香京julia种子在线播放

C57Bl6 mice (Jackson Laboratory) were ordered at 8 weeks of age and acclimated in the mouse facility for 2 weeks. Mice were maintained in a 12-h light/dark cycle with ad libitum water and regular NIH-07 chow diet (22.5% protein, 4.5% fat, 4.5% fiber). The animals were maintained and the experiment performed according to the animal protocols approved by the NIDDK Animal Care and Use Committee (NIH) (#K023-LCBB-16).

Mice were single-housed 1 week before the experiment. Pregnancy of C57Bl6 mice was assessed by visualization of the vaginal plug and defined as E0.5. Pregnant mice were fed a quarter pellet/day of Rodent BLT’s^TM (Bio-serv) onto which were pipetted either: (1) water (Ctrl); (2) 10 μL of water with 0.1 mg of sucralose and 0.25 mg of acesulfame-K [equivalent to upper limit of ADI for human consumption (ADI1x)]; (3) 20 μL of water with 0.2 mg of sucralose and 0.5 mg of acesulfame-K [twice the ADI (ADI2x)] (Figure 1A). The ADI for sucralose and acesulfame-K are 5 and 15 mg/kg of bodyweight/day, respectively. Pellets were added into a plastic cup in the cage every day and supplemented the regular chow diet. Pellet ingestion was assessed visually every day.

One day prior to sacrifice, mice were fasted overnight (16 h). Glycaemia was measured at the mouse tail using a Contour apparatus (Bayer). Mice were sacrificed after isoflurane anesthesia (Baxter).

Retro-orbital blood collection was performed under isoflurane anesthesia.

Following sacrifice, the entire digestive tract from stomach to rectum was dissected and flushed with 10 mL of phosphate buffer saline (PBS) buffer. The gut flush was then concentrated by collecting the retained fraction on 0.22 μm filter paper.

Milk collection was performed on gestational day 15. The litter was separated from the mother 2 h prior to milking. Under isoflurane anesthesia, mice were injected intraperitoneally with 2IU of oxytocin (VetOne). After a few minutes, milk was collected by squeezing each mammary gland (Willingham et al., 2014). All samples were flash-frozen and stored at -80°C.

Sucralose and acesulfame-K concentrations in plasma, urine, feces and breast milk were measured in triplicate using liquid chromatography-mass spectrometry (LC-MS). Analyses were performed with an Acquity I-Class UPLC (Waters Corp., Milford, MA, United States) and an Acquity UPLC BEH C-18 column (2.1 mm × 50 mm, 1.7 μm) coupled with a Q-Exactive MS (Thermo Scientific, Waltham, MA, United States) with a HESI-II electrospray source. Analyses were performed using isotopically labeled sucralose as the internal standard for sucralose measurements and isotopically labeled acesulfame-K for acesulfame-K measurements. The %CV for the samples throughout experiments was 4.5% for sucralose and 0.8% for acesulfame-K, respectively, and for samples between days was 5.8% for sucralose and 1.0% for acesulfame-K, respectively.

A non-targeted metabolomics analysis was performed on feces and plasma samples by Metabolon^®.

RNeasy Mini Kit (Qiagen) was used to extract RNA from 100 mg of liver. Four microliter of qScript cDNA Supermix (Quanta bioscience) was added to 500 μg of RNA extract and run on a thermocycler to provide cDNA using the following program: 25°C/5 min; 42°C/30 min; 85°C/5 min. cDNA quantity and quality were then assessed using nanodrop spectrophotometer. cDNA samples were analyzed with RT² Profiler PCR Array- Mouse Drug Metabolism: Phase I Enzymes or Mouse Drug Metabolism: Phase II Enzymes (Qiagen) according to the manufacturer protocol. Samples were processed on 7900HT Fast Real-Time PCR system (Applied Biosystems).

Feces DNA samples were obtained from gut flushes (see sample collection) using the Fast DNA Spin Kit feces (MPBio). DNA quantity and quality were then assessed using a Nanodrop spectrophotometer.

