- Email: [email protected]
Dyeing to Learn More about the Gut Microbiota
Cell Host & Microbe
Previews Dyeing to Learn More about the Gut Microbiota Mohamed S. Donia1 and Michael A. Fischbach1,* 1Department of Bioengineering and Therapeutic Sciences and the California Institute for Quantitative Biosciences, University of California, San Francisco, San Francisco, CA 94158, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.chom.2013.01.011
The switch from culture-based enumeration to deep sequencing has enabled microbial community composition to be profiled en masse. In a new article, Maurice et al. (2013) report the use of fluorescence-activated cell sorting (FACS) to perform a high-throughput analysis of gut microbiota community function. The big advance that sets the latest wave of microbiome research apart from earlier studies is deep sequencing: the ability to enumerate all of the cells in a complex microbial community at once. The switch from a low-throughput technique, culture-based enumeration, to the high-throughput technology of deep sequencing enabled the ‘‘magic’’ of viewing community composition from 30,000 feet (Figure 1). Not surprisingly, many of the key insights from the last few years of microbiome research—spatiotemporal variation in the microbiome (Costello et al., 2009), the effect of diet on the gut community (Turnbaugh et al., 2009), development of the infant microbiome (Dominguez-Bello et al., 2010; Koenig et al., 2011; Yatsunenko et al., 2012), and the response of the gut community to antibiotic treatment (Dethlefsen and Relman, 2011)—would have been difficult to glean from culture-based studies. There is a great deal yet to learn about the microbiome from deep sequencing; many of the key questions that remain unanswered concern the temporal dynamics of the microbiota and disease-specific changes in community composition. Nevertheless, a consensus is emerging in the microbiome research community that questions about community composition—which are addressed by deep sequencing— should be accompanied by new lines of inquiry into community function (http:// grants.nih.gov/grants/guide/rfa-files/RFARM-12-021.html). Taking microbiome insights from bench to bedside, the argument goes, will require a molecular-level understanding of function: metabolism of dietary inputs, synthesis of diffusible molecules and surface antigens, and modulation of host signaling pathways. Such a detailed description of host-
microbiota interactions would reveal how the composition and function of the gut community relate to disease; how they can be modulated by small molecule drugs, probiotics, and prebiotics; and what the goals of those perturbations should be. Classically, the study of microbial function has been a low-throughput endeavor. Notable papers have explored the biological role of a single molecule produced by an individual microbial species (Mazmanian et al., 2005; Shin et al., 2011). An important exception has been a series of metabolomic studies of the microbiota from Nicholson and coworkers, which have highlighted key microbial metabolites and unexpected similarities and differences in function among gut microbial communities (Nicholson et al., 2012). In an exciting new manuscript, Turnbaugh and colleagues adapt a highthroughput technique pioneered for the analysis of aquatic microbial communities to study the metabolic state of the gut microbiota en masse (Figure 1) (Maurice et al., 2013). This technique consists of treating an intact microbial community (e.g., a human fecal sample) with the fluorescent dyes SYBR Green, propidium iodide (Pi), and DiBAC and using fluorescence-activated cell sorting (FACS) to determine the proportion of dye-positive versus dye-negative cells. SYBR Green binds to DNA and reports on the total quantity of DNA in a cell. The level of SYBR Green fluorescence can therefore distinguish between cells with high and low nucleic acid content (HNA and LNA, respectively); HNA cells are presumed to be actively dividing and/or to have an increased metabolic activity, while LNA cells are not. Propidium iodide is excluded by cells with an intact membrane, and DiBAC can enter depolarized cells,
so cells that are Pi+ or DiBAC+ are presumed ‘‘damaged.’’ The authors used this FACS-based assay to profile the metabolic activity of tens of thousands of cells from each of 21 fecal samples from three individuals, both fresh and after treatment with antibiotics or other drugs. Three of their findings are particularly notable. First, they show that while more than half of the cells in the community are active, 17% of cells are Pi+ and 27% are DiBAC+, indicating a sizable minority of damaged cells. By combining FACS with 16S rRNA sequencing, they could show that the active and damaged subsets are both dominated by members of the Firmicutes order Clostridiales, although different genera dominate the active and damaged subpopulations. These results imply a key functional difference between Firmicutes and the other major gut phylum, Bacteroidetes: different subsets of the Firmicutes are more likely to be metabolically active and dying, indicating a greater level of cell turnover among them than the Bacteroidetes. They also suggest the tantalizing finding that, in the authors’ words, ‘‘members of the gut microbiota may inhabit distinct ecological niches, defined not only by physical location and resource utilization but also by their level of metabolic activity.’’ Second, antibiotics have a direct and rapid physiological effect on the gut microbiota. Notably, the Pi+ subset doubled from 12% to 23%, and the DiBAC+ subpopulation increased from 33% to 44%. This finding, which had been suspected but not demonstrated, shows the uneven manner in which antibiotic treatment affects cells in the population over time. A panel of six host-targeted drugs did not increase the proportion of Pi+ and DiBAC+ cells, showing that this effect is specific to compounds that
Cell Host & Microbe 13, February 13, 2013 ª2013 Elsevier Inc. 119
Cell Host & Microbe
Previews
target bacteria. Surprisingly, enable reporter assays of Low Throughput High Throughput the proportion of HNA cells signal transduction or metais not significantly affected bolic gene function. However by antibiotic treatment, this new field unfolds, one Community indicating that even broadthing is certain: dyeing is just Composition spectrum antibiotics do not the beginning. Culture-Based have an immediate effect Deep Sequencing Enumeration on a substantial portion of ACKNOWLEDGMENTS actively dividing bacteria: more than half of the Work in the authors’ laboratory is supported by a Medical Research population. Community Program Grant from the W.M. Keck For the third notable Function Foundation, a Fellowship for Science finding, Turnbaugh and coand Engineering from the David workers turned from their and Lucile Packard Foundation, and grants from the NIH (OD007290, FACS-based analysis of AI101018, and AI101722). Single-Species Fluorescence Activated metabolic activity to metatranscriptomics, a commuREFERENCES nity-wide profiling technique Figure 1. An Evolution from Low- to High-Throughput Microbial with more precedent. They Community Profiling Techniques Costello, E.K., Lauber, C.L., HamThe low-throughput technique of culture-based enumeration has given way to show that treatment with antiady, M., Fierer, N., Gordon, J.I., deep sequencing, which enables a more thorough view of community compobiotics and other xenobiotics and Knight, R. (2009). Science 326, sition. Turnbaugh and coworkers introduce a high-throughput technique, 1694–1697. increases the expression of based on FACS, that enables microbiota metabolic function to be profiled en masse. genes involved in drug resisDethlefsen, L., and Relman, D.A. (2011). Proc. Natl. Acad. Sci. USA tance and metabolism; for 108(Suppl 1 ), 4554–4561. example, multiple antibiotics increased the expression of drug trans- tant phenotype or disease? FACS-seq will Dominguez-Bello, M.G., Costello, E.K., Contreras, porters, while the histamine H2-receptor prove to be especially useful, both as M., Magris, M., Hidalgo, G., Fierer, N., and Knight, R. (2010). Proc. Natl. Acad. Sci. USA 107, 11971– antagonist nizatidine induced the expres- a diagnostic and as a tool to study 11975. sion of genes that might be involved in the the molecular underpinnings of disease Koenig, J.E., Spor, A., Scalfone, N., Fricker, A.D., reduction of its terminal nitro group. etiology, if a microbiome-related disorder