Effect of Electro-Acupuncture on Gut Microbiome in Cisplatin-Induced Premature Ovarian Failure Mice

Background Growing evidences showed that gut microbiota is associated with premature ovarian failure (POF). Many case reports had shown that electro-acupuncture was effective in the treatment of POF. However, there was little research on regulating gut microbiome of POF mice by electro-acupuncture. Therefore, this study aimed to verify whether electro-acupuncture could improve ovarian function and regulate gut microbiome in mice with POF. In addition, the therapeutic effects of electro-acupuncture at acupoints and non-acupoints were compared.


Electro-acupuncture treatment
According to the "Experimental acupuncturolog", Guanyuan (CV4), Sanyinjiao (SP6) and non-acupoints (5 mm horizontally beside CV4 and 3 mm vertically above SP6) were selected. The non-acupoints did not belong to any meridians. Subsequently, all acupoints and non-acupoints were disinfected by 75% alcohol before electro-acupuncture. The 0.25mm×13mm acupuncture needles (Suzhou Medical Appliance Factory, JiangSu, China) were be selected. The depth of acupuncture was about 3-5mm. The current output mode of the electro-acupuncture instrument (Model SDZ-II; Suzhou Medical Appliance Factory, Jiangsu, China) was alternate of sparse wave and dense wave (intermittent wave: 4 Hz; irregular wave: 50 Hz). The POF mice in EA group and EN group were treated by electro-acupuncture 30 min daily for 3 weeks (Fig. 1A and B).
Afterwards, all mice were sacri ced by enucleated eyes at the end of electro-acupuncture treatment.
Observation of estrous cycle 20µL of 0.9%NaCl solution was infused into the vagina of mice by the pipette gun and pumped gently several times. Then, the solution was sucked out onto the glass slide and stained. Vaginal exfoliated cells were observed under the light microscope. The above steps were repeated daily during the period of experiment.

Histopathology examination
Ovaries were collected and dehydrated in 10% paraformaldehyde for 48 hours after mice were sacri ced, and embedded with para n after ethanol dehydration. Subsequently, ovaries were sliced to a thickness of 5mm, and the morphology of ovary was observed under light microscope after stained.
Enzyme Linked Immunosorbent Assay (ELISA) Blood samples were collected via enucleated eyes. Blood samples were centrifugated (10000rmp 10 min 4℃) to obtain the serum, and the FSH, E 2 , luteinizing hormone (LH), anti mullerian hormone (AMH) and β-glucuronidase were detected by ELISA kit. All experimental processes were according to the protocol provided by the manufacturer.

Quantitative Real-time PCR (qPCR)
The relative expressions of PI3K, AKT and mTOR in ovary were detected by qPCR. Then, the ovary was put into phosphate buffer solution (PBS) and homogenized on ice. The homogenate was centrifuged (3000r/min 15min 4℃) and the upper liquid was collected. Afterwards, total RNA was extracted by the method of TRIpure and reverse transcribed into cDNA. The CT value was derived and the relative expression of PI3K, AKT and mTOR were calculated according to the 2 −△△ CT value after ampli cation.
Detection and analysis of gut microbiome Fecal samples were collected in the sterile laboratory before sacri ced. The samples of fecal were placed into sterile centrifuge tubes with PBS and mixed. All samples were centrifuged (4000r/min 4℃ 30min) and precipitates were retained. After, total DNA was extracted according to the protocol of kit (Thermo Fisher, DNAzol, USA). The DNA of bacillus coli was applied as a template for ampli cation. Clone databases of 16S rDNA gene were constructed according to DNA Sample Prep Kit (Illumina, TruSeqTM, USA), and DNA sequences were detected (Illumina, MiSeq, USA). Finally, original sequences were spliced and ltered.
Principal Component Analysis (PCA) was applied to compare the bacterial diversity of gut microbiome.
The analysis of similarities (ANOSIM) was applied to analyze whether there was comparability between each group. In addition, the linear discriminant analysis (LDA) was applied to screen out the different species through LEfSe. Finally, the function of gut microbiome was annotated according to Kyoto Encyclopedia of genes and genomes database (KEGG database).

Weight
All mice were weighted after treated for 3 weeks. The weight of POF group was signi cantly lower than control group after modeling. Although the weight of EA group showed a recovery trend and besides, a signi cant increase could also be observed than the POF group, the weight of EA group was still slight lower than control group. Compared with the POF group, the weight of EN group had no signi cant difference (Fig. 1C).

