 10 for panels A–D and panel F, = 7–8 for panel E, = 12 for panels G–I). Exe exercise, Sur surgery, NS normal saline, Val valeric acid.

Fig. 9. Exercise via regulating gut microbiota and blood valeric acid attenuated surgery-induced learning and memory impairment in old mice.

Nineteen-month old male mice with or without exercise for 4 weeks were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia (panels A–D). In second experiment, 19-month old mice were treated with antibiotics to eliminate their native gut microbiota and then transplanted with feces from exercise mice (Trans-Exe) or control mice (Trans-Control) before the surgery (panels E–G). In third experiments, 19-month old male mice with or without exercise for 4 weeks in the presence or absence of intraperitoneal injection of valeric acid (one injection per week) were subjected to surgery (panels H–J). A, E, H Training sessions of Barnes maze test. B, F, I Memory assessment of Barnes maze test. C, D, G, J Novel object recognition test. Results are mean ± S.D. with (panels B and G) or without (panels A, E and H) presentation of value of individual mouse or median ± interquartile range (panels F, I, and J) with presentation of value of individual mouse in other panels (n = 11–12 for panels A–D, = 9–10 for panels E–G, = 9–11 for panels H–J). Exe exercise, Sur surgery, NS normal saline, Val valeric acid. *P < 0.05 compared the two curves.

Fig. 10. Exercise stabilized gut microbiota of mice with surgery in old mice.

Nineteen-month old male mice with or without exercise for 4 weeks were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia. Presentation of α diversity is in (panels A, C, E and G). Presentation of β diversity is in (panels B, D, F and H) (n = 16 for panels A and B, and = 8 for other panels). OC-OS old control mouse samples harvested before surgery, OE-OES old exercise mouse samples harvested before surgery, OS old surgery mouse samples harvested just before surgery, OSP old surgery mouse samples harvested on post-surgery day 7, OES old exercise mouse samples harvested before surgery, OESP old exercise mouse samples harvested on post-surgery day 7.

Fig. 11. Exercise attenuated surgery-induced GDNF decrease, immune and inflammatory responses and dendritic arborization impairment in old mice.

Nineteen-month old male mice with or without exercise for 4 weeks were subjected to left carotid artery exposure (surgery) under isoflurane anesthesia. A GDNF expression. B C3 expression. C Representative Iba-1 and C3ar immunostaining images. D Quantitative data of Iba-1 and C3ar immunostaining. E IL-1β and IL-6 expression. F Representative GFAP and BrdU immunostaining images. G Quantitative data of GFAP and BrdU immunostaining. H Representative Golgi staining. I Quantitative data of intersections among dendritic branches and spine density. Results in panels A, D and E are median ± interquartile range with presentation of value of individual mouse and results in panels B, G and I are mean ± S.D. with presentation of value of individual mouse (n = 10 for panels A, = 12 for panel B, = 6 for panels D and G, = 8 for panels E and I). Exe exercise, Sur surgery.

Fig. 12. Diagrammatic presentation of the findings from this study.

Surgery may induce gut dysbiosis to lead to increased valeric acid in the blood, which then activates complement 3 signaling to result in postoperative cognitive dysfunction. Low intensity exercise before surgery inhibits this detrimental pathway to reduce postoperative cognitive dysfunction. GDNF glial cell-derived neurotrophic factor, C3ar complement 3a receptor, POCD postoperative cognitive dysfunction.