The Food Additive Xanthan Gum Drives Adaptation of the Human Gut Microbiota

The diets of industrialized countries reflect the increasing use of processed foods, often with the introduction of novel food additives. Xanthan gum is a complex polysaccharide with unique rheological properties that have established its use as a widespread stabilizer and thickening agent1. However, little is known about its direct interaction with the gut microbiota, which plays a central role in digestion of other, chemically-distinct dietary fiber polysaccharides. Here, we show that the ability to digest xanthan gum is surprisingly common in industrialized human gut microbiomes and appears to be contingent on the activity of a single bacterium that is a member of an uncultured bacterial genus in the family Ruminococcaceae. We used a combination of enrichment culture, multi-omics, and recombinant enzyme studies to identify and characterize a complete pathway in this uncultured bacterium for the degradation of xanthan gum. Our data reveal that this keystone degrader cleaves the xanthan gum backbone with a novel glycoside hydrolase family 5 (GH5) enzyme before processing the released oligosaccharides using additional enzymes. Surprisingly, some individuals harbor a Bacteroides species that is capable of consuming oligosaccharide products generated by the keystone Ruminococcaceae or a purified form of the GH5 enzyme. This Bacteroides symbiont is equipped with its own distinct enzymatic pathway to cross-feed on xanthan gum breakdown products, which still harbor the native linkage complexity in xanthan gum, but it cannot directly degrade the high molecular weight polymer. Thus, the introduction of a common food additive into the human diet in the past 50 years has promoted the establishment of a food chain involving at least two members of different phyla of gut bacteria.


32
The diets of industrialized countries reflect the increasing use of processed foods, often with the 33 introduction of novel food additives. Xanthan gum is a complex polysaccharide with unique 34 rheological properties that have established its use as a widespread stabilizer and thickening 35 agent 1 . However, little is known about its direct interaction with the gut microbiota, which plays 36 a central role in digestion of other, chemically-distinct dietary fiber polysaccharides. Here, we 37 show that the ability to digest xanthan gum is surprisingly common in industrialized human gut 38 microbiomes and appears to be contingent on the activity of a single bacterium that is a member  To identify enzymes responsible for XG hydrolysis and determine their cellular location, 193 we grew three independent cultures in liquid medium containing XG and subjected cell-free  Table 3). 205 While most of the proteins were either detected in low amounts or lacked functional predictions 206 consistent with polysaccharide degradation, one of the most abundant proteins across all three 207 samples was one of two GH5 enzymes (RuGH5a) encoded in the previously identified R. GH5+CBMs, hereafter referred to as RuGH5a for simplicity). All but the full-length construct 221 yielded reasonably pure proteins, but only constructs with the GH5 and all three CBMs showed 222 activity on xanthan gum, suggesting a critical role in catalysis for these CBMs. (Extended Data 223 6). The alternate GH5 (RuGH5b) was also expressed in a variety of forms but did not display any 224 activity on XG (Extended Data 6). 225 Analysis of the reaction products showed that RuGH5a releases pentasaccharide  where this locus is in the process of streamlining or expanding. Additional support for the 262 involvement of this locus in XG degradation is provided by RNA-seq based whole genome 263 transcriptome analysis, which showed the induction of genes in this cluster when the community 264 was grown on XG compared to another polysaccharide (polygalacturonic acid, PGA) that also 265 supports R. UCG13 abundance (Extended Data 9). 266 267 B. intestinalis cross-feeds on XG oligosaccharides with its xanthan utilization PUL 268 Although R. UCG13 was recalcitrant to culturing efforts, we isolated several bacteria 269 from the original consortium, including a representative strain of the Bacteroides intestinalis that 270 was the most abundant ( Figure 1c) and also harbors a highly expressed candidate PUL for XG 271 degradation (Figure 2 or all of the sugars contained in the XGOs (Figure 4a). Consistent with the candidate B. 280 intestinalis XGOs PUL being involved in this phenotype, all of the genes in this locus were 281 activated >100-fold (and some >1000-fold) during growth on XGOs compared to a glucose-282 grown reference (Figure 4b). Whole genome RNA-seq analysis of the B. intestinalis strain 283 grown on XGOs revealed that the identified PUL was the most highly upregulated in the 284 genome, further validating its role in metabolism of XGOs (Extended Data 9). Interestingly,

285
RuGH5a XGOs treated with PL8 continued to support B. intestinalis growth, but tetramer 286 generated from the P. nanensis GH9 and PL8 failed to support any growth (Extended Data 9). 287 Growth was rescued in the presence of glucose but not in the presence of RuGH5a XGOs to 288 upregulate the PUL (Extended Data 9), suggesting that either the B. intestinalis transporters or 289 enzymes are incapable of processing this isomeric substrate.

290
To further test the role of the identified B. intestinalis PUL in XGOs degradation, we 291 tested recombinant forms of its constituent enzymes for their ability to degrade XGOs, and 292 confirmed the activity of several enzymes. The carbohydrate esterase domain C-terminal to the 293 PL-CE bimodular protein was able to remove acetyl groups from acetylated xanthan 294 pentasaccharides (Extended Data 8). While we were unable to detect xanthan lyase activity for 295 the PL-CE enzyme on full length XG or oligosaccharides it is likely that this enzyme or another 296 lyase acts to remove the terminal mannose residue since the GH88 was able to remove the 297 corresponding 4,5 unsaturated glucuronic acid residue from the corresponding tetrasaccharide 298 that would be generated by its action (Extended Data 8). The GH92 was active on the 299 trisaccharide produced by the GH88 as observed by loss of the trisaccharide and formation of 300 cellobiose (Extended Data 8). Finally, the GH3 was active on cellobiose, but did not show 301 activity on either tri-or tetra-saccharide, suggesting that this enzyme may be the final step in B.  310 To determine if the original consortium was representative of all our XG-degrading 311 cultures, we performed metagenomic sequencing on 20 additional XG-degrading communities 312 and retrieved 16 high-quality and 3 low-quality R. UCG13 MAGs as well as an unbinned contig 313 affiliated with R. UCG13 (Supplemental Table 2). We found that the R. UCG13 XG utilization 314 locus is extremely well conserved across these cultures with only one variation in gene content, 315 the insertion of a GH125 coding gene, and >95% amino acid identity (Extended Data 10). The  as hunter-gatherers, that are less likely to be exposed to this food additive. Using each locus as a 332 query, we searched several publicly available fecal metagenome datasets collected from 333 populations worldwide. All modern populations sampled displayed some presence of the R.

334
UCG13 XG locus, with the Chinese and Japanese cohorts being the highest (up to 51% in one 335 cohort) ( Figure 5). The B. intestinalis locus was less prevalent, with two industrialized 336 population datasets (Japan and Denmark/Spain) lacking any incidence. Where the locus was 337 present, its prevalence ranged from 1-11%. The three hunter-gatherer or non-industrialized   Table 4). We also found 12 hits for the B. intestinalis XGOs  polysaccharides. These data suggest that the R. UCG13 XG locus is more broadly present in 387 mammalian gastrointestinal microbiomes and can at least be recovered through XG-feeding. and our finding that R. UCG13 can colonize infants at an early age highlight the profound 397 impacts that XG may be having on the assembly, stability, and evolution of industrialized human 398 microbiomes.

399
The discovery of XG loci in an environmental sample and mouse microbiome, raises 400 ecological questions about the transfer and evolution of XG utilization between host and non-401 host associated environments. Although the mouse microbiome with a XG locus could have been 402 exposed to XG through herbivory of X. campestris infected plants, mice are affiliated with 403 human activities as pests and XG is used as a food additive in various domesticated animal 404 foodstuffs (e.g. in calf milk replacers 44 ), further solidifying a link between these loci and human 405 activities. Since XG is a naturally biosynthesized exopolysaccharide, it is also intriguing to 406 speculate about the role of R. UCG13's XG locus with respect to exopolysaccharides that other 407 microbes may be producing locally in the gut.   427 The original culture was isolated from a survey of 80 healthy adults using a bacterial culture 428 strategy designed to enrich for members of the Gram-negative Bacteroidetes, a phylum that 429 generally harbors numerous polysaccharide-degrading enzymes 23 . The original culture was the 430 only XG-degrading culture isolated from this initial survey, likely due to its bias for 431 Bacteroidetes. For subsequent surveys and further culturing fecal samples were collected into 432 pre-reduced phosphate buffered saline, then transferred to an anaerobic chamber (10% H2, 5% 433 CO2, and 85% N2; Coy Manufacturing, Grass Lake, MI) maintained at 37°C. Fecal suspensions 434 were used to inoculate cultures and passaged using partially Defined Medium (DM), which was 435 generally prepared as a 2x stock then mixed 1:1 with 10 mg/mL carbon source (e.g. xanthan 436 gum). Each L of prepared DM medium (pH=7.2) contained 13.6 g KH2PO4 (Fisher, P284),  Samples were bead beaten on high for 2-3 minutes with a Mini-BeadBeater-16 (Biospec 465 Products, USA), then centrifuged at 18,000 g for 5 mins. The aqueous phase was recovered and 466 mixed by inversion with 500 µL of phenol:chloroform, centrifuged at 18,000 g for 3 mins, and 467 the aqueous phase was recovered again. The sample was mixed with 500 µL chloroform, 468 centrifuged, and then the aqueous phase was recovered and mixed with 0.1 volumes of 3 M 469 sodium acetate (pH 5.2) and 1 volume isopropanol. The sample was stored at -80 C for ≥30 470 mins, then centrifuged at ≥20,000 g for 20 mins at 4 C. The pellet was washes with 1 mL room 471 temperature 70% ethanol, centrifuged for 3 mins, decanted, and allowed to air dry before  Sequencing FASTQ files were analyzed using mothur (v.1.40.5) 46 using the Silva reference 493 database 11 . OTUs with the same genus were combined and displayed using R 47 with the 494 packages reshape2 48 , RColorBrewer 49 , and ggplot2 50 . 496 An overnight culture was serially diluted in 2x DM. Serial dilutions were split into two 50 mL 497 tubes and mixed 1:1 with either 10 mg/mL xanthan gum or 10 mg/mL monosaccharide mixture 498 (4 mg/mL glucose, 4 mg/mL mannose, 2 mg/mL sodium glucuronate), both of which also had 1 499 mg/mL L-cysteine. Each dilution and carbon source was aliquoted to fill a full 96-well culture   P73 Xeon with data stored to 2Tb SSD), followed by base calling using Guppy v3.2.10 in 'fast' 553 mode. This generated in total 3.59 Gb of data. The Nanopore reads were further processed using 554 Filtlong v0.2.0 (https://github.com/rrwick/Filtlong), discarding the poorest 5% of the read bases, 555 and reads shorter than 1000 bp.

556
The quality processed Nanopore long-reads were assembled using CANU 62 v1.9 with the 557 parameters corOutCoverage=10000 corMinCoverage=0 corMhapSensitivity=high 558 genomeSize=5m redMemory=32 oeaMemory=32 batMemory=200. An initial polishing of the 559 generated contigs were carried out using error-corrected reads from the assembly with  Bellefonte, PA) with the following conditions: injector volume, 2 μl; injector temperature, 240 605 ºC; detector temperature, 300 ºC; carrier gas (helium), velocity 1.9 meter/second; split ratio, 1:2; 606 temperature program was 160 ºC for 6 min, then 4 ºC/min to 220 ºC for 4 min, then 3 ºC/min to 607 240 ºC for 5 min, and then 11 ºC/min to 255 ºC for 5 min.  Each culture fraction was mixed 1:1 with 5 mg/mL xanthan gum and incubated at 37 C for 24 617 hours. Negative controls were prepared by heating culture fractions to 95 C for 15 mins, then 618 centrifuging at 13,000 g for 10 mins before the addition of xanthan gum. All reactions were 619 halted by heating to ≥85 C for 15 mins, then spun at 20,000 g for 15 mins at 4 C. Supernatants 620 were stored at -20 C until analysis by thin layer chromatography.         using RuGH5a) and were carried out at pH 6.0. Reactions were incubated overnight at 37°C, 732 halted by heating at ≥ 95°C for 5-10 minutes, and centrifugation at ≥20,000 g for 10 mins. isolates for growth on xanthan oligosaccharides. Some isolates (e.g. Parabacteroides distasonis) 770 required the inclusion of 5 mg/mL beef extract (Sigma, B4888) to achieve robust growth on simple monosaccharides; in these cases, beef extract was included across all carbon conditions.

772
Unless otherwise specified, carbon sources were provided at a final concentration of 5 mg/mL.

773
Isolates were grown overnight in TYG media, subcultured 1:50 into DM -BE -glucose and grown 774 overnight, then subcultured 1:50 into DM -BE with either various carbon sources. Final cultures 775 were monitored for growth by measuring increase in absorbance (600 nm) using 96-well plates 776 as previously described. phase at OD600 ~0.85 whereas XG cultures were harvested at late-log phase at OD600 ~1.2 to 796 allow liquification of XG, which was necessary to extract RNA from these cultures. As before, 797 cultures were harvested by centrifugation, mixed with RNA Protect (Qiagen) and stored at -80 798 °C until further processing. RNA was purified as before except that multiple replicates of DM-799 XG RNA were pooled together and concentrated with Zymo RNA Clean and Concentrator TM -25 800 to reach acceptable concentrations for RNA depletion input. rRNA was depleted twice from the 801 purified total RNA using the MICROBExpress TM Kit, each followed by a concentration step using the Zymo RNA Clean and Concentrator TM -25. About 90% rRNA depletion was achieved 803 for all samples. B. intestinalis RNA was sequenced using NovaSeq and community RNA was 804 sequenced using MiSeq. The resulting sequence data was analyzed for differentially expressed 805 genes following a previously published protocol 76 . Briefly, reads were filtered for quality using           X a n t h a n