Observation of estrous cycle
Vaginal exfoliated cells examination was applied to con rm the estrous cycle in mice. The vaginal exfoliated cells of all mice were evaluated, and the representative pictures which could clearly show the estrous cycle were selected. In the control group and EA group, nucleated epithelial, corni ed epithelial, and leukocytes were observed alternately. There were long-term existence of nucleated epithelial and leukocytes, but corni ed epithelial was not observed in the POF group and EN group. These results indicated that POF mice were established successfully. (Fig.S1) Histopathology examination As shown in Fig. 3, the antral follicles and corpus luteum could be observed in the control group. On the contrary, the number of atresia follicles increased in the POF group, which indicated the POF model was established successfully. There were more intact follicles in the EA group than POF group. Moreover, atresia follicles were decreased in the EA group. Although there were a few antral follicles in the EN group, atresia follicles and the structure of vacuoles could still be observed (Fig. 2).
In this study, it was found that the high levels of FSH and LH in POF mice. On the contrary, the levels of E 2 , AMH, and β-glucuronidase were decreased signi cantly in POF mice. Compared with the POF group, the levels of FSH and LH were signi cantly decreased, while the levels of E 2 , AMH, and β-glucuronidase were increased after EA. There was no signi cant difference in the levels of FSH, LH, AMH and βglucuronidase between EN group and POF group. However, the level of E 2 in EN group was signi cantly higher than that in POF group (Fig. 3).

QPCR examination
The results showed that the relative expressions of PI3K, AKT and mTOR in POF group were lower than those in control group. Obviously, there were no signi cant difference in the relative expression of PI3K, AKT and mTOR between EA group and control group. In addition, the expressions of PI3K, AKT and mTOR in EA group were signi cantly increased from those in EN group (Fig. 4).

Relative content of gut microbiome
The dual terminal sequence data were obtained by MiSeq sequencing platform. Non-repetitive sequences of OTUs were clustered according to more than 97% similarity. Moreover, chimeras were removed in the clustering process to obtain OTU sequences. The rarefaction cures were constructed according to the sobs index of each sample at different sequencing depths. The sequencing depth of all samples was more than 25000. With the increase of sequencing depth, the dilution curve tended to be at. All in all, the sequencing depths of all samples were reasonable (Fig. 5A).
In this study, there were no signi cant difference in the sobs index of OUT level between control group, POF group and EN group. However, the sobs index of OUT level in EA group was signi cantly higher than POF group (Fig. 5B).

Diversity of gut microbiome
As shown in Fig.S2-S5, ANOSIM analysis and PCA analysis indicated that the diversity of gut microbiome in each group was signi cant different. PCA analysis showed that the gut microbiome of POF group and EA group had obvious dispersion, and there was signi cant difference between POF group and EA group (Fig. 6). Similarly, although the PCA showed that diversity of gut microbiome in POF group and EN group had poor dispersion, there was signi cant difference between POF group and EN group (Fig. 7).

Species Changes of gut microbiome
To screen the different species of microbiome in each group, LDA was applied. And the top three most diverse and highest expressions dominant microbiomes in each group were selected. According to the results, the dominant microbiomes of the control group were Lactobacillus, Muribaculaceae, Monoglobus. Meanwhile, the dominant microbiomes in POF group were Lachnospiraceae, Eubacterium coprostanoligenes and Blautia. Then, Anaeroplasma, Ruminococcaceae and Eubacterium ventriosum were the dominant microbiomes of EA group. In addition, the dominant microbiomes of EN group were Tannerellaceae, Parabacteroides and Streptococcaceae (Fig. 8).
To compare the relative abundance of gut microbiome, three highest expressions dominant microbiomes in the control group were compared with other groups. The relative abundance of Lactobacillus in EA group was signi cantly higher than that in POF group and EN group. Then, the relative abundance of Muribaculaceae in EA group was signi cantly lower than that in control group. However, there was no signi cant difference in the relative abundance of Muribaculaceae between EA group and POF group. Furthermore, the relative abundance of Monoglobus in the control group was signi cantly higher than that in other groups, but there was no signi cant difference between POF group and EA group (Fig. 9).
The Firmicutes/Bacteroidetes (F/B) ratio was considered to be an important indicator of gut microbiome homeostasis [11]. Although the F/B ratio in EA group was signi cantly higher than that in control group, EA group was signi cantly lower than that in POF group. Meanwhile, there was signi cant different between EA group and EN group. (Fig. 10) Functionality of the gut microbiome According to the KEGG database, estrogen signaling pathway, oocyte maturation and PI3K-AKT signaling pathway were selected which were the most related to POF and the number of sequences was calculated.
The results showed that the numbers of sequence which associated with estrogens signaling, oocyte nutrition and PI3K-AKT signaling pathway were increased in POF group. Meanwhile, the numbers of sequence in EA group was lower than that of POF group. There was no signi cant difference between EA group and control group. At the same time, there was no signi cant difference in the numbers of sequence between EN group and POF group (Fig. 11).