Stombaugh, J., Knight, R., Angenent, L.T., and Three important questions remain. is characterized by differences in commu- Ley, R.E. (2011). Proc. Natl. Acad. Sci. USA 108(Suppl 1 ), 4578–4585. First, how general are the findings? If nity metabolic activity. Importantly, the nearly all unperturbed gut communi- authors show that xenobiotic-perturbed Maurice, C.F., Haiser, H.J., and Turnbaugh, P.J. (2013). Cell 152, 39–50. ties—independent of composition— communities exhibit interindividual and show similar proportions of HNA and temporal differences in their active and Mazmanian, S.K., Liu, C.H., Tzianabos, A.O., and Kasper, D.L. (2005). Cell 122, 107–118. damaged cells, then Turnbaugh and damaged subsets, so it is likely that signifcoworkers’ results will be a true mile- icant differences will be seen in any Nicholson, J.K., Holmes, E., Kinross, J., Burcelin, R., Gibson, G., Jia, W., and Pettersson, S. (2012). stone, but their technique will be of limited disease state that results in a similar Science 336, 1262–1267. use in assaying functional differences perturbation (e.g., the oxidative stress of Shin, S.C., Kim, S.H., You, H., Kim, B., Kim, A.C., among communities. In contrast, if inflammation). Lee, K.A., Yoon, J.H., Ryu, J.H., and Lee, W.J. communities differ dramatically in their Third, can the authors’ approach (2011). Science 334, 670–674. proportions of HNA and damaged cells, be extended to other high-throughput, Turnbaugh, P.J., Ridaura, V.K., Faith, J.J., Rey, then this technique could be of broad community-wide measurements of func- F.E., Knight, R., and Gordon, J.I. (2009). Sci. Transl. Med. 1, ra14. utility for distinguishing among function- tion? Fluorophore-labeled metabolites ally distinct communities. might allow the metabolism of a specific Yatsunenko, T., Rey, F.E., Manary, M.J., Trehan, I., Dominguez-Bello, M.G., Contreras, M., Magris, M., Second, can the proportion of active dietary molecule to be read out, while Hidalgo, G., Baldassano, R.N., Anokhin, A.P., et al. and damaged cells be linked to an impor- anaerobic fluorescent proteins could (2012). Nature 486, 222–227.
120 Cell Host & Microbe 13, February 13, 2013 ª2013 Elsevier Inc.
Previews Dyeing to Learn More about the Gut Microbiota Mohamed S. Donia1 and Michael A. Fischbach1,* 1Department of Bioengineering and Therapeutic Sciences and the California Institute for Quantitative Biosciences, University of California, San Francisco, San Francisco, CA 94158, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.chom.2013.01.011
The switch from culture-based enumeration to deep sequencing has enabled microbial community composition to be profiled en masse. In a new article, Maurice et al. (2013) report the use of fluorescence-activated cell sorting (FACS) to perform a high-throughput analysis of gut microbiota community function. The big advance that sets the latest wave of microbiome research apart from earlier studies is deep sequencing: the ability to enumerate all of the cells in a complex microbial community at once. The switch from a low-throughput technique, culture-based enumeration, to the high-throughput technology of deep sequencing enabled the ‘‘magic’’ of viewing community composition from 30,000 feet (Figure 1). Not surprisingly, many of the key insights from the last few years of microbiome research—spatiotemporal variation in the microbiome (Costello et al., 2009), the effect of diet on the gut community (Turnbaugh et al., 2009), development of the infant microbiome (Dominguez-Bello et al., 2010; Koenig et al., 2011; Yatsunenko et al., 2012), and the response of the gut community to antibiotic treatment (Dethlefsen and Relman, 2011)—would have been difficult to glean from culture-based studies. There is a great deal yet to learn about the microbiome from deep sequencing; many of the key questions that remain unanswered concern the temporal dynamics of the microbiota and disease-specific changes in community composition. Nevertheless, a consensus is emerging in the microbiome research community that questions about community composition—which are addressed by deep sequencing— should be accompanied by new lines of inquiry into community function (http:// grants.nih.gov/grants/guide/rfa-files/RFARM-12-021.html). Taking microbiome insights from bench to bedside, the argument goes, will require a molecular-level understanding of function: metabolism of dietary inputs, synthesis of diffusible molecules and surface antigens, and modulation of host signaling pathways. Such a detailed description of host-