Library preparation for sequencing of the V3-V4 region of 16S ribosomal DNA was performed using two PCRs. First, 16S DNA was amplified using 1 μg of feces DNA, 2× MyTaq Red Mix (Bioline) and the following primers: 16S Forward: TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGG NGGCGGCWGCAG; 16S Reverse: GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACH VGGGTATCTAATCC. Amplification was performed on a thermocycler using the following program: 95°C/3 min; 95°C/30 s; 55°C/30 s; 72°C/30 s (repeat three last steps 35×); 72°C/5 min; 4°C/∞. To assess PCR quality, samples were run on 2% agarose E-gel (Invitrogen). Then, 2 μL of PCR product was used as a matrix for the second “adapters” PCR. Each 16S amplicon was then amplified using a unique couple of barcode-containing primers for dual-indexing of samples during the sequencing process (Fadrosh et al., 2014). Addition of the adapters on 16S amplicons was checked on 2% agarose gel. Finally, amplicon bands were extracted from agarose gels using QIAquick Gel Extraction Kit (Qiagen). This step cleaned up the final PCR product and removed eventual high molecular weight primer dimers, thus increasing the sequencing yield. A mixed sample of about 30 samples (1:1 ratio) was used for sequencing.

Sequencing was performed by the sequencing core (NIDDK- National Institutes of Health) using MiSeq Reagent Kit v2 (600-cycle) (Illumina).

Sequencing output was first de-multiplexed and sequences were then analyzed with macQIIME workflow in “open reference OTU picking” protocol. Alpha- and Beta-diversity (Principal Coordinate Analysis- PCoA) were performed on R Studio using the phyloseq package.

Two sets of primers were used to determine the quantity of A. muciniphila in feces samples (5→3): A. muciniphila Forward: CAGCACGTGAAGGTGGGGAC; A. muciniphila Reverse: CCTTGCGGTTGGCTTCAGAT; 16S Forward: TCGTCGGCAGCGTCAGATGTGTATAAGAG ACAGCCTACGGGNGGCGGCWGCAG; 16S Reverse: GTCTCGTGGGCTCGGAGATGTGTATAAG AGACAGGACTACHVGGGTATCTAATCC. One μg of DNA was mixed with fast SYBR Green mastermix (Applied Biosystems), two primers (100 μM) and amplified on 7900HT Fast Real-Time PCR system (Applied Biosystems). Each determination was performed in triplicate and A. muciniphila abundance was measured relative to 16S expression.

Spearman correlation, t-test and ANOVA analysis, Principal Component Analysis and Volcano plots were performed and plotted using R. The scripts for the analysis were posted in GITHUB¹ .

Akkermansia muciniphila was purchased from ATCC (BAA-835) and maintained under anaerobic condition at 37°C either in Remel PRAS Brain Heart Infusion Broth (Thermo Scientific) or on Trypticase Soy Agar plates with Sheep Blood (BD BBL). A. muciniphila growth experiment was conducted in 96-well plate liquid culture by measuring the absorbance at 600 nm on the POLARstar Omega spectrophotometer (BMG Labtech). Media was supplemented with 100 μg/mL of ampicillin, 100 ng/mL of acesulfame-K and 35 ng/mL of sucralose (“low NNS”) or 100 μg/mL of acesulfame-K and 35 μg/mL of sucralose (“high NNS”).

Microbiome and metabolomics data have been published on Open Science Framework (OSF)².

All scripts for analysis and/or plotting in this article have been published on GitHub¹.

