The neighborhood a child grows up in may influence their health for years to come in previously invisible ways.
A long-term study of 2,000 children born in England and Wales and followed to age 18 found that young adults raised in communities marked by more economic deprivation, physical dilapidation, social disconnection and danger display differences in the epigenome – the proteins and chemical compounds that regulate the activity of their genes.
The researchers say the study lends support to the hypothesis that gene regulation may be one biological pathway through which neighborhood disadvantage “gets under the skin” to engender long-term health disparities.
The differences were identified in genes previously linked to chronic inflammation, exposure to tobacco smoke, outdoor air pollution, and lung cancer and may put these people at risk for poorer health later in life.
Epigenetic differences remained even after taking into account the socioeconomic conditions of children’s families, and were seen in young adults who did not smoke or display evidence of high inflammation.
“These findings may help explain how long-term health disparities among communities emerge,” said Aaron Reuben, a Ph.D. candidate at Duke who was the study’s lead author.
“They also tell us that children who look the same physically and are otherwise healthy may enter adulthood wired at the cellular level for different outcomes in the future.”
It’s not possible to know yet whether these differences are lasting or could be modified, Reuben said. “That is something we will need to continue to evaluate.”
The study, appearing this month in the journal JAMA Network Open, drew from diverse data sources to characterize the physical, social, economic, and health and safety characteristics of children’s neighborhoods across their childhood and adolescence.
Data were gathered from local government and criminal justice databases, systematic observation of neighborhood conditions (via Google Street View) and detailed surveys of neighborhood residents. Researchers combined this high-resolution multi-decade neighborhood data with epigenetic information derived from blood drawn from participants at at age 18.
“The research is an important reminder that geography and genes work together to shape our health,” said Avshalom Caspi, the Edward M. Arnett Professor of Psychology & Neuroscience at Duke and a senior author on the study.
In a journal commentary that accompanied the study, psychiatric epidemiologist at Harvard Medical School Erin Dunn noted that neighborhood-induced gene regulation differences “are likely implicated in many adverse health outcomes, spanning from mental health disorders to cancer, obesity, and metabolic diseases.”
She writes, “I hope that studies like this by Reuben and colleagues will prompt researchers to explore these complex concepts and to bridge social determinants of health with epigenetic processes.”
The microbial community in the human digestive tract, the gut microbiota, is highly complex and also displays strong variation across individuals (Human Microbiome Project Consortium. 2012. Structure, function and diversity of the healthy human microbiome. Nature 486:207–214. doi:10.1038/nature11234,–Qin J, Li R, Raes J, Arumugam M, Burgdorf KS, Manichanh C, Nielsen T, Pons N, Levenez F, Yamada T, Mende DR, Li J, Xu J, Li S, Li D, Cao J, Wang B, Liang H, Zheng H, Xie Y, Tap J, Lepage P, Bertalan M, Batto JM, Hansen T, Le Paslier D, Linneberg A, Nielsen HB, Pelletier E, Renault P, Sicheritz-Ponten T, Turner K, Zhu H, Yu C, Li S, Jian M, Zhou Y, Li Y, Zhang X, Li S, Qin N, Yang H, Wang J, Brunak S, Dore J, Guarner F, Kristiansen K, Pedersen O, Parkhill J, Weissenbach J, et al. . 2010. A human gut microbial gene catalogue established by metagenomic sequencing. Nature 464:59–65. doi:10.1038/nature08821.).
Variability in gut microbiota composition is related to many factors, including medication, diet, and genetics (Blekhman R, Goodrich JK, Huang K, Sun Q, Bukowski R, Bell JT, Spector TD, Keinan A, Ley RE, Gevers D, Clark AG. 2015. Host genetic variation impacts microbiome composition across human body sites. Genome Biol 16:191. doi:10.1186/s13059-015-0759-1.,–Yatsunenko T, Rey FE, Manary MJ, Trehan I, Dominguez-Bello MG, Contreras M, Magris M, Hidalgo G, Baldassano RN, Anokhin AP, Heath AC, Warner B, Reeder J, Kuczynski J, Caporaso JG, Lozupone CA, Lauber C, Clemente JC, Knights D, Knight R, Gordon JI. 2012. Human gut microbiome viewed across age and geography. Nature 486:222–227. doi:10.1038/nature11053.). The gut microbiota has a variety of functions within the host, such as metabolism of certain compounds (Backhed F, Ley RE, Sonnenburg JL, Peterson DA, Gordon JI. 2005. Host-bacterial mutualism in the human intestine. Science 307:1915–1920. doi:10.1126/science.1104816,–Rowland I, Gibson G, Heinken A, Scott K, Swann J, Thiele I, Tuohy K. 2017. Gut microbiota functions: metabolism of nutrients and other food components. Eur J Nutr doi:10.1007/s00394-017-1445-8), and its composition is correlated with several diseases, such as Crohn’s disease and colorectal cancer (Burns MB, Lynch J, Starr TK, Knights D, Blekhman R. 2015. Virulence genes are a signature of the microbiome in the colorectal tumor microenvironment. Genome Med 7:55. doi:10.1186/s13073-015-0177-8.,–Sinha R, Ahn J, Sampson JN, Shi J, Yu G, Xiong X, Hayes RB, Goedert JJ. 2016. Fecal microbiota, fecal metabolome, and colorectal cancer interrelations. PLoS One 11:e0152126. doi:10.1371/journal.pone.0152126).
In mice, certain microbial communities can lead to changes in the host’s weight and overall health, suggesting that there is a reciprocal effect between the host and the gut microbiota (Goodrich JK, Waters JL, Poole AC, Sutter JL, Koren O, Blekhman R, Beaumont M, Van Treuren W, Knight R, Bell JT, Spector TD, Clark AG, Ley RE. 2014. Human genetics shape the gut microbiome. Cell 159:789–799. doi:10.1016/j.cell.2014.09.053.).
Recent work in mice has explored host gene expression and gene regulation in response to microbiome exposure (Camp JG, Frank CL, Lickwar CR, Guturu H, Rube T, Wenger AM, Chen J, Bejerano G, Crawford GE, Rawls JF. 2014. Microbiota modulate transcription in the intestinal epithelium without remodeling the accessible chromatin landscape. Genome Res 24:1504–1516. doi:10.1101/gr.165845.113.,–Donohoe DR, Garge N, Zhang X, Sun W, O’Connell TM, Bunger MK, Bultman SJ. 2011. The microbiome and butyrate regulate energy metabolism and autophagy in the mammalian colon. Cell Metab 13:517–526. doi:10.1016/j.cmet.2011.02.018). These studies suggested that the microbiome can induce both epigenetic changes and binding of specific transcription factors (Camp JG, Frank CL, Lickwar CR, Guturu H, Rube T, Wenger AM, Chen J, Bejerano G, Crawford GE, Rawls JF. 2014. Microbiota modulate transcription in the intestinal epithelium without remodeling the accessible chromatin landscape. Genome Res 24:1504–1516. doi:10.1101/gr.165845.113., Davison JM, Lickwar CR, Song L, Breton G, Crawford GE, Rawls JF. 2017. Microbiota regulate intestinal epithelial gene expression by suppressing the transcription factor Hepatocyte nuclear factor 4 alpha. Genome Res 27:1195–1206. doi:10.1101/gr.220111.116, Krautkramer KA, Kreznar JH, Romano KA, Vivas EI, Barrett-Wilt GA, Rabaglia ME, Keller MP, Attie AD, Rey FE, Denu JM. 2016. Diet-Microbiota interactions mediate global epigenetic programming in multiple host tissues. Mol Cell 64:982–992. doi:10.1016/j.molcel.2016.10.025.,–Semenkovich NP, Planer JD, Ahern PP, Griffin NW, Lin CY, Gordon JI. 2016. Impact of the gut microbiota on enhancer accessibility in gut intraepithelial lymphocytes. Proc Natl Acad Sci U S A 113:14805–14810. doi:10.1073/pnas.1617793113.).
