Page 1/14

Effects of perioperative managements on ocularsurface microbiota in intravitreal injection patientsYaguang Hu 

First A�liated Hospital of Xi'an Jiaotong UniversityQiong Wu 

Northwest Women's and Children's HospitalFang Sui 

First A�liated Hospital of Xi'an Jiaotong UniversityMing Zhang 

First A�liated Hospital of Xi'an Jiaotong UniversityZhen Zhang 

First A�liated Hospital of Xi'an Jiaotong UniversityRui Shi 

Shaanxi Provincial People's HospitalNa Hui 

Xi'an Huaxia Eye HospitalLi Qin 

First A�liated Hospital of Xi'an Jiaotong UniversityLi Li  ( eyelili2010@mail.xjtu.edu.cn )

First A�liated Hospital of Xi'an Jiaotong University

Research Article

Keywords: microbiota, intravitreal injection, ocular surface, perioperative managements, high-throughputsequencing

Posted Date: February 11th, 2021

DOI: https://doi.org/10.21203/rs.3.rs-154655/v1

License: This work is licensed under a Creative Commons Attribution 4.0 International License.  Read Full License

Page 2/14

AbstractBackground: We investigated the effects of perioperative managements on ocular surface microbiota inpatients who received intravitreal injections.

Methods: Samples of ocular surface microbiota were obtained from 41 eyes of 41 patients who receivedintravitreal injections at the First A�liated Hospital of Xi'an Jiaotong University and People’s Hospital inShaanxi Province. The microbial 16s rDNA from samples were analyzed using an Illumina MiSeq high-throughput sequencing platform. Operating classi�cation units (OTU) clustering and alpha/beta diversityanalysis was performed for species classi�cation.

Results: High-throughput sequencing yielded 1697337 sequences and 1396, indicating an equivalentnumber of ocular surface microbiota. OTU was signi�cantly lower in the eyes that received intravitrealinjections, compared with those who did not received injections (P<0.05). The number of OTUs wasinversely correlated with the administered of perioperative managements and the times of intravitrealinjections. Beta diversity showed a signi�cant difference among each groups (P<0.05). With theincreased perioperative managements, the prevalence of gram-positive bacteria (Clostridium) haddecreased and the prevalence of gram-negative bacteria (Neisseria) had increased.

Conclusions: Microbiota on the ocular surface were signi�cantly different between treated and controleyes. And the composition of microbiota on the ocular surface was dramatically changed over time ineyes that received intravitreal injections. Perioperative managements might disrupt the balance of theocular surface microbiota, increasing risk of infectious disease. Therefore, perioperative managementsincluding local antibiotic eye drops should be applied cautiously before intraocular surgery.

BackgroundIntravitreal injections of anti-vascular endothelial growth factor (VEGF) were preferred to inhibit thepathological angiogenesis, which is a main feature of many posterior segment pathologies and causeirreversible visual �eld defect or vision loss. And millions of injections are performed each year all overthe world. However, it carried an inherent risk of infectious endophthalmitis. According to the mostcomprehensive review to date, with data from 14,866 intravitreal injections, the estimated prevalence ofinfectious endophthalmitis is 0.02% to 0.09%1. Nowadays, in clinical works several procedures wereconducted during the intravitreal injection to prevent infectious endophthalmitis, including application ofantibiotic eye drop and sterilizing conjunctival sac with 5% povidone iodine. Doctors would like to believethese perioperative managements might decrease the pathogens. However, some published articlesshowed that using antibiotics eye drops could not make the infection rate lower 2, 3. Some study showedthat local antibiotic eye drops might increase microbial resistance and opportunistic infection4. 

