A new technique for use in the study of the microbiome: An evaluation of a three-dimensional cell culture technique in maintaining the gastrointestinal microbiome of four Balb/c female mice and implications for future studies

Fluctuations in oxygen, pH, nutrients, or other factors such as food or pharmaceuticals, may perturb the microbiota of the gastrointestinal (GI) tract. This environmental variation is a cause for concern given dysbiosis of the microbiome is correlated with disease states; thereby, model organisms are utilized to study microbial communities during, after, or before shifts in microbes since intact ex vivo microbiomes have historically been challenging to utilize. The objective of this study is to culture an explant microbiome of 4 Balb/c, laboratory bred mice to develop an ex vivo tool for future microbiome studies. We cultured homogenates of the distal colon of 4 mice in three dimensional, 24 well plate culture dishes. These dishes were incubated for 24 hours in two different oxygen concentration levels, 0% and 20%. The pH of the plate was tested before and after incubation. To analyze the integrity of the microbiome, we utilized 16S sequencing. Further, we utilized 16S metagenomics to characterize fecal samples and colon samples to speculate whether future studies may utilize feces in constructing an explant microbiome to spare animal lives. We found that pH and familial relationship had a profound impact on community structure while oxygen did not have a significant influence. The feces and the colon were similar in community profiles, which lends credence to utilizing feces in future studies. In addition, our efforts successfully cultured archaea, which included difficult to culture strains such as Miscellaneous Crenarchaeota group (MCG) and Methanobacteria. Ultimately, further attempts to culture and preserve an animal’s microbiome needs to control for and maintain stable pH.


Introduction (Level 1)
The microbiome forms a symbiotic relationship with its host [1]. Essentially, microbes have a 7 135 Interferences (CoNet) [36] application for Cytoscape [37]. Feces and colon data were removed 136 before CoNet analysis. CoNet has been utilized in previous studies to investigate defined 137 interactions between microbes [38][39][40]. Spearman correlation coefficient with a cutoff ratio of 138 0.6 was utilized, and to focus the network, only microbes with sequence counts greater or equal 139 to 20 were included. 1,000 permutations were accomplished through a bootstrapping method 140 with an FDR correction [39]. 141 Results (Level 1) 142 pH and oxygen (Level 2) 143 pH readings of each plate were taken before and after incubation. As shown in Table 1, pH 144 fluctuated from the original homogenate and baseline (before 24-hour incubation). In addition, 145 mouse samples maintained varying levels of pH which correlated with differences in oxygen 146 concentration (Table 1) mice succumbed to euthanasia, colon samples were harvested to complement fecal shedding. 153 Mouse 1, 2, and 4 shed two fecal samples each while mouse 3 shed three fecal samples; 154 therefore, we had a total of 9 colon and 9 fecal samples across 4 mice. After sequence quality 155 filtering we had a total sample size of 111 samples and 3,133,666 sequences total ( Fecal and colon comparison (Level 2) 159 The feces and the colon samples were characterized for microbial composition at the phylum and 160 family level (Figs 1A and 1B, respectively). Community composition was dominated by the 161 phyla Firmicutes and Bacteroidetes ( Fig 1A), with means of 47% and 49%, respectively, and 162 standard deviations (SD) of 23%. In addition, the Bacteroidales family S24-7 was highly 163 abundant with a mean of 42% and a SD of 20% ( Fig 1B). These results are consistent with recent 164 microbiome studies of mice [41][42][43]. Beta diversity analysis for each of the feces and colon 165 samples showed no difference in composition (Table 3), and alpha diversity analysis ( Fig 1C) 166 also revealed no difference (KW ANOVA, chi-squared = 6.64, df = 7, p = 0.47). Therefore, all 167 colon samples were pooled together, and all feces samples were pooled together for a statistical 168 comparison of feces and colon. The Shannon diversity index was utilized for comparison of the bulk samples ( Fig 1D). Results    In the explanted microbiomes, Firmicutes and Bacteroidetes were the dominant phyla (Fig 2A).

185
Firmicutes had the highest average relative abundance, 70% (SD 28%), with Bacteroidetes 186 averaging 18% (SD 16%). Across all cultures, Enterococcus was highly abundant (Fig 2B) 187 having a mean of 47% (SD 32%). The Shannon diversity index was different between the explanted cultures and the microbiome of the mice (KW ANOVA, chi-squared = 73.58, df = 8, p 189 < 0.01). Post-hoc analysis shows that, compared to the microbiome, mouse cultures 1 and 2 were 190 the same while mouse cultures 3 and 4 differed ( Fig 2C). The microbial profile of each sample 191 revealed a difference in beta diversity between the plates and the microbiome (Table 4), which is 192 also reflected in the PCoA plot ( Fig 2D).  communities also showed a marked difference between plates of varying pH levels (Table 5). 214 Additionally, CCA and PERMANOVA revealed that oxygen did not contribute to community 215 clustering (Table 6; Fig 4). The only significant effect fitted in the PERMANOVA is sibling relationship (fixed factor) while oxygen (fixed factor) is not significant. Sibling relationship explains the changes in community composition as time elapsed in the incubators.