In brief, C57Bl6 mice (n = 31) were fed either water (control), or a combination of sucralose and acesulfame-K at the ADI (Sucralose ADI: 5 mg/kg/day; Acesulfame-K ADI: 15 mg/kg/day) or twice the ADI (ADI2x) during pregnancy and lactation (Figure 1A). NNS concentrations were measured in mothers’ samples (blood, feces, breastmilk) after fasting and sacrifice post-weaning (Supplementary Figures S1A,B and Supplementary Table S1). While no NNS were observed in the control group, NNS were found in mothers’ blood and feces in both NNS groups. As expected, based on previous pharmacokinetic studies, sucralose was mainly found in mother’s feces whereas acesulfame-K was found in equal amounts in feces and blood. Both sweeteners were found in breastmilk (d15) at considerably lower concentrations than in blood and/or feces, and concentrations were highly variable between individual animals. A total of 226 20-day old pups distributed across the 3 groups were analyzed in the present study. Pups’ blood and feces samples were collected at d20 after fasting and sacrifice (Supplementary Figures S1C,D and Supplementary Table S1). Overall, both NNS were either unmeasurable or present at low concentrations in the pups’ blood and feces compared to their mother’s. Sucralose was found mostly in the feces and was roughly correlated with the amount of sucralose in mothers’ breast milk. Acesulfame-K concentrations were below the detection limit in pups’ blood and feces, but measurable in the pups’ urine when analyzed in newborns and 14 day-old pups (Supplementary Figure S1E and Supplementary Table S1). These measurements confirmed that both NNS were transmitted prenatally. Litter size and sex distribution were comparable in NNS vs. Ctrl pups (Supplementary Table S2). Measurement of the pups’ fasting glucose at d20 showed a decrease in ADI2x pups (p = 0.02) compared to water, while a non-significant difference was observed between the ADI1x and Ctrl groups (p = 0.16) (Figure 1B). The weight of the ADI2x pups was also reduced compared to Ctrl animals with weight curves gradually separating in the first week of life (Figure 1C).

To further define the consequences of the NNS exposure, we analyzed both feces and plasma metabolites in the pups using a non-targeted metabolomics approach. The analysis highlighted 594 and 708 metabolites in plasma and feces, respectively, across 18 samples, (6 in each group). In the NNS pups’ plasma, more than 100 metabolites were deregulated compared to the Ctrl pups and almost 200 were deregulated in feces at p <= 0.05 threshold (Supplementary Figure S2). Volcano plots representing changes of both ADI1x and ADI2x vs. control group revealed extensive metabolic deregulation in both NNS groups (Figure 2A,B and Supplementary Tables S3, S4). Compared to Ctrl, NNS feces samples showed the most deregulated metabolites (at least 5 up and 35 down) compared to plasma (at least 3 up and 9 down) for both ADI1x and ADI2x pups. All three groups of pups separated significantly in a principal component analysis (PCA) and by ADONIS test (feces r² = 0.30 and p < 0.001; plasma r² = 0.28 and p < 0.001), highlighting distinct metabolomes (Figure 2C,D). Of note, a pool of metabolites was uniquely deregulated in ADI2x compared to ADI1X (plasma: 59 ups and 28 downs; feces: 107 up and 58 down), suggesting dose-dependent metabolic deregulation.

Amino acid metabolism was the most deregulated class of metabolites in NNS pups (Figure 3A, 2A,B and Supplementary Tables S3, S4). Along with primary amino acid metabolites, numerous acetylated, sulfated or methylated amino acids as well as products of amino acid metabolism (picolinate, indole-3-carboxylic acid) were downregulated in NNS pups; N-acetyl cadaverine and tyramine, products of amino acid breakdown, were highly upregulated. Many glycine-containing dipeptides were also significantly decreased in NNS pups. Similarly, numerous products of glycine metabolism were downregulated such as bile acids, heme breakdown products, hippurate and propionylglycine.

Both carbohydrate and lipid metabolism were found to be altered in the NNS pups (Figure 3A, 2A,B and Supplementary Tables S3, S4). The presence of sucralose as well as the decreased plasma glucose levels in NNS fed pups were also confirmed by metabolomics. GlcNAc-6-sulfate was at least 6 times higher in pups’ feces in both NNS groups, suggesting a glycosaminoglycan processing deficiency following NNS exposure. In the lipids, modified glycerylphosphorylcholine and glycerylphosphoethanolamine as well as myoinositol were significantly downregulated in NNS exposed pups while structural lipids such as phosphoethanolamine, hexadecasphinganine and sphinganine were upregulated.

Numerous deregulated metabolites were associated with improper liver clearance and detoxification pathways (Figure 3A, 2A,B and Supplementary Tables S3, S4). Of note, both primary and secondary bile acids, N-acetylated amino acids and sulfated metabolites were downregulated in NNS pups. In addition, accumulation of cys-glutathione disulfide was consistent with unprocessed reactive intermediary metabolites normally removed through phase 2 detoxification (Figure 3A).