Discussion
Patients with POF usually have symptoms such as amenorrhea for more than four months, hot ashes, night sweats, low libido and infertility [2]. Cisplatin as one of the most common-seen drugs for chemotherapy and considered to activate PTEN through the PI3K-AKT signaling pathway, which might potentially lead to follicular atresia [12]. Herein, the POF mice were established by cisplatin in this study.
The pathological morphology of ovary and vaginal exfoliated cells could re ect the ovarian status of mice. According to the above results, we found that EA could repair the injured of ovary and restore estrous cycle in cisplatin-induced POF mice. On the contrary, EN had little effect on the regulation of ovarian pathological morphology and estrus cycle.
FSH, LH, E2 and AMH are the important sex hormones that can re ect ovarian function [2], while both FSH and LH are thought to be secreted by hypophysis, while E2 is mainly secreted by granulosa cells in ovary. Serum AMH is applied to detect ovarian reserve function in clinic as reported [13]. The decrease of AMH level indicated premature depletion of primordial follicles. E2 can affect the levels of FSH and LH through the mechanism of negative feedback in HPO axis, which makes the sex hormones maintain a dynamic balance [8]. Follicular atresia causes the decrease of granulosa cells in follicles when POF occurs [14]. Therefore, POF can cause E2 and AMH to maintain at a low level. The long-term maintenance of low serum E2 level will break the negative feedback mechanism between ovary and pituitary, and lead to the levels increased of FSH and LH [15]. Therefore, both the high levels of FSH and LH in patients with POF, while low levels of E2 and AMH. According to the results, it was found that EA could effectively restore the levels of serum FSH, LH, E2 and AMH in POF mice. However, EN did not restore the levels of FSH, LH, E2 and AMH. These results might be related to EA increasing the number of antral follicles in ovary. It showed that acupoints could improve ovarian function more effectively than non-acupoints.
PI3K-AKT signaling pathway is involved in the process of oocyte growth, primordial follicle development and granulosa cell proliferation [16]. FSH is a glycoprotein hormone synthesized by hypophysis. FSH can combine with the receptor on the membrane of ovarian granulosa cells to activate PI3K-AKT pathway to promote the maturation of ovarian granulosa cells [17]. On the other hand, E2 can combine with the estrogen receptor α (ER-α) and estrogen receptor β (ER-β) in ovary [18]. PI3K has subunits of p85 and P110. ER-α combines with the subunit p85 in PI3K to activate AKT [19,20]. Meanwhile, mTOR is activated by AKT, which leads to proliferation and development of follicle [21]. In this study, we found that the PI3K-AKT signaling pathway in ovary of cisplatin-induced POF mice was inhibited. The relative expression of PI3K, AKT, and mTOR were reversed to normal in POF mice by EA. The result indicated that PI3K-AKT signaling pathway could be activated by electro-acupuncture at acupoints to promote the proliferation of ovarian cells. The results were consistent with the previous research results of other subjects.
There are a large number of microbiomes in human intestine, including probiotics, pathogenic bacteria and conditioned pathogen [22]. Gut microbiome is considered to be closely related to human health [23]. Growing evidence shows that the disorder of gut microbiome can lead to ovarian diseases, endocrine diseases and cardiovascular diseases [24,25]. It had been reported that the disorder of gut microbiome would break the barrier of intestinal mucosa [26]. Pathogenic bacteria and their metabolites reached the ovary through the circulation of blood to reduce ovarian function [27]. On the other hand, the enterohepatic circulation of estrogen was an important way to maintain the stability level of human serum estrogen [28]. Serum free estrogen is converted into conjugated estrogen by the liver. Then, the conjugated estrogen is excreted into the intestine with the bile [29]. β-glucuronidase was considered to be one of the metabolites of gut microbiome. Conjugated estrogen was converted into free estrogen by βglucuronidase [30]. Free estrogen enters the circulatory system of blood through intestinal resorption to promote growth of follicle and relieve the symptoms of POF.
In the control group, the three highest relative expression dominant microbiomes were Lactobacillus, Muribaculaceae and Monoglobus. Lactobacillus plays an important role in maintaining human health and is considered as one of the most important probiotics in human intestinal [31]. It had been reported that Lactobacillus could inhibit the growth and reproduction of pathogenic bacteria [32]. Muribaculaceae is widely distributed in the intestines of mice. Muribaculaceae can reduce the colonization of Clostridium di cile in the intestine [33]. Meanwhile, Muribaculaceae can degrade carbohydrates. In addition, Monoglobus can degrade pectin and maintain intestinal health [34]. According to the above results, we found that the three highest relative expression dominant microbiomes were probiotics in control group. They are important to maintain the health physiological function of intestine.