microbiota interactions would reveal how the composition and function of the gut community relate to disease; how they can be modulated by small molecule drugs, probiotics, and prebiotics; and what the goals of those perturbations should be. Classically, the study of microbial function has been a low-throughput endeavor. Notable papers have explored the biological role of a single molecule produced by an individual microbial species (Mazmanian et al., 2005; Shin et al., 2011). An important exception has been a series of metabolomic studies of the microbiota from Nicholson and coworkers, which have highlighted key microbial metabolites and unexpected similarities and differences in function among gut microbial communities (Nicholson et al., 2012). In an exciting new manuscript, Turnbaugh and colleagues adapt a highthroughput technique pioneered for the analysis of aquatic microbial communities to study the metabolic state of the gut microbiota en masse (Figure 1) (Maurice et al., 2013). This technique consists of treating an intact microbial community (e.g., a human fecal sample) with the fluorescent dyes SYBR Green, propidium iodide (Pi), and DiBAC and using fluorescence-activated cell sorting (FACS) to determine the proportion of dye-positive versus dye-negative cells. SYBR Green binds to DNA and reports on the total quantity of DNA in a cell. The level of SYBR Green fluorescence can therefore distinguish between cells with high and low nucleic acid content (HNA and LNA, respectively); HNA cells are presumed to be actively dividing and/or to have an increased metabolic activity, while LNA cells are not. Propidium iodide is excluded by cells with an intact membrane, and DiBAC can enter depolarized cells,
so cells that are Pi+ or DiBAC+ are presumed ‘‘damaged.’’ The authors used this FACS-based assay to profile the metabolic activity of tens of thousands of cells from each of 21 fecal samples from three individuals, both fresh and after treatment with antibiotics or other drugs. Three of their findings are particularly notable. First, they show that while more than half of the cells in the community are active, 17% of cells are Pi+ and 27% are DiBAC+, indicating a sizable minority of damaged cells. By combining FACS with 16S rRNA sequencing, they could show that the active and damaged subsets are both dominated by members of the Firmicutes order Clostridiales, although different genera dominate the active and damaged subpopulations. These results imply a key functional difference between Firmicutes and the other major gut phylum, Bacteroidetes: different subsets of the Firmicutes are more likely to be metabolically active and dying, indicating a greater level of cell turnover among them than the Bacteroidetes. They also suggest the tantalizing finding that, in the authors’ words, ‘‘members of the gut microbiota may inhabit distinct ecological niches, defined not only by physical location and resource utilization but also by their level of metabolic activity.’’ Second, antibiotics have a direct and rapid physiological effect on the gut microbiota. Notably, the Pi+ subset doubled from 12% to 23%, and the DiBAC+ subpopulation increased from 33% to 44%. This finding, which had been suspected but not demonstrated, shows the uneven manner in which antibiotic treatment affects cells in the population over time. A panel of six host-targeted drugs did not increase the proportion of Pi+ and DiBAC+ cells, showing that this effect is specific to compounds that
Cell Host & Microbe 13, February 13, 2013 ª2013 Elsevier Inc. 119
Cell Host & Microbe
Previews
target bacteria. Surprisingly, enable reporter assays of Low Throughput High Throughput the proportion of HNA cells signal transduction or metais not significantly affected bolic gene function. However by antibiotic treatment, this new field unfolds, one Community indicating that even broadthing is certain: dyeing is just Composition spectrum antibiotics do not the beginning. Culture-Based have an immediate effect Deep Sequencing Enumeration on a substantial portion of ACKNOWLEDGMENTS actively dividing bacteria: more than half of the Work in the authors’ laboratory is supported by a Medical Research population. Community Program Grant from the W.M. Keck For the third notable Function Foundation, a Fellowship for Science finding, Turnbaugh and coand Engineering from the David workers turned from their and Lucile Packard Foundation, and grants from the NIH (OD007290, FACS-based analysis of AI101018, and AI101722). Single-Species Fluorescence Activated metabolic activity to metatranscriptomics, a commuREFERENCES nity-wide profiling technique Figure 1. An Evolution from Low- to High-Throughput Microbial with more precedent. They Community Profiling Techniques Costello, E.K., Lauber, C.L., HamThe low-throughput technique of culture-based enumeration has given way to show that treatment with antiady, M., Fierer, N., Gordon, J.I., deep sequencing, which enables a more thorough view of community compobiotics and other xenobiotics and Knight, R. (2009). Science 326, sition. Turnbaugh and coworkers introduce a high-throughput technique, 1694–1697. increases the expression of based on FACS, that enables microbiota metabolic function to be profiled en masse. genes involved in drug resisDethlefsen, L., and Relman, D.A. (2011). Proc. Natl. Acad. Sci. USA tance and metabolism; for 108(Suppl 1 ), 4554–4561. example, multiple antibiotics increased the expression of drug trans- tant phenotype or disease? FACS-seq will Dominguez-Bello, M.G., Costello, E.K., Contreras, porters, while the histamine H2-receptor prove to be especially useful, both as M., Magris, M., Hidalgo, G., Fierer, N., and Knight, R. (2010). Proc. Natl. Acad. Sci. USA 107, 11971– antagonist nizatidine induced the expres- a diagnostic and as a tool to study 11975. sion of genes that might be involved in the the molecular underpinnings of disease Koenig, J.E., Spor, A., Scalfone, N., Fricker, A.D., reduction of its terminal nitro group. etiology, if a microbiome-related disorder Stombaugh, J., Knight, R., Angenent, L.T., and Three important questions remain. is characterized by differences in commu- Ley, R.E. (2011). Proc. Natl. Acad. Sci. USA 108(Suppl 1 ), 4578–4585. First, how general are the findings? If nity metabolic activity. Importantly, the nearly all unperturbed gut communi- authors show that xenobiotic-perturbed Maurice, C.F., Haiser, H.J., and Turnbaugh, P.J. (2013). Cell 152, 39–50. ties—independent of composition— communities exhibit interindividual and show similar proportions of HNA and temporal differences in their active and Mazmanian, S.K., Liu, C.H., Tzianabos, A.O., and Kasper, D.L. (2005). Cell 122, 107–118. damaged cells, then Turnbaugh and damaged subsets, so it is likely that signifcoworkers’ results will be a true mile- icant differences will be seen in any Nicholson, J.K., Holmes, E., Kinross, J., Burcelin, R., Gibson, G., Jia, W., and Pettersson, S. (2012). stone, but their technique will be of limited disease state that results in a similar Science 336, 1262–1267. use in assaying functional differences perturbation (e.g., the oxidative stress of Shin, S.C., Kim, S.H., You, H., Kim, B., Kim, A.C., among communities. In contrast, if inflammation). Lee, K.A., Yoon, J.H., Ryu, J.H., and Lee, W.J. communities differ dramatically in their Third, can the authors’ approach (2011). Science 334, 670–674. proportions of HNA and damaged cells, be extended to other high-throughput, Turnbaugh, P.J., Ridaura, V.K., Faith, J.J., Rey, then this technique could be of broad community-wide measurements of func- F.E., Knight, R., and Gordon, J.I. (2009). Sci. Transl. Med. 1, ra14. utility for distinguishing among function- tion? Fluorophore-labeled metabolites ally distinct communities. might allow the metabolism of a specific Yatsunenko, T., Rey, F.E., Manary, M.J., Trehan, I., Dominguez-Bello, M.G., Contreras, M., Magris, M., Second, can the proportion of active dietary molecule to be read out, while Hidalgo, G., Baldassano, R.N., Anokhin, A.P., et al. and damaged cells be linked to an impor- anaerobic fluorescent proteins could (2012). Nature 486, 222–227.
120 Cell Host & Microbe 13, February 13, 2013 ª2013 Elsevier Inc.