Metabolic detoxification occurs primarily in the liver and is divided into 3 phases: transformation, conjugation and transport (Almazroo et al., 2017) (Figure 3A). To confirm the unbiased metabolomics findings that NNS pups have altered detoxification metabolism, gene expression was analyzed in the livers of control and ADI1x pups by phase 1 and 2 drug metabolism qPCR array (Figure 3B and Supplementary Tables S5, S6). Transcriptomics of 20-day old pups’ livers showed downregulated transcripts encoding enzymes in phase 2 metabolism. Indeed, the bile acid-CoA: amino acid N-acetyl transferase Baat, the acyl-CoA synthetase Acsl3, the glycine modifying enzyme Gnmt and glutathione S-transferase Gstm5 were highly downregulated. These transcriptional results are consistent with improper phase 2 detoxification and accumulation of intermediary metabolites. Furthermore, whitening of NNS pups’ livers was consistently observed in the ADI2x group and to a lesser degree in the ADI1x group (Supplementary Figure S3A). Intriguingly, common histologic analysis of the NNS livers did not reveal macroscopic damage to the liver (Supplementary Figure S3B – H&E and Masson). By immunohistochemistry analysis, there was no evidence of either altered glycogen (PAS), lipid droplets (H&E), iron (Perl) or bile (Hall) storages in the offspring of NNS exposed mothers (Supplementary Figure S3B).

In addition to major changes in metabolism, metabolomics analysis identified deregulation of specific microbial metabolites (Figure 2, 3A). These observations suggested major changes in the microbiome in addition to amino acid metabolic deregulation in the NNS pups. Many bacterial metabolites (Gottschalk, 2012) differed in their abundance in NNS pups such as N-formyl-methionine, 6-oxolithocholate, 7-ketodeoxycholate, hippurate, daidzein and genistein. Considering the existing literature on NNS and adult microbiome, we hypothesized that NNS could also alter the pups’ gut microbiome.

Bacteroidetes, verrucomicrobia and firmicutes represented 44.34, 30.53, and 23.62% of the Ctrl pups’ microbiota, respectively (Figure 4A and Supplementary Table S7). In NNS pups, firmicutes doubled (ADI1x = 44.45%; ADI2x = 45.91%), including clostridiales families lachnospiraceae and ruminococcaceae (g_Oscillospira). In contrast, verrucomicrobia, represented by a single species A. muciniphila, was significantly depleted in NNS pups (ADI1x = 2.97%; ADI2x = 6.88%). These observations were confirmed by analysis of the beta-diversity principal coordinate analysis (PCoA), showing separation of Ctrl, ADI1x and ADI2x pups (Figure 4B). All three groups of pups also separated significantly by ADONIS test (Unweighted Unifrac: r² = 0.07 and p < 0.001; Weighted Unifrac: r² = 0.18, p < 0.001) (Supplementary Figure S4D). A significant increase in alpha-diversity was also observed for both NNS groups, suggesting an increased variety of the gut microbiota (Figure 4C and Supplementary Figure S4E). This may be attributed to the loss of the highly abundant A. muciniphila in the gut in combination with increase of the firmicutes species.

Offspring’s microbiome dysbiosis is often explained by changes in the mother’s microbiome. Indeed, the microbiota is shaped from birth by successive exposure to the mother’s vaginal, skin and breast milk microbiomes (Palmer et al., 2007). Consequently, the same phyla were found in the mothers’ and pups’ microbiome (Supplementary Figure S4A and Supplementary Table S8). Ctrl and NNS-fed mothers’ microbiota were not statistically different. Similarly, Ctrl and NNS mothers’ microbiomes did not separate on the alpha or beta-diversity analysis (PCoA plot) or by ADONIS test (Supplementary Figures S4B–E). Therefore, the divergence between Ctrl and NNS pups’ microbiota cannot be explained by an alteration in the mothers’ microbiota. Mother-to-pup’s transmission can be analyzed by calculation of Spearman correlation coefficients for the different groups. A decrease in correlation was observed with NNS exposure, suggesting that the divergence between mothers’ and pups’ microbiomes were due to increasing NNS concentrations (Figure 4D).

Alteration of NNS pups’ microbiome also suggested metabolic imbalance, which correlated with the metabolomic data. Indeed, an increased firmicutes/bacteroidetes ratio is associated with obesity (Ley et al., 2005) while depletion of A. muciniphila is frequently a sign of metabolic dysregulation (Everard et al., 2013; Schneeberger et al., 2015; Figure 4E).

