Gut Microbial Variation May Predict Neonatal Jaundice and Microbial Alterations After Treatment


 Background: Neonatal jaundice is a common disease that affects up to 60% of newborns and the relationship between early gut microbiome and development of neonatal jaundice is not fully understood. This study aims to characterize gut microbiome of newborns and to explore its association with risk of neonatal jaundice.Methods: We collected 257 fecal samples from 58 infants at 5 time points of 0, 1, 3, 6 and 12 months prospectively, and finally 114 samples from 6 neonatal jaundice infants (NJI) with treatment and 19 matched non-NJI completed Miseq sequencing and analysis. We characterized gut microbiome, identified microbial differences and gene functions. Results: Meconium microbial diversity from NJI was decreased versus non-NJI. Genus Gemella was decreased in NJI versus non-NJI. Eleven predicted microbial functions including fructose-1,6-bisphosphatase III and Pyruvate carboxylase subunit B decreased, while 3 functions including acetyl CoA acyltransferase increased in NJI. After treatments, microbial community presented a significant alterations based beta-diversity. Phylum Firmicutes and Actinobacteria were increased, while Proteobacteria and Fusobacteria were decreased. Microbial alterations were also analyzed between 6 recovery NJI and 19 non-NJI. Conclusion: Gut microbiota was unique in meconium microbiome from NJI, implying early gut microbiome intervention could be promising for the management of neonatal jaundice. Alterations of gut microbiota from NJI can be of great value to bolster evidence-based prevention against ‘bacterial dysbiosis’.

The extracted DNA used as the template to amplify the V3 to V4 regions of 16S rRNA gene. The forward primer (341F) was 5'-CCTACGGGNGGCWGCAG-3' and the reverse primer (805R) was 5'-GACTACHVGGGTATCTAATCC-3'. The PCR ampli cation was performed in a EasyCycler 96 PCR system (Analytik Jena Corp., AG) using the following program: 1 cycle of 95 ℃ for 3 min; 21 cycles of (94℃ 30 s; 58℃ 30 s; 72℃ 30 s); 1 cycle of 72℃ 5 min. The products from different samples were indexed and mixed at equal ratios for sequencing according to the manufacturer's instructions, and the sequencing was performed on the Illumina MiSeq platform at the Shanghai Mobio Biomedical Technology Co. Ltd. Raw Illumina read data for all samples were deposited in the European Bioinformatics Institute European Nucleotide Archive database under accession number PRJNA680178 and PRJNA665920.
Operational Taxonomic Units (OTUs) and taxonomy annotation Equal numbers of reads were randomly chosen from all samples, and then OTUs were binned using UPARSE pipeline [11]. Sequences with 97% similarity level were clustered into OTUs. The software RDP classi er version 2.6 [12] was used to assign sequences to the new bacterial taxonomy.
Bacterial diversity and taxonomic analysis Bacterial diversity was assessed by sampling-based analysis of OTUs and presented by ACE index, which was calculated using R program package "vegan". Principal coordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) based on OTU abundance and distribution was conducted by R package (http://www. R-project. org/) to analyze microbial communities [13]. The weighted and unweighted Unifrac distances were calculated with phyloseq package [14]. A heatmap of the identi ed key variables was generated by the Heatmap Builder.
Bacterial differences were compared at the phylum and genus levels. Fecal microbial characteristics were analyzed by linear discriminant analysis (LDA) effect size (LEfSe) method (http://huttenhower.sph.harvard.edu/lefse/) [15]. Using a normalized relative abundance matrix, LEfSe performs the Kruskal-Wallis rank sum test to determine characteristics with signi cantly different abundances between assigned bacterial and uses LDA to assess the effect size of each characteristics.

Functional annotation of gut microbial 16S rRNA gene
To predict the functional pro les of microbial communities based on 16S rRNA gene sequences, we utilized phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt) version 1.0.0 pipeline [16] and human version 0.99 [17] to establish KEGG orthologie (KO) and KEGG pathway/module pro le.

Statistical analysis
The Wilcoxon rank sum test was used to compare continuous variables between both groups. Fisher's exact test was used to compare categorical variables. The infant weight-for-length z score were Gut microbiomes alterations in NJI at 0 month We collected meconium from NJI and matched non-NJI at 0 month. Compared with the non-NJI, fecal microbial diversity, as estimated by ACE estimator, was markedly decreased in NJI (P < 0.05, Fig. 2a, Supplementary Table S1). Moreover, a Venn diagram displaying the overlaps showed that 201 of the total 330 OTUs were shared between NJI and non-NJI (Fig. 2b). Notably, 52 of 253 OTUs were unique for NJI.
To assess similarity among microbial communities, we performed PCoA and NMDS analysis based on unweighted UniFrac distance (Fig. 2c&d, Supplementary Table S2&3).
A heatmap of the identi ed key variables were revealed and demonstrated that a total of 27 key OTUs were signi cantly different between NJI and non-NJI ( Supplementary Fig. 1, Supplementary Table S4). We further analyzed infant gut microbiota composition and alterations at the phylum and genus levels between two groups. Fecal bacterial composition in each sample at the phylum and genus levels were shown in Supplementary Fig. 2a&b (Supplementary Table S5&S6). Average composition of microbial community at the phylum and genus levels between two groups were shown in Fig. 2e&f We used the linear discriminant analysis (LDA) Effective Size (LEfSe) to determine the speci c bacterial taxa related to neonatal jaundice. A cladogram representative of fecal microbial structure and their predominant bacteria displayed the greatest differences in taxa between NJI and non-NJI (all P < 0.05, Supplementary Fig. 3, Supplementary Table S10). Meanwhile, the cladogram of fecal microbial structure between NJI and non-NJI also showed the greatest differences in taxa (all P < 0.05, LDA > 2, Fig. 3a, Supplementary Table S10), which suggested gut microbial alterations in NJI.
Microbial metabolic functioning predictions using PICRUSt pipeline [16]  Gut microbiota temporary dysbiosis between pre-treatment and post-treatment The Venn diagram showed overlaps between pre-treatment (0 month) and post-treatment (1 month), revealing that 113 of 214 OTUs were shared between two groups (Fig. 4a). To display microbiome space between pre-treatment and post-treatment, we performed PCoA and NMDS analysis based on unweighted UniFrac distance (P < 0.05, Fig. 4b&c, Supplementary Table S12&13). Moreover, PCoA was conducted based on weighted UniFrac distances to assess microbial distribution among 0, 1, 3, 6 and 12 months (Fig. 4d, Supplementary Table S14). PCoA analysis indicated samples tended to be uniform at 0 and 12 month, no obvious separation was observed in YCM treatment. Notably, samples are most heterogeneous at age 1-6 months.
A heatmap of the identi ed key variables were revealed and demonstrated that a total of 26 key OTUs were signi cantly different between two groups ( Supplementary Fig. 4, Supplementary Table S15). Fecal bacterial composition and difference at the phylum and genus levels between two groups were shown in Fig. 4e&f&g&h, respectively (all P < 0.05, Supplementary Table S16-19).
We detected the greatest differences in taxa between pre-treatment and post-treatment using the LEfSe method and LDA scores, as shown in Fig. 5a and Supplementary Fig. 5a (all P < 0.05, LDA > 2.4, Supplementary Table S20).
The predominant fecal microbial functions between pre-treatment and post-treatment were shown by a cladogram and LDA analysis (all P < 0.05, LDA > 2, Fig. 5b and Supplementary Fig. 5b, Supplementary  Table S21). These data revealed signi cant differences between both groups.

Gut microbiota alterations in recovery NJI and non-NJI
The Venn diagram displaying the overlaps showed that 139 of 248 OTUs were shared between recovery NJI and non-NJI at 1 month (Fig. 6a). To display microbiome space between recovery NJI and non-NJI among 0, 1, 3, 6 and 12 months, we performed PCoA analysis based on weighted UniFrac distance (Fig. 6b, Supplementary Table S22). This data revealed that distinct separation bacterial communities were present between recovery NJI and non-NJI at early age, while the microbial communities became more uniform over time.
A heatmap of the identi ed key variables were revealed and demonstrated that a total of 13 key OTUs were signi cantly different between recovery NJI and non-NJI at 1 month ( Supplementary Fig. 6, Supplementary Table S23). Fecal bacterial composition and difference at the phylum and genus levels between two groups at 1 month were shown in Fig. 6c&d&e&f, respectively (all P < 0.05, Supplementary  Table S24-27).
We detected the greatest differences in taxa between recovery NJI and non-NJI at 1 month using the LEfSe method and LDA scores, as shown in Fig. 7a and Supplementary Fig. 7a (all P < 0.05, LDA > 3, Supplementary Table S28).
The predominant fecal microbial functions between recovery NJI and non-NJI at 1 month were shown by a cladogram and LDA analysis (all P < 0.05, LDA > 2, Fig. 7b and Supplementary Fig. 7b, Supplementary Table S29).

Effect of medication on infant growth
These observations prompted us to explore the potential relationship between YCM treatment and infant growth. Thus, we examined whether weight-for-length z score at 12 month differed between recovery NJI and non-NJI. Using t-test for independent samples compare the development and growth of recovery NJI and non-NJI at the 12-Month-old. The infant weight-for-length z score were calculated according to World Health Organization standards. [18] Infant weight-for-length z score at 12 month did not differ signi cantly between recovery NJI and non-NJI (P < 0.05). Our study demonstrates that YCM treatment in early life is independent of growth at the 12 month of age.

Discussion
We illustrated that neonatal jaundice was associated with altered composition and function of gut microbiota, as well as decrease of α-diversity. Recent studies have reported that higher level of α-diversity was associated with lower risk of necrotizing enterocolitis, atopic eczema, and neonatal sepsis [19][20][21].
Our study suggested that altered microbial community might play an important role during neonatal jaundice initiation and development and higher α-diversity of gut microbiome could also be a protective factor for infants at risk of jaundice. Moreover, the bacterial community of neonate at risk of jaundice was separated from that of non-neonatal jaundice.
The gut microbiota is indispensable to the health of the host, healthy infants individuals may share some key microbiota structural features, whereas neonate at risk of jaundice may have aberrant patterns and lack some key bacteria, leading to a 'dysbiosis' state. LEfSe analyzed in neonatal jaundice risk present a decrease of some probiotics and butyrate-producing bacteria. Butyrate plays an important role in bacterial energy metabolism and intestinal mucosa health in humans, as the major energy source of the intestinal mucosa, and as an important regulator of gene expression, in ammation, differentiation and apoptosis in host cells [22][23][24][25]. It is noteworthy that short chain fatty acids (SCFAs) (particularly propionate and butyrate) initiate within the intestinal mucosa several complementary mechanisms issuing in the activation of intestinal gluconeogenesis [26]. LEfSe shown that some butyrate-producing bacteria, such as Blautia and Pseudobutyrivibrio were increased in non-NJI versus NJI. Moreover, members of the genus Blautia produce acetate, ethanol, hydrogen, lactate, or succinate that can provide the energy for the host [27]. A study indicated that Pseudobutyrivibrio, a butyrate, lactic acid and formic acid producer [28]. In addition, members of the genus Blautia produce acetate, ethanol, hydrogen, lactate, or succinate that can provide the energy for the host. Moreover, non-NJI harbor more bene cial populations such as Lachnospiraceae, one of the major taxonomic groups of the gut microbiota, which degrade complex polysaccharides to SCFAs, including acetate, butyrate, and propionate, which can be used by the host as energy [29]. These results indicated the gut microbial community alteration might play an key role during neonatal jaundice initiation. fructose-1,6-bisphosphatase III and Pyruvate carboxylase subunit B enzymes involved in gluconeogenesis decreased clearance of lactate. In addition, in infancy, the body has a great demand for energy and the body may produce more ATP through the glycolysis to maintain energy metabolism, resulting in increased lactate. Thus, the neonate with limited capacity to metabolism of lactate via the intestinal gluconeogenesis may be associated with neonatal jaundice.
Our study analyzed the effect of YCM treatment on gut microbiota of NJI at 0,1,3,6 and 12 months. We found that gut microbiota difference within the YCM treatment group were completely decreased over time, suggesting that YCM treatment can only temporarily perturbation of NJI gut microbiota. Moreover, we conducted a longitudinal study of recovery NJI and non-NJI, PCoA revealed that the microbial community of recovery NJI was clustered with that of non-NJI over time, suggesting microbiota of recovery NJI tends to be recovers to similar with non-NJI gradually, and effect of YCM treatment on gut microbiota is temporary. Importantly, infants growth and development at 12 month did not differ signi cantly between recovery NJI and non-NJI. Taken together, these results demonstrated that gut microbiota composition were in uenced by YCM treatment at early, these differences were absent over time and the gut microbiota gradually recovers. Thus, we propose that YCM treatment may have little long term effect on infant healthy.