In humans, work has been done to correlate microbiota composition, cellular response, and phenotype, thus illuminating the interplay between microbe and host. For example, studies have shown that expression and RNA splicing changes correlate with changes in the gut microbiota in vivo (Haberman Y, Tickle TL, Dexheimer PJ, Kim MO, Tang D, Karns R, Baldassano RN, Noe JD, Rosh J, Markowitz J, Heyman MB, Griffiths AM, Crandall WV, Mack DR, Baker SS, Huttenhower C, Keljo DJ, Hyams JS, Kugathasan S, Walters TD, Aronow B, Xavier RJ, Gevers D, Denson LA. 2014. Pediatric Crohn disease patients exhibit specific ileal transcriptome and microbiome signature. J Clin Invest 124:3617–3633. doi:10.1172/JCI75436. ,–Morgan XC, Kabakchiev B, Waldron L, Tyler AD, Tickle TL, Milgrom R, Stempak JM, Gevers D, Xavier RJ, Silverberg MS, Huttenhower C. 2015. Associations between host gene expression, the mucosal microbiome, and clinical outcome in the pelvic pouch of patients with inflammatory bowel disease. Genome Biol 16:67. doi:10.1186/s13059-015-0637-x. ).
However, in humans, it is challenging to perform large-scale studies and to account for environmental effects, such as host diet, health status, and medication use.
Recently, we described an in vitro approach based on human epithelial cells inoculated with live microbial communities (Richards AL, Burns MB, Alazizi A, Barreiro LB, Pique-Regi R, Blekhman R, Luca F. 2016. Genetic and transcriptional analysis of human host response to healthy gut microbiota. mSystems 1:e00067-16. doi:10.1128/mSystems.00067-16.) that is well suited to the study of the effects of the microbiome on human gene regulation.
In short, we cultured human colonocytes under hypoxic conditions in order to recapitulate the gut environment. We then exposed the colonocytes to live gut microbiotas derived from human fecal samples under hypoxic conditions and measured the levels of response of host cells via transcriptome sequencing (RNA-seq).
Using this technique, we identified differentially expressed (DE) genes that were enriched among genes associated with microbiome-related diseases, such as obesity (Goodrich JK, Waters JL, Poole AC, Sutter JL, Koren O, Blekhman R, Beaumont M, Van Treuren W, Knight R, Bell JT, Spector TD, Clark AG, Ley RE. 2014. Human genetics shape the gut microbiome. Cell 159:789–799. doi:10.1016/j.cell.2014.09.053, Turnbaugh PJ, Backhed F, Fulton L, Gordon JI. 2008. Diet-induced obesity is linked to marked but reversible alterations in the mouse distal gut microbiome. Cell Host Microbe 3:213–223. doi:10.1016/j.chom.2008.02.015.), and among genes that were differentially expressed in germ-free mice exposed to gut microbiota (28). The advantage of this system is that we can study human cellular response to gut microbiota in a highly scalable way.
Here we sought to use this in vitro system to determine the extent to which and the mechanism by which interindividual variations in microbiome composition drive differences in gene expression in the host cells.
In order to assess the mechanism of cellular response to microbiotas, we considered chromatin accessibility via ATAC-seq (Buenrostro JD, Giresi PG, Zaba LC, Chang HY, Greenleaf WJ. 2013. Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat Methods 10:1213–1218. doi:10.1038/nmeth.2688), which utilizes the Tn5 transposase for fragmentation and tagging of accessible DNA. Similar to DNase-seq, ATAC-seq can capture open chromatin regions, which are regions accessible to transcription factor (TF) binding.
Changes in chromatin accessibility can modulate transcription factor binding and thereby influence gene expression (Semenkovich NP, Planer JD, Ahern PP, Griffin NW, Lin CY, Gordon JI. 2016. Impact of the gut microbiota on enhancer accessibility in gut intraepithelial lymphocytes. Proc Natl Acad Sci U S A 113:14805–14810. doi:10.1073/pnas.1617793113. , Ackermann AM, Wang Z, Schug J, Naji A, Kaestner KH. 2016. Integration of ATAC-seq and RNA-seq identifies human alpha cell and beta cell signature genes. Mol Metab 5:233–244. doi:10.1016/j.molmet.2016.01.002. ). The specific TFs binding in regions of open chromatin can be identified through the motif sequence model and footprinting analysis of ATAC-seq data.
We also sought to determine if specific microbial taxa drive gene expression variation. These open issues are crucial for understanding the causal role of the microbiome in host physiology and for designing targeted therapies revolving around interventions in the gut microbiome.
Exposure to microbiota influences host gene expression.