The ocular surface is a microecosystem colonized by speci�c �ora. It serves as a natural biologicalbarrier between inside and outside of the eye. As intestinal and oral diseases were related to microbiotacomposition, the ocular diseases were also affected by the ocular surface microbiota. Previous studies

Page 3/14

shown that microbiota contribute to immune regulation by promoting the production of dendritic cellsand Treg cells, as well as regulating the secretion of cytokines through natural killer T lymphocytes5, 6.Furthermore, microbiota on the ocular surface may be disrupted by perioperative managements like theapplication of local antibiotic eye drops, and induce the immune balance of ocular surface to bedamaged.

To our knowledge, few previous studies have investigated the effect of perioperative management onocular surface microbiota in intravitreal injection patients. This study was designed to investigate thepotential effect of perioperative management of intravitreal injection on the ocular surface microbiota inthose patients who received intravitreal injections.

Materials And MethodsStudy design

This study was approved by the institutional review committee of the First A�liated Hospital of Xi'anJiaotong University, and all methods was conducted in accordance with the Declaration of Helsinki.Patients received intravitreal injections in the ophthalmic outpatient department of the First A�liatedHospital of Xi'an Jiaotong University or the ophthalmic inpatient department of Shaanxi ProvincialPeople's Hospital, during the period from January 2018 to December 2018. A research assistant collecteda total of 41 samples according to a standardized training protocol. The control group (Group A) included19 patients who had not been treated with any perioperative managements. For each patient, sampleswere obtained from ocular surface in each eye. In treated group (Group B), 22 samples were obtainedfrom the eyes that had been treated with antibiotic eye drops (levo�oxacin)(Santen Pharmaceutical Co.,Ltd., Japan) prior to intravitreal injections. Samples were analyzed with a Hiseq2500 sequencer (Illumina,US), Thermomixer (Eppendorf, Germany), low-temperature centrifuge (Eppendorf, Germany), Vortexgenerator (Haimen Its Klingbeil, China), Qubit �uorescence spectrometer (Thermo�sher, Malaysia), and aQubit dsDNA BR assay kit (Thermo�sher, Malaysia).

In Group B1 (11 samples), patients had received only once intravitreal injection. Meanwhile, 11 patients inGroup B2 had received more than twice intravitreal injections. The antibiotic eye drops used for treatmentwere levo�oxacin, which were administered for 12 times before the injections. Patients who did notcomplete the follow-up process were excluded from the study, as were those patients with ocular surfacediseases, and those received dexamethasone, antibiotic eye drops or ointments within three months.Patients who had undergone any eye surgery (except for intravitreal injection) within the past 3 monthswere also excluded. Informed consent was obtained from all subjects for all examinations andprocedures.

Sample collection and DNA storage

Patients were asked to look upward and eyelids were everted. Samples were obtained by mild sweepingthe inferior fornices of eye for several times with sterile cotton swabs. Then the conjunctival swabs were

Page 4/14

stored in sterile tubes and transported to the laboratory in an icebox. The samples were stored at -80℃,and DNA (16Rs) was extracted within one week.

DNA extraction and PCR ampli�cation

Tip of the cotton swab was inactivated and transferred to a 2-mL centrifuge tube. After pyrolysis andcooling, 1 mL phenol/chloroform/isoamyl alcohol was added. Tubes were then centrifuged at 12,000 rpmfor 10 min at 25℃. Supernatant (900 µL) was removed. Pre-cooled isopropanol and 3 mL 10% sodiumacetate were added to the centrifuge tube for overnight precipitation at -20℃. Samples were centrifugedfor more than three times. After being dried for 3-5 minutes, samples were dissolved in appropriate buffer,and then extracted. Using the bacteria in the sample as DNA template, the V3-V4 region of the bacterial16S rDNA gene was ampli�ed with 341F (5'-ACTCCTACGGGAGGCAGCAG-3') and 806R (5'-GGACTACHVGGGTWTC- TAAT-3') as joint sequence and a universal primer as the barcode sequence7, 8.The polymerase chain reaction (PCR) system was set up with 30 ng of quali�ed genomic DNA samplesand corresponding fusion primers. The PCR reaction parameters were adjusted for PCR ampli�cation.The results of PCR ampli�cation were puri�ed with Agencourt AM Pure XP magnetic beads, thendissolved in elution buffer and labeled to establish a database. An Agilent 2100 Bioanalyzer was used todetect the fragment range and concentration of the DNA library. Based on the size of the insertedfragments, the HiSeq platform was selected for use in high-throughput sequencing.