A closer look at pups’ A. muciniphila level and their corresponding mothers by qPCR confirmed an improper transmission in NNS pups (Figure 5A). It is important to understand why A. muciniphila colonization was defective in NNS fed pups since this bacterium is widely regarded as beneficial (Everard et al., 2013; Schneeberger et al., 2015). However, A. muciniphila level analyzed by qPCR showed various levels single housed mothers 1 day before mating (Supplementary Figure S5). Breastfeeding is one of the major sources of bacteria and probiotics for the pups. Nevertheless, analysis of the newborn pups showed similar level of A. muciniphila in Ctrl and ADI1x animals right after first nursing (d1), suggesting that mothers properly transmitted A. muciniphila at birth to their pups (Figure 5B). However, pups’ A. muciniphila level over time (Figure 5C) showed an improper progression of colonization in the ADI1x group. Indeed, after the establishment of gut anaerobic environment (Pantoja-Feliciano et al., 2013), A. muciniphila colonized the Ctrl but not the ADI1x pups.

We then directly exposed A. muciniphila to various NNS concentrations or ampicillin and measured bacterial growth in vitro (Figure 5D). While ampicillin completely inhibited A. muciniphila growth, no difference was observed between control or NNS supplemented media suggesting that A. muciniphila can grow in the presence of NNS.

We found that NNS exposure of pregnant and nursing mothers causes critical changes in amino acid and detoxification metabolism in the progeny, identified by a non-targeted metabolomics analysis of feces and blood samples. Eleven of the significantly deregulated metabolites in NNS pups are part of standard screening tests for inborn errors of metabolism (IEM) in humans (El-Hattab, 2015). IEMs are mostly due to mutations in specific enzymes or transport proteins of important metabolic pathways. Frequently observed symptoms in IEM are growth failure, weight loss, developmental delay, liver failure and/or hypoglycemia, many of which are observed in the NNS exposed pups. Despite different etiologies (genetic mutation vs. environmental exposure), this suggests similarly interrupted processes of nutrient processing and detoxification.

Among the various deregulated metabolic pathways, the pups’ amino acid metabolism is most affected particularly the glycine metabolism. Glycine is used for synthesis of glutathione, bile acids, heme, hippurate and propionylglycine (Wang et al., 2013), all under-represented in the metabolomics data. Glycine shortage has been associated with suboptimal growth, impaired immune system function (Lewis et al., 2005) as well as many other conditions (Wijekoon et al., 2004). NNS pups also have a striking deficit in numerous compounds related to detoxification. A relationship with NNS has been suggested previously, demonstrated in adult rats by the upregulation of the multidrug resistance transporter P-gp and the cytochrome P450 Cyp3a4 and Cyp2d1 following sucralose exposure (Abou-Donia et al., 2008). In our hands, Cyp3a11, the mouse homolog of Cyp3a4, is also upregulated but did not reach statistical significance, potentially due to the short exposure. Furthermore, several phase 2 detoxification metabolism enzymes are downregulated in NNS pups. Among the deregulated enzymes is bile acid-CoA: amino acid N-acetyl transferase (Baat), which catalyzes the addition of glycine or taurine to unconjugated bile acids, consistent with the observed decrease of primary bile acids (O’Byrne et al., 2003). Decreased glycine modifying enzyme Gnmt expression is consistent with a deregulated glycine metabolism and lower Acyl-CoA synthetase Acsl3 expression also implies an inefficient phase 2 metabolism. Finally, glutathione-S transferase Gstk1 plays a major role in phase 2 detoxification participating in the maintenance of oxidative balance (Deponte, 2013). The balance between phase 1 and phase 2 enzymes is critical for proper detoxification. Food, smoking, alcohol consumption, age and genetics are known factors impacting detoxification enzymes producing intermediary harmful metabolites faster than they can be detoxified, ultimately leading to increased oxidative stress and tissue damage (Liska, 1998). Our data suggest that sucralose and/or acesulfame-K also belong to these pathogenic factors. This is consistent with a previous study showing an association of sucralose consumption and microbiome-dependent liver inflammation (Bian et al., 2017a). Since our findings were most impressive in the pups, we speculate that the pre- and postnatal period represent a highly vulnerable developmental phase (Abou-Donia et al., 2008). At 20-days postpartum, pups are still developing their detoxification systems and metabolism. At this critical developmental stage exposure to NNS might be more consequential. Also, 20-days-old pups have a limited nutritional influx, mostly breast milk. At weaning, the increased variety of nutrients may further challenge metabolic detoxification. Indeed, endotoxins, end products of metabolism, bacterial as well as intermediary metabolites might accumulate, creating a toxic environment impossible to overcome.

Following pre- and post-natal exposure to NNS, pups have a dramatically different microbial composition, which is not due to alteration of mother’s microbiome. Indeed, mothers and pups’ microbiome are expected to look alike, consequences of a high microbe exchange through vaginal and physical contact, breastfeeding and murine coprophagic behavior (Faith et al., 2013; Bäckhed et al., 2015).

Studies in adult mice have previously shown NNS’ detrimental effects on the intestinal microbiome, mostly increases in clostridiales, lactobacilli, and enterobacteria (Abou-Donia et al., 2008; Cowan et al., 2013; Palmnäs et al., 2014; Suez et al., 2014; Bian et al., 2017b). A bacteriostatic effect has also been suggested in microbes in the oropharynx (Prashant et al., 2012). The microbiome signature of the NNS pups is a large increase in firmicutes combined with a marked depletion of Verrucomicrobia, represented exclusively by A. muciniphila. Both of these alterations are clinically relevant and have been associated with increased incidence of obesity and/or metabolic syndrome in rodents and humans (Ley et al., 2005; Hansen et al., 2012; Schneeberger et al., 2015). The increase in firmicutes observed in NNS pups is mainly due to the clostridiales order, the lachnospiraceae family and oscillospira genus. Those three clostridia class members are known to participate in normal metabolism of bile acids, carbohydrates and amino acids and participate in colonic epithelium health by producing short chain fatty acids (Donohoe et al., 2011). Therefore, increases in clostridia can influence colonocyte homeostasis. Interestingly, increases in exposed pups’ phospholipids and sphingolipids also are consistent with an increase gut cell proliferation.

Akkermansia muciniphila, depleted in NNS pups, is a beneficial bacterium associated with a young and lean metabolism, inversely correlated with fat mass gain, type 1 diabetes and Inflammatory bowel syndrome (Ley et al., 2005; Hansen et al., 2012; Schneeberger et al., 2015). A decrease in A. muciniphila was previously observed in response to long saccharin exposure in adult rats (Suez et al., 2014). Interestingly, A. muciniphila encodes a GlcNAc-6-sulfatase and the reduction in A. muciniphila in the NNS-pups’ gut is consistent with the abundance of GlcNAc-6-Sulfate in NNS pups’ feces. In vitro culture of A. muciniphila is not affected by the presence of NNS but the colonization is clearly deficient in the pups. Therefore, the improper colonization of A. muciniphila may be the result of a change in the mucin profile, fucosylated milk oligosaccharides and/or gut pH due to changes in metabolites, microbe abundance or inflammation status (Aakko et al., 2017;Van Herreweghen et al., 2017).

Metabolomics analysis is also suggestive of substantial gut microbiome alterations. Among the deregulated metabolites, N-formyl-methionine is used for protein synthesis in bacteria only (Gottschalk, 2012). Secondary bile acids are also exclusively produced by gut microbes from primary bile acid release in the GI tract. A decrease in secondary bile acids similar to ours, has been associated with decreased fasting glucose in mice (Kuno et al., 2018). Heme breakdown products biliverdin and bilirubin are also transformed by microbes to produce urobilinogen. From dietary aromatic compounds, the gut microbiome also metabolizes benzoate, the primary source of hippurate. Finally, gut microbes convert natural isoflavone glycosides into daidzein and genistein, which are readily absorbed. Gut microbiome function is also tightly linked to detoxification mechanisms. Indeed, microbial metabolites are detoxified by the host (Spanogiannopoulos et al., 2016) and in a complementary fashion, the gut microbiota transforms drugs and bile acids (Wahlström et al., 2016). Therefore, improper detoxification might be the result or the consequence of microbiome dysbiosis.