The human microbiome is one of the fastest growing areas of research in the past two decades (Sommer and Bäckhed, 2013; Gilbert et al., 2018). Many gut bacteria have been shown to modulate physiological functions and pathologies (Gilbert et al., 2016; Elinav et al., 2019; Lynch et al., 2019). Nevertheless, mechanistic understanding is still lacking (Fischbach, 2018; Lawson et al., 2019; Vrancken et al., 2019). Various tools were developed to visualize and detect gut commensals (Valm et al., 2011; Earle et al., 2015; Welch et al., 2016, 2017; Tropini et al., 2017; Whitaker et al., 2017; Sheth et al., 2019). Commonly used fluorescent labeling methods, such as Green Fluorescent Proteins (GFP) and derivatives, require oxygen to mature and fluoresce and hence are inapplicable to live monitoring of gut bacteria in mammals, which are mostly anaerobes. To overcome this challenge, we recently developed a method to fluorescently label live, culturable, and anaerobic gut bacteria with fluorophores that do not require oxygen to mature and fluoresce. This method was based on metabolic oligosaccharide engineering and bioorthogonal click chemistry (Geva-Zatorsky et al., 2015). Briefly, bacterial strains are grown in the presence of azido-modified sugars. During the bacterial growth, these azido-sugars are incorporated in bacterial biomolecules. Then, the bacteria are incubated with fluorescently tagged cyclooctynes, which covalently react with the azide group, turning the bacteria fluorescent (Figure 1A). This method, however, is limited to studying only a few bacteria at a time, while the gut microbiota consists of a plethora of bacterial species residing in functional multi-species communities. Here, we report development of COMBInatorial cliCK-chemistry (COMBICK) labeling of human-derived gut bacteria that enables studying diverse communities of live gut symbionts in their natural environments.

In COMBICK we metabolically label each bacteria with a combination of different fluorophores. Each combination allows unique identification of the corresponding bacteria. The number of labeling combinations depends on the number of fluorophores available (“n”), the machine specifications, and the number of fluorophores used to label each bacteria (“k”), and can be calculated by the formula: (nk)=n!k!⁢(n-k)!. For example, with 5 fluorophores available, and each bacteria labeled with a combination of two, 10 different bacteria can be simultaneously differentiated, according to the formula: (52)=5!2!*⁢3!=10. Downstream methods such as flow cytometry and confocal microscopy could be used to differentiate and analyze the labeled bacterial populations (Figures 1B, 2A).

To apply this method, we first chose one bacteria, Bifidobacterium longum, and applied 2-color COMBICK labeling using 5 different fluorophores (AF594, AF430, TAMRA, AF647, and AF488) (Figure 1B). This created 10 different populations of B. longum that could be differentiated by flow cytometry and analyzed either by manual, classical analysis (Supplementary Figure 1), or by semi-supervised, semi-automatic analysis as applied here (Figures 1C–E). Briefly, for each fluorescent channel we created an analogous binary parameter. The binary parameter equals 1 if the single bacteria was positive for the relevant channel otherwise it equals 0. Then, a new parameter was derived, which contains the number of positive fluorophores for each event. Using this parameter, we could discard unlabeled bacteria and bacteria aggregates such that only relevant events were gated. Next, based on the binary parameters we performed t-SNE analyses for dimensionality reduction to identify bacterial populations according to their fluorophores profile. We confirmed the applicability of COMBICK dual-labeling on 15 phylogenetically diverse human-derived gut commensals (Supplementary Figure 2 and Supplementary Table 1).

Using COMBICK to study bacterial consortia is not restricted to flow cytometry only, but can also be applied with other methods such as confocal microscopy (Figure 2A). By confocal microscopy we could differentiate between 6 different dual-labeled B. longum populations using 4 fluorophores (AF405, AF488, AF594, and AF647) (Figure 2B).

Recent studies have demonstrated the importance of the microbiome at the strain level (Yang et al., 2020; Hajjo and Geva-Zatorsky, 2021). These studies emphasized functional differences of different strains of the same species, albeit their genomes are largely similar. However, technologies that enable studying different strains in parallel are limited. COMBICK enables us to differentiate between several strains of the same species simultaneously. To validate this, we applied COMBICK on several different strains of B. fragilis in a multi-strain community (Supplementary Table 2). To do so, we further developed COMBICK to enable labeling of each bacteria with 1, 2, or 3 fluorophores simultaneously. Each strain was labeled with a unique combination (all possible combinations with these 3 fluorophores: single-, dual-, or triple-labeled). This enabled us to differentiate between 7 different strains of B. fragilis at once using merely 3 fluorophores, utilizing all possible combinations simultaneously (Figures 3A–C).

We further developed COMBICK to enable high-order-fluorophore labeling. We could successfully label single bacteria with combinations of 4 distinct fluorophores simultaneously (Figure 3D). Theoretically, with ‘n’ fluorophores available, COMBICK enables labeling of bacterial populations with combinations of 1, 2, 3 … and up to ‘n’ fluorophores simultaneously. The major limit is the available fluorophores, and machine specifications (i.e., the number of fluorophores the microscopes or flow cytometers can simultaneously detect). Hence, the total number of distinct bacteria that could be differentiated by COMBICK is 2ⁿ−1 (Figures 3E,F). The number of bacteria that could be labeled simultaneously grows exponentially depending on the number fluorophores used according to the formula: (n1)+(n2)+(n3)+⋯⁢(nn)=∑k=1n(nk)=2n-1. The semi-supervised analysis we applied here, enables semi-automatic analyses of such large consortia.

The Enterobacteriaceae family of bacteria are primary colonizers of mammalian intestines consisting of both symbionts and pathobionts that can cause intestinal diseases (Kelly et al., 2017). Antibiotic resistant Enterobacteriaceae are an arising global health hazard due to their increasing prevalence and limited effective eradication treatments. In many cases, gut colonizing strains are the source for Enterobacteriaceae diseases (Dickstein et al., 2016; Bar-Yoseph et al., 2021). Thus, developing methods that enable studying and understanding Enterobacteriaceae mechanisms of colonization, dynamic interactions with other gut microbiota members and their effects on the host in health and disease is timely and clinical relevant. Therefore, we aimed to validate COMBICK on a panel of Enterobacteriaceae species, both lab strains and clinical isolates, including antibiotic resistant isolates (Supplementary Table 3). Our panel included bacteria from Citrobacter, Klebsiella and Escherichia genera. We succeeded to label different strains of Escherichia coli, Citrobacter rodentium, and more (Figures 4, 5 and Supplementary Figures 3, 4). Among the Enterobacteriaceae genera that we tested, Klebsiella was the one that was inefficiently labeled (Supplementary Figure 5).

Lastly, we applied COMBICK on anaerobic bacteria directly isolated from stool. The bacteria were isolated from the stool of “humanized” mice (i.e., mice colonized with a human microbiota), and dual-labeled by COMBICK using AF488 and AF594. Out of 11 isolates 10 were successfully labeled and only one was not (Figure 6 and Supplementary Figure 6).

To summarize, this method is relatively simple, fast, and applicable to bacteria from across phyla and ecosystems, both lab strains and clinical isolates. The synergism between multi-combinatorial labeling and semi-supervised analysis makes COMBICK applicable to study and differentiate mixed populations of live bacteria, including multi-strains consortia. Furthermore, COMBICK allows future studies on microbial colonization, population dynamics, microbe-microbe, microbe-host, and bacteria-phage interactions, as well as analyses of individual bacterial populations in the context of communities in their natural environments. The method presented here opens a window for real-time studies on high-order, diverse gut bacterial communities – commensals, symbionts and pathobionts (i.e., Enterobacteriaceae). As such, we believe this method is timely and can significantly facilitate mechanistic and translational research of the gut microbiota.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethical review and approval was not required for the animal study because in the experiments involving mice we only collected stool from the mice and did not conduct experiments on them.

HH, NB, and TG conceived the project, analyzed the data, interpreted the results, and wrote the manuscript. NG-Z conceived the project, interpreted the results, and wrote the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.