Sequencing and data processing

The Miseq PE301+8+8+301 IIIumina platform was used for double-ended sequencing. Low-quality readswere �ltered out and stored o�ine; the remaining clean data were used for subsequent analysis. FLASHsoftware was used for the creation of mosaics, �ltering, and chimera removal. OTU analysis wasconducted with USEARCH software (v7.0.1090). OTU analysis included assessments of speciescomplexity, differences in species composition between groups, and correlation analysis9, 10. TheGreenGene database was annotated based on the results obtained. Master software was used tocalculate the alpha diversity index of each sample. The statistical methods used for analysis included therank-sum test (Wilcoxon test) and linear discriminant effect size (LEfSe) analysis. P < 0.05 wasconsidered statistically signi�cant.

ResultsOTU comparison

Demographic characteristics of the patients (age, gender and disease type) were shown in Table 1. Therewas no signi�cant difference between each group (P>0.05). 

OTU stand for operational taxonomic unit (strain, genus, species, grouping, etc.), which is used to classifygroups of closely related individuals. Sequences with >97% similarity are grouped into OTUs. Group A hada total OTU number of 973 (369 unique); group B1 had a total OTU number of 760 (213 unique); group B2

Page 5/14

had a total OTU number of 668 (166 unique). The number of OTUs with overlap among the three groupswas 357. The number of OTUs in Group A was higher than that in Group B1, and the number of OTUs inGroup B1 was higher than that in Group B2. These �ndings indicate that ocular surface microbialcommunity diversity was inversely correlated with the perioperative managements and the number ofintravitreal injections administered (Figure 1a). The results of PLS-DA analysis based on OTU abundancerevealed the similar changes about the microbiota on ocular surface among Group A, Group B1, andGroup B2 (Figure 1b).

Microbial diversity analysis

After obtaining the results of OTU clustering analysis, Ace, Chao, Simpson, and Shannon evaluationindexes were used to analyze the alpha diversity of all samples. There was no statistically signi�cantdifference among each sample in microbial community alpha diversity (P>0.05) (Figure 2a), indicatingthat there was no statistically signi�cant difference in microbial community composition in each group.

Beta diversity analysis was performed to measure differences among multiple groups. The results of betadiversity analysis revealed a signi�cant difference in  composition of microbiota on ocular surfacebetween Groups A, B1 and B2 (P=1.84×10-5) (Figure 2b). 

Species composition of the conjunctival microbiota

In Figure 3, species abundance was compared among the groups at the phylum, class, order,family, genus, and species levels. All of the microbiota in each group and each level were shown in theleft. We selected the top three microbiota in Group A and analyzed their variations in three groups. Theresults were shown by histogram in the middle. Furthermore, we compared each microbiota in differentgroups, and found out the top �ve microbiota that increased (Red) and decreased (Green) signi�cantly.We ranked the microbiota by the results of fold changes (B2/A). These results were shown in the righttable.

In Group B1, WPS-2 increased for 54.27 folds and S24-7 increased greatest for 11.1 folds. Furthermore,Streptococcus_euqi dramatically increased for 61088.15 folds. In Group B2 Streptococcus_euqiincreased for 298 folds. Instead, Synergistetes decreased for 0.08 fold in Group B1 and 0.12 fold in GroupB2. Deinococci, Deinococcales, Deinococcaceae and Deinococcus were all signi�cantly decreased inGroup B1 and B2. Overall, microbiota on the ocular surface was obviously changed after perioperativemanagements. S24-7 and Streptococcus_euqi were gram-negative bacteria. Synergistetes, Deinococci,Deinococcales, Deinococcaceae and Deinococcus were all gram-positive bacteria.