To determine the impact of variation in the gut microbiota on host cells, we treated human colonic epithelial cells (HCoEpiC) with live gut microbiota extracted from 5 healthy, unrelated human individuals (Fig. 1A; see also Fig. S1A to D in the supplemental material). The composition of these samples is representative of that other healthy gut microbiome samples from the Human Microbiome Project (Fig. S1E) (Human Microbiome Project Consortium. 2012. Structure, function and diversity of the healthy human microbiome. Nature 486:207–214. doi:10.1038/nature11234, Human Microbiome Project Consortium. 2012. A framework for human microbiome research. Nature 486:7402215–7402221).
We then separately assessed changes in gene expression and microbial composition following 1, 2, and 4 h of exposure. The overall changes in gene expression among microbiome treatments and controls are clustered first by time point (by principal-component analysis and by hierarchical clustering as shown in Fig. 1B [see also Table S1 in the supplemental material]), where the strongest response (3,240 genes across any of the five microbiota samples) occurred at 2 h following exposure (see equation 3 in Text S1 in the supplemental material; Benjamini-Hochberg false-discovery rate [BH FDR] < 10%, |log2fold change [FC]| > 0.25).
Among these, we identified 669 transcripts (188 genes) that were differentially expressed in all five treatments following 2 h of treatment (Fig. 1C; see also 1-h and 4-h comparisons in Fig. S1F and G).
To identify genes whose expression levels changed consistently across the treatments at each time point, we removed the individual effect from the model and considered the 5 microbiota treatments as replicates.
Two representative genes, PDLIM5 and DSE, that were found to be differentially expressed at each time point are shown in Fig. 1D. Notably, the results from analysis of those genes show that we were able to identify various expression patterns through this model as long as the 5 microbiota treatments led to the same response at a given time point (Fig. S1H and I; see also Table S1).
These 5,413 genes with shared expression changes were enriched for genes that function in protein translation, as well as those associated with the cell surface, such as in adherens junctions (BH FDR < 10−6) (Fig. S1J; see also equation 2 in Text S1), suggesting a biological function that may relate to the host cell’s interaction with the microbiota.
Supplemental methods and results: supplementary text for materials and methods. Download Text S1, PDF file, 0.2 MB.
Impact of microbiota exposure on host gene expression. (A) Relative abundances of the bacterial phyla in each of the five different microbial communities at 0 (uncultured), 1, 2, and 4 h with and without (+/-) coculturing with colonocytes. (B) Alpha diversity rarefaction plots for each of the microbial communities in this study. Error bars represent the standard deviations of the results from the multiple rarefaction trials. (C) Principal-coordinate analysis (PCoA; unweighted UniFrac) of the samples showed a clear separation by microbial community (left panel), while neither coculture conditions (center panel) nor time (right panel) had a dramatic effect on the community structures. (D) PCoA of samples using weighted UniFrac, colored by microbial community. (E) PCoA representing HMP samples and samples from the current study, with the HMP fecal samples indicated in red and all other HMP samples in gray. The five fecal samples used in the current study are indicated in yellow. The top panel shows PC1 (x axis) versus PC2 (y axis), while the bottom panel shows PC1 (x axis) versus PC3 (y axis). (F) Plots generated by the use of the UpSetR package to show comparisons of genes that were differentially expressed at 1 h following exposure to each of the five microbiota. (G) Comparison of differentially expressed genes across the five treatments at 4 h. (H) Plot generated by the use of the UpSetR package to show comparisons of genes across time points in the model where the five treatments were treated as replicates. The blue bars to the left indicated the total number of differentially expressed genes in the given set. The gray vertical bars show the number of genes that are in the set denoted below them. Sets with a single dark gray circle show the number of differentially expressed genes unique to that sample. (I) Heat map showing gene expression changes from the model in which the five microbiota treatments at a given time point were used as replicates. Purple denotes an increase in gene expression (green shows a reduction) compared to the gene expression in colonocytes cultured alone. Only genes that were differentially expressed in at least one time point are indicated here. (J) Plot showing gene ontology enrichment for genes that were differentially expressed at any time point in the model in which each treatment was considered to represent a replicate (and in which the treatments led to similar expression changes). The top 10 enrichments are shown in this plot, excluding categories where the expected number of genes was less than 10 or greater than 500. The size of the point corresponds to the –log10 of the adjusted P value. Download FIG S1, PDF file, 1.1 MB.