240
Enterococcus significantly increased between mice cultures and the microbiome (Fig 6A). 241 Additionally, Proteobacteria strains were more abundant in cultures reaching a high pH; 242 although, cultures reaching pH 6 were equivalent in Proteobacteria compared to cultures 243 reaching pH 9 (Fig 6B). Enterococcus strains also followed a similar dynamic in which they 244 increased in cultures reaching a lower pH, 6 and 7 (Fig 6C). Subsequently, cultures with the 245 lowest pH, 6, had a significantly high abundance of Lactobacillus (Fig 6D). Archaea were more 246 abundant in plates with pH 9 and 10 (Fig 6E), and further, Clostridium strains were more likely 247 to be present in mouse 1 and 2 cultures compared to mouse 3 and 4 ( Fig 6F). partitioned by which mouse, M, it originated. "ns", non significant; "***", p < 0.001; "**", p < 255 0.01; "*", p ≤ 0.05.

256
Microbial network (Level 2) 257 The OTUs in the microbial network represent 88% of the relative sequence count for the cultured 258 well plates (Fig 7). Much of the interactions were positive in nature meaning copresence in a 259 shared-niche is the most abundant interaction type. Negative, mutually exclusive interactions are 260 only between OTUs from the genus Enterococcus and several OTUs from the order 261 Bacteroidales (Fig 7). The interaction between the 4 mutually exclusive OTUs account for 50% 262 of all sequences. Many studies of the microbiome utilize germ-free mice, which are expensive to house and breed 270 and require sacrificing animals to study the GI microbiome [14,16]. In this study, we attempted 271 to culture an ex vivo microbiome in 3D well plates to decrease the cost associated with studying 272 animal microbiomes. We found that cultures for mouse 1 and 2 were comparable to the gut 273 microbiome in Shannon diversity (Fig 2C), which is very promising. Ultimately, our explant 274 microbiome was significantly different than the in vivo microbiome, but we were able to culture 275 a diverse number of prokaryotic strains utilizing our method. Optimizing efforts in culture 276 media, detection, and atmospheric gradients is extremely important in culturing desired microbes [44]. With very little optimization, we were able to culture many gut microbial species including  (Fig 6E). 280 The upsurge of Enterococcus in cultured plates (Fig 6A) is explained by its competitiveness 281 outlined in Fig 7. A likely scenario is that Enterococcus outcompeted strains within the order  Post-incubation, a shift in pH was seen amongst the various 12 wells of the plates (Table 1). This 289 shift in pH is accounted for by the increase in gram negative Proteobacteria in plates with a more 290 basic pH (Fig 6B). Creation of the amine groups from this phylum perhaps led to the increase in 291 pH [55-57]. Further, the decrease in pH may be related to the increase in the lactic acid 292 fermenter Enterococcus in plates with a pH of 7 (Fig 6C) [52] while plates with a pH of 6 may 293 be partially explained by the increase in both Enterococcus and Lactobacillus (Fig 6C, D)

294
[52,58,59]. Additionally, pH was a strong influencer in the growth of archaea. Archaea grew 295 more readily in plates with a higher pH (Fig 6E). Not only are archaea difficult to culture, but

305
The microbiome is passed on from mother to litter [8,62]. Our ex vivo microbiome was highly 306 impacted by familial relationship (Table 6, Fig 4). Further, mice differed in the amount of  Essentially, our results indicate that the feces and large distal colon are highly similar; therefore, 314 it is reasonable to consider avoiding mouse sacrifice by culturing feces. Future experiments will 315 need to control pH shifts to avoid media-related population dynamics. Since none of the plates 316 maintained the original baseline pH or even homogenate pH, we assume that additional buffering 317 capacity or equilibrating media and culturing in a CO 2 incubator may create a more stable 318 explanted microbiome, possibly maintaining diversity more similar to in vivo microbiota. Even 319 with the pH swings seen here, we were able to culture bacteria that are difficult to routinely 320 culture. Ultimately, this study found that pH was a stronger influencer of community 321 composition than oxygen. pH has a strong influence on the establishment of the microbes that