Analysis of species differences

The LefSe histogram shows the microorganisms with signi�cantly differential abundance in Group A,Group B1, and Group B2. Column length represents relative abundance of a given species (Figure 4a)11,

12. The bacteria that were signi�cantly more common in Group B2 were Alphaproteobacteria,

Page 6/14

Sphingomonadales, Sphingomonadaceae, Rhizobiales, Rhizobiaceae, Trueperaceae, Proteus,Aquabacterium, Oxalobacteraceae, Pelomonas, and Sphingopyxis. Facklamia was signi�cantly morecommon in Group B1. Mollicutes, Tenericutes, and Moraxella were signi�cantly more common in GroupA. At the phylum level, the bacteria with greatest abundance in Group A were Mollicutes, while the mainbacteria in Group B2 were Alphaproteobacteria. The results showed that antibiotic treatment altered thedominant and core bacterial species present on the ocular surface. This �nding is consistent with thecladogram of species abundance LDA (Figure 4b).

DiscussionIn China, the patients were asked to use antibiotic eye drop for more than 12 times before receivedintravitreal injection. Most of the ophthalmologists recognized that the application of antibiotic eye dropwould signi�cantly lower the risk of infectious endophthalmitis. Nerveless, there was no strong evidencethat the application of antibiotic eye drops on the ocular surface before intraocular surgery is a preferredway to prevent intraocular infection13-15. Many system analyses had shown that the infectiousendophthalmitis incidence with antibioprophylaxis was about 0.052%. Instead, the incidence in controlgroup without antibiotic eye drops was 0.048% in million patients. There was no signi�cant differencebetween the two groups16. Thus, in-depth explored the reasons that why the effect of antibioprophylaxiswas not dramatic in preventing infectious endophthalmitis was the most important thing. 

The structure and diversity of microbiota was a foundation for analyzing the microecosystem on theocular surface. Through this study, we detected that main composition of the ocular surface microbiotabefore and after intravitreal injections. Furthermore, we found that the composition of the ocular surfacemicrobiota also signi�cantly changed when the patients received several times of intravitreal injections. Itmeant that perioperative procedures conducted including local antibiotic eye drops might disrupt theoriginal balance of the ocular surface microbiota, which may deeply in�uence the homeostasis of ocularsurface, even might increase the risk of infectious endophthalmitis.

The published clinical study showed the majority pathogens nearly 75% that induced intraocularinfections were gram-negative bacteria13. Besides, infections occurred on the ocular surface were mainlyinduced by gram-positive bacteria17, 18. Our research had illustrated this question based on in-depthanalyzing ocular surface microbiota. As above results shown, the relative abundance of gram-negativebacteria increased after perioperative managements. Thus, perioperative managements reduced thespecies diversity of the microbiota on the ocular surface. This �nding was similar to those studies onintestinal, and oral microbes19, 20. In brief, the perioperative managements we conducted beforeintravitreal injection, including application of antibiotic eye drop, had obviously changed the compositionof ocular surface microbiota. And we ought to pay attention whether the increasing relative abundance ofgram-negative bacteria had some relationship with the infectious endophthalmitis.

Altering the relative abundance of symbiotic and non-pathogenic microbiota, which might induce thegrowth of pathogenic bacteria and thus induce devastating eye infections, such as bacterial keratitis and

Page 7/14

endophthalmitis13, 21. Staphylococcus epidermidis was typically non-pathogenic bacteria, which couldinduce infectious ocular diseases in individuals with weakened immune system22. In our study, atthe species level, the relative abundance of Staphylococcus epidermidis decreased after perioperativemanagements applied but the other microbiota like Streptococcus_equi was increased. Perioperativemanagements including the application of levo�oxacin eye drop, tobramycin �ushing �uid and 5%povidone iodine �ushing �uid. They acted on different bacteria. Levo�oxacin eye drop sterilized gram-positive bacteria, and tobramycin mainly sterilized gram-negative bacteria. Moreover, 5% povidone iodine�ushing �uid sterilized both of them. All of perioperative managements were likely to alter thecomposition of microbiota on the ocular surface, which might involve in the infectious endophthalmitis.In addition, in intestine, indolepropionic acid, a metabolite produced exclusively by the microbiota fromdietary tryptophan, participated in the process of immune regulation23. The metabolites of microbiotamight destroy the microenvironment of ocular surface and the function of biological barrier wasdiminished. Further studies should be conducted to illustrate these problems. And we should consideratethe perioperative managements before intravitreal injection carefully.

Notably, our study had some limitations. We could not obtain samples from the patients who receivedintravitreal injections without perioperative managements. And we could not obtain samples frompatients with infectious endophthalmitis, in order to estimate whether the microbiota induced infectiousendophthalmitis was accordance with the changes of microbiota on ocular surface. Besides, futurestudies should include a larger sample size, in order to identify the signi�cant differences in ocularsurface microbial diversity between eyes with and without the local antibiotic treatment prior tointravitreal injection. Additional studies are necessary to validate this new scienti�c concept.

ConclusionOur study showed that with the perioperative managements, relative abundance of microbiota on theocular surface such as Staphylococcus epidermidis, Acinetobacter, and Actinomyces, were decreased.Furthermore, perioperative managements, prior to intravitreal injection increased the relative abundanceof conditional pathogenic bacteria such as Neisseria. It demonstrated that perioperative managementsincluding antibiotic eye drops destroyed the relative balance of microbiota on the ocular surface, allowingpathogenic bacteria to proliferate and increasing the risk of infectious endophthalmitis. Thus, thenecessary of perioperative managements before intravitreal injection should be deeply researched.

AbbreviationsOTU operating classi�cation units

VEGF vascular endothelial growth factor

PCR polymerase chain reaction

Page 8/14

DeclarationsEthics approval and consent to participate

This study was approved by the institutional review committee of the First A�liated Hospital of Xi'anJiaotong University, and all methods was conducted in accordance with the Declaration of Helsinki.Informed consent was obtained from all subjects for all examinations and procedures.

Consent for publication

Not applicable. 

Availability of data and materials

The datasets of 41 participants generated and analyzed during the current study are available inthe supplementary material.

Competing interests

The authors report no con�icts of interest.

Funding

This work was supported by Key Research and Development Program of Shaanxi Province in China(Grant no. 2017SF-028), First A�liated Hospital of Xi'an Jiaotong University of China.

Authors’ contribution

Yaguang Hu and Qiong Wu wrote the main manuscript text.

Yaguang Hu and Fang Sui prepared �gures and tables. 

Ming Zhang, Zhen Zhang, Rui Shi and Na Hui collected the samples.

Li Qin guided the sample collection and communicated with participates.

Li Li designed and supervised this work.

All authors reviewed the manuscript.

Acknowledgement

Not applicable. 

References

Page 9/14

1. Jager RD, Aiello LP, Patel SC, Cunningham ET, Jr. Risks of intravitreous injection: a comprehensivereview. Retina (Philadelphia, Pa) 2004;24:676-698.10.1097/00006982-200410000-00002

2. Tuni-Picado J, Martinez-Palmer A, Fernandez-Sala X, et al. Infectious postoperative endophthalmitisafter cataract surgery performed over 7 years. The role of azithromycin versus cipro�oxacin eyedrops. Rev Esp Quimioter 2018;31:15-21

3. Asghar A, Ellhai I, Obaid N, Sughra U. Role of topical antibiotics in prophylaxis againstendophthalmitis following intravitreal antibiotics. Pak J Med Sci 2018;34:1283-1287.10.12669/pjms.345.14817

4. Avery RL, Bakri SJ, Blumenkranz MS, et al. Intravitreal injection technique and monitoring: updatedguidelines of an expert panel. Retina (Philadelphia, Pa) 2014;34 Suppl 12:S1-S18.10.1097/IAE.0000000000000399

5. Hunyor AP, Merani R, Darbar A, Korobelnik JF, Lanzetta P, Okada AA. Topical antibiotics andintravitreal injections. Acta Ophthalmol 2018;96:435-441.10.1111/aos.13417

�. Ding T, Schloss PD. Dynamics and associations of microbial community types across the humanbody. Nature 2014;509:357-360.10.1038/nature13178

7. Caporaso JG, Lauber CL, Walters WA, et al. Ultra-high-throughput microbial community analysis onthe Illumina HiSeq and MiSeq platforms. ISME J 2012;6:1621-1624.10.1038/ismej.2012.8

�. Caporaso JG, Kuczynski J, Stombaugh J, et al. QIIME allows analysis of high-throughput communitysequencing data. Nat Methods 2010;7:335-336.10.1038/nmeth.f.303

9. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics2010;26:2460-2461.10.1093/bioinformatics/btq461

10. Fan Q, Teo YY, Saw SM. Application of advanced statistics in ophthalmology. Invest Ophthalmol VisSci 2011;52:6059-6065.10.1167/iovs.10-7108

11. Segata N, Izard J, Waldron L, et al. Metagenomic biomarker discovery and explanation. Genome Biol2011;12:R60.10.1186/gb-2011-12-6-r60

12. Knop N, Knop E. Conjunctiva-associated lymphoid tissue in the human eye. Invest Ophthalmol VisSci 2000;41:1270-1279. PMID 10798640

13. Ham B, Hwang HB, Jung SH, Chang S, Kang KD, Kwon MJ. Distribution and Diversity of OcularMicrobial Communities in Diabetic Patients Compared with Healthy Subjects. Curr Eye Res2018;43:314-324.10.1080/02713683.2017.1406528

14. Mela EK, Drimtzias EG, Christo�dou MK, Filos KS, Anastassiou ED, Gartaganis SP. Ocular surfacebacterial colonisation in sedated intensive care unit patients. Anaesth Intensive Care 2010;38:190-193.10.1177/0310057X1003800129

15. Shin H, Price K, Albert L, Dodick J, Park L, Dominguez-Bello MG. Changes in the Eye MicrobiotaAssociated with Contact Lens Wearing. mBio 2016;7:e00198.10.1128/mBio.00198-16

1�. Benoist d'Azy C, Pereira B, Naughton G, Chiambaretta F, Dutheil F. Antibioprophylaxis in Prevention ofEndophthalmitis in Intravitreal Injection: A Systematic Review and Meta-Analysis. PLoS One

Page 10/14

2016;11:e0156431.10.1371/journal.pone.0156431

17. Speaker MG, Milch FA, Shah MK, Eisner W, Kreiswirth BN. Role of external bacterial �ora in thepathogenesis of acute postoperative endophthalmitis. Ophthalmology 1991;98:639-649; discussion650.10.1016/s0161-6420(91)32239-5

1�. Mah FS, Davidson R, Holland EJ, et al. Current knowledge about and recommendations for ocularmethicillin-resistant Staphylococcus aureus. J Cataract Refract Surg 2014;40:1894-1908.10.1016/j.jcrs.2014.09.023

19. Daroy ML, Lopez JS, Torres BC, Loy MJ, Tuano PM, Matias RR. Identi�cation of unknown ocularpathogens in clinically suspected eye infections using ribosomal RNA gene sequence analysis. ClinMicrobiol Infect 2011;17:776-779.10.1111/j.1469-0691.2010.03369.x

20. Suto C, Morinaga M, Yagi T, Tsuji C, Toshida H. Conjunctival sac bacterial �ora isolated prior tocataract surgery. Infect Drug Resist 2012;5:37-41.10.2147/IDR.S27937

21. Zhang Y, Liu ZR, Chen H, et al. Comparison on conjunctival sac bacterial �ora of the seniors with dryeye in Ganzi autonomous prefecture. Int J Ophthalmol 2013;6:452-457.10.3980/j.issn.2222-3959.2013.04.08

22. Kodjikian L, Salvanet-Bouccara A, Grillon S, et al. Postcataract acute endophthalmitis in France:national prospective survey. J Cataract Refract Surg 2009;35:89-97.10.1016/j.jcrs.2008.10.018

23. Dodd D, Spitzer MH, Van Treuren W, et al. A gut bacterial pathway metabolizes aromatic amino acidsinto nine circulating metabolites. Nature 2017;551:648-652.10.1038/nature24661

Table  Group A Group B1 Group B2

Age (years) 63.52±9.72 60.82±10.50 58.18±14.89Gender

(Male/Female)8 (42.1%)/11

(57.9%)9 (81.8%)/2

(18.2%)5 (45.5%)/6

(54.5%)Diseases (%) DR 47.3 45.4 45.4

AMD 31.6 18.1 27.2PCV 10.5 9.1 0NVG 10.5 9.1 9.1CRVO 0 18.2 9.1BRVO 0 0 9.1

Table 1 The demographic characteristics of patients in three groups. DR, diabeticretinopathy; AMD, age-related macular degeneration; PCV,  poly-poidal choroidalvasculopathy; NVG, neovascular glaucoma; CRVO, central retinal vein occlusion; BRVO,branch retinal vein occlusion. 

Figures

Page 11/14

Figure 1

(a) In this Venn diagram illustrating the relationships among various OTUs, colors represent differentsamples or different groups, and the number of overlapping pairs is the number of OTUs shared betweensamples or groups. The number of areas of overlap indicates the number of OTUs common to multiplesamples or groups. (b) In this schematic presenting the results of OTU PLS-DA, dots of different colors orshapes represent sample groups under different environments or conditions. The scales of the horizontaland vertical axes are relative distances, which have no practical signi�cance. The abscissa and ordinaterepresent the factors suspected to in�uence microbial species diversity.

Figure 2

Page 12/14

Alpha and Beta diversity boxplots. (a) OTU clustering analysis including Ace, Chao, Simpson andShannon evaluation indexes were used to analyze the alpha diversity of all samples. From top to bottom,the �ve lines represent: minimum, �rst quartile, median, third quartile and maximum. (b) Beta diversityanalysis was performed to measure differences among multiple groups.

Figure 3

Page 13/14

Classi�cation of OTUs by (a) phylum, (b) class, (c) order, (d) family, (e) genus and (f) species levels. Colorhistograms (left) represented species abundance. Black and white histograms (middle) showed top threemicrobiota in Group A and the the variation of them in three groups. Color table (Right) showed the top�ve microbiota that increased (Red) and decreased (Green) signi�cantly in three groups. It was ranked bythe results of fold changes (B2/A).For analyses at all other levels, species with abundance <0.5% areclassi�ed as “Other”.

Figure 4

Page 14/14

(a) LEfSe analysis of the LDA diagram. Colors represent different groups of microbiota with signi�cantdifferences in relative abundance among groups, as re�ected by LDA scores with absolute value > 2.0.Bar colors represent different treatment groups. Bar length represents LDA score as a measure ofdifferential relative abundance among groups. (b) LEfSe analysis yielded a cluster tree, with differentcolors representing different groups. Nodes of different colors represent microbial communities thatplayed an important role in the same group. Colored circles represent biomarkers, and yellow nodesrepresent microbial groups that did not play an important role in any of the groups investigated. Forconcentric circles, moving from the center to the outside, circles represent the level of phylum, class, order,family and genus.