Lachnospiraceae, Eubacterium coprostanoligenes and Blautia were the dominant microbiomes in the POF group. Lachnospiraceae was considered to protect the intestinal mucosa [35]. On the contrary, Lachnospiraceae also could damage the pathway of glucose metabolism and promote the process of in ammation [36]. Therefore, Lachnospiraceae was considered as a conditional pathogen. Eubacterium coprostanogenes was considered to be pathogenic bacteria. It had been reported that Eubacterium coprostanogenes could reduce the intestinal absorption of cholesterol [37,38]. Blautia could prevent in ammation and promote the production of short chain fatty acids (SCFAs) to maintain intestinal homeostasis, so it was considered to be a potential probiotic [39,40]. The existence of bene cial bacteria in the rst three dominant bacteria of POF mice whether be related to the self-healing function of mice, and it still needed to be further researched.
In the EA group, Anaeroplasma, Ruminococcaceae, and Eubacterium ventriosum were the three microbiomes with the highest relative expression. Anaeroplasma was an agent of anti-in ammatory. In addition, Anaeroplasma could maintain the homeostasis of immune in intestinal mucosal [41]. Similarly, Ruminococcaceae and Eubacterium ventriosum had the function of anti-in ammatory, and both were the common bene cial bacteria in the intestine [42,43]. As a member of SCFAs producer, Ruminocaceae was considered to maintain intestinal immune homeostasis. All the rst three dominant microbiomes in POF mice were probiotics after treating by EA. The result indicated that EA could increase the abundance and diversity of probiotics.
The dominant microbiomes of three highest relative expressions in the EN group were Tannerellaceae, Parabacteroides, and Streptococcaceae. Several studies had found that Tannerellaceae promoted the occurrence of intestinal in ammation [44]. Parabacteroides were considered to be bene cial bacteria [45]. Moreover, it had been reported that Parabacteroides could alleviate liver injury and reduce the expression of in ammatory genes in liver [46]. Streptococceae is a conditioned pathogen, which was a common microbiome in human oral cavity, skin, intestinal, and upper respiratory tract [47]. Although the content of probiotics could be increased by EN, the dominant bacteria were pathogenic bacteria.
According to the results, EA could signi cantly reduce the F/B ratio, which indicated that EA had a good effect on improving the structure of gut microbiome. The abundance and diversity of gut microbiome in POF mice could be regulated by EA. According to the results of KEGG pathway analysis, the gut microbiome which related to estrogen signaling pathway, oocyte nutrition and PI3K-AKT signaling pathway were regulated by EA. These ndings indicated that EA might play a role in the treatment of POF by regulating the abundance and diversity of gut microbiome.
In the future, antibiotic cocktail mice would be selected to further research the therapeutic mechanism of EA on POF by fecal microbial transplantation and macrogene sequencing.

Conclusion
EA could restore ovarian pathological morphology and estrous cycle of mice with POF. At the mean time, EA was also found to regulate sex hormones and PI3K-AKT signaling pathway in mice with POF. Further, we speculated that estrogen level might be restored by EA through regulating the abundance and the diversity of gut microbiome in POF mice. Estrogen combined with the estrogen receptor on follicular granulosa cells to activate PI3K-AKT signaling pathway, and that could promote the growth and development of follicles.
The disease from POF would become an increasing public health burden. Regulating the gut microbiome by EA to impact ovarian function provides an exciting future therapeutic. This study provided a preliminary veri cation for revealing the mechanism of EA in the treatment of POF.

Declarations
The selected points and weight of mice. A and B: acupoints and non-acupoints. C, the weight of mice in each group (*mean signi cant difference from the control group at P<0.05; # mean signi cant difference from the POF group at P<0.05; △mean signi cant difference from the EA group at P<0.05.).

Figure 2
The pathological examination of ovaries in mice of each group. (A and a, mean control group; B and b, mean POF group; C and c, mean EA group; D and d, mean EN group)   The histogram of LDA value distribution in each group. Three microbes with the most difference signi cantly were selected in each group.