Differentially expressed genes with various models. The table lists all transcripts that were differentially expressed in each of 5 models. (A) Model considering transcripts that were DE at each time point and with each treatment. (B) Model considering transcripts that were DE at each time point (with the 5 microbiota treatments considered replicates). (C) Model considering transcripts that were DE across treatments as determined using the likelihood ratio test. (D) Model considering transcripts that were DE relative to the mean-centered baseline abundance of the denoted taxon. (E) Model considering transcripts that were DE following coculture of colonocytes with the microbiome plus various concentrations of Collinsella aerofaciens. Genes were considered differentially expressed if at least one transcript was differentially expressed. Download Table S1, TXT file, 16.8 MB.
Each microbiota sample was derived from a different individual with a unique diet and genetic makeup. Therefore, we expect that the microbial composition and diversity of all samples differ. Examining the uncultured microbiomes, we found variability in their microbial composition and diversity (Simpson’s index values ranged between 0.94 and 0.98).
In contrast, we found that the microbial communities in the cultured microbiomes did not change dramatically over time and that the microbiomes maintained their individual-specific composition characteristics during culture (Fig. S1A to C).
Next, we considered how the human colonocytes influence microbial composition, and we found that most taxa were unaffected by the presence of human cells, while 13 taxa (of 112 tested) showed various levels of abundance dependent on the presence of host cells (likelihood ratio test; BH FDR < 10%) (examples are shown in Fig. S2A and B, table shown in Fig. S2C).
In order to determine how the compositions of the microbiota samples influenced host gene expression differently, we utilized a likelihood ratio test to compare models that included or excluded the individual microbiota effect. This test is able to incorporate gene expression changes over time and to compare the trajectories of expression in response to the 5 different microbiota treatments.
In this way, we identified 409 genes (1,484 transcripts) (likelihood ratio test; BH FDR < 10%; see Table S1) with expression patterns that significantly differed in response to the five microbiota samples. Two examples of representative genes with differing gene expression patterns are WNT7A and SPARC (Fig. 1E).
These examples demonstrate that no two microbiota samples induce the same response in the genes identified by the likelihood ratio test and further showed that the genes had different responses to the same treatment.
These data demonstrate that the host and the microbiota influence each other and that interindividual variation in the microbiome can lead to different gene expression responses in interacting host cells.
FIG S2 – Changes to the microbiome and concordance of taxon model and Collinsella spike-in. (A and B) Genera (Faecalibacterium [A] and Bacteroides [B]) whose changes in abundance following coculturing were inconsistent in a manner dependent on the microbiota sample in which they were found. Each color denotes samples exposed to a particular microbiota sample as follows: red, Ind1; blue, Ind2; yellow, Ind3; green, Ind4; teal, Ind5. (C) Table denoting 13 taxa whose abundance was affected by culturing with colonocytes as determined by the likelihood ratio test in comparisons between models with and without a treatment effect (BH FDR < 10%). (D and E) Correlations within the microbial taxon clusters shown in Fig. 2A. (D) Microbes positively associated with genes in cluster 2. (E) Microbes associated with genes in cluster 1. Red edges indicate a positive correlation. Blue edges indicate a negative correlation. Edge thickness indicates the magnitude of the correlation. Correlations and corresponding significance data from comparisons between pairs of microbial taxa were generated using SparCC, which estimates linear Pearson correlations, with 100 bootstrap iterations (72). Correlations were filtered by the use of a BH FDR of <5% and a correlation magnitude of >0.5. (F) QQ plot of P values from the model of the five-microbiota experiment corresponding to abundances of baseline Collinsella (gray). The red points indicate the same values as those indicated in gray but that were grouped in subsets to only include the 1,570 differentially expressed genes (DEG) in the validation spike-in experiment performed with Collinsella aerofaciens. Download FIG S2, PDF file, 0.2 MB.
More information: Aaron Reuben et al, Association of Neighborhood Disadvantage in Childhood With DNA Methylation in Young Adulthood, JAMA Network Open (2020). DOI: 10.1001/jamanetworkopen.2020.6095