Conclusions
This study comprehensively characterized gut microbiome in NJI and demonstrated the association between early meconium microbiome and subsequent diagnosis of neonatal jaundice. The combination of early gut microbiome intervention and currently used treatment methods may further bene t NJI. Moreover, we illustrated gut microbial alterations and development in NJI with treatment, which may provide a solid foundation for future health outcomes through microbiota intervention.

Author contributions
Prof JD and Ms XM conceptualized and designed the study, drafted the initial manuscript, and reviewed and revised the manuscript. Ms WL and QX, and Dr LH, XZ, AL, ZL and HR designed the data collection instruments, collected data, carried out the initial analyses, and reviewed and revised the manuscript. Dr ZR conceptualized and designed the study, coordinated and supervised data collection, and critically reviewed the manuscript for important intellectual content. All authors approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.                                    Identi cation of speci c bacterial taxa and microbial functions between between recovery NJI and non-NJI. (a) The greatest differences in taxa between recovery NJI and non-NJI are presented according to the LDA scores (log10). (b) Differences in gut microbial functions between recovery NJI and non-NJI based on the LDA scores (log10). NJI: neonatal jaundice infants; LDA: linear discriminant analysis.

Figure 7
Page 51/56 Identi cation of speci c bacterial taxa and microbial functions between between recovery NJI and non-NJI. (a) The greatest differences in taxa between recovery NJI and non-NJI are presented according to the LDA scores (log10). (b) Differences in gut microbial functions between recovery NJI and non-NJI based on the LDA scores (log10). NJI: neonatal jaundice infants; LDA: linear discriminant analysis.

Figure 7
Identi cation of speci c bacterial taxa and microbial functions between between recovery NJI and non-NJI. (a) The greatest differences in taxa between recovery NJI and non-NJI are presented according to the LDA scores (log10). (b) Differences in gut microbial functions between recovery NJI and non-NJI based on the LDA scores (log10). NJI: neonatal jaundice infants; LDA: linear discriminant analysis.

Figure 7
Identi cation of speci c bacterial taxa and microbial functions between between recovery NJI and non-NJI. (a) The greatest differences in taxa between recovery NJI and non-NJI are presented according to the LDA scores (log10). (b) Differences in gut microbial functions between recovery NJI and non-NJI based on the LDA scores (log10). NJI: neonatal jaundice infants; LDA: linear discriminant analysis. Figure 7