FDAAs can be covalently incorporated into bacterial peptidoglycan (PGN) by the catalysis of endogenous D,D or L,D-transpeptidases of the bacteria, which provides a versatile strategy for examining PGN synthesis during the bacterial growth and division (Kuru et al., 2019). Recently, it has been demonstrated that multiple FDAAs could be chronologically incorporated into bacterial PGN in vivo (Hudak et al., 2017; Lin L. et al., 2020). Here, to reveal the growth and division patterns of gut bacteria in rats, we used the STAMP protocol to label their gut microbiotas with two different FDAAs: TAMRA-amino-D-alanine (TADA) and Cy5-amino-D-alanine (Cy5ADA), containing TAMRA (tetramethylrhodamine) or Cy5 (Cyanine 5) on their side chains, respectively (Figure 1A).

In previously reported mouse microbiota labeling, the FDAAs were given at an interval of 3 h (Lin L. et al., 2020). Considering that the gastric emptying and intestinal motility time of rats are comparable to mice (Lu et al., 2017), here we adopted a similar in vivo labeling protocol. The collected cecal microbes after STAMP showed strong dual-color labeling, and the green and red signals observed by confocal microscopy revealed the PGN synthesis processes of the labeled gut bacteria (Figure 1B). Flow cytometry also showed high labeling coverages of the two FDAAs (Supplementary Figure S1), indicative of an efficient in vivo labeling of the rat microbiotas.

The two-color fluorescence imaging presented a high morphological diversity of rat gut microbes with strong FDAA labeling throughout the cells (Figure 1B). The different distributions of labeling signals indicated distinct growth and division patterns: septum synthesis division (Figure 1C, No.1), segmented/continuous diffuse synthesis (No. 2 and 3), and asymmetric/polar synthesis (No. 4), to name a few. We also observed bacteria with only TAMRA or Cy5 signals (No. 5 and 6), which suggested that these cells might have different growth rates during the two labeling stages.

After obtained the resourceful imaging of the FDAA-labeled rat gut microbiota, we set out to identify the growth patterns of bacteria on the species level. FISH has proved to be a powerful method for visualization and identification of microbes in complex environments (Sunde et al., 2003). To determine the taxonomic composition of the microbiota, we performed both 16S rDNA and metagenomic sequencing (Supplementary Figure S2 and Supplementary Table S1). Based on the results of metagenomic sequencing, 25 species with relatively high abundances (>0.12%, covering ∼35% of the total population, Supplementary Table S1) were selected and taxonomically labeled with corresponding FISH probes (Supplementary Tables S2,S3). The FISH sequences were either based on previous reports or designed in this study using an algorithm that was previously developed and recently optimized (Kong et al., 2010; Lin L. et al., 2020). To verify the specificities of the newly designed FISH probes, the labeling patterns and cell morphologies of each stained species were carefully examined. Only the species with highly consistent patterns and morphologies (Supplementary Figure S3) were reported herein. Among the 25 FISH probes, we were able to confirm the specificities of 15 sequences based on the criteria (listed in Supplementary Table S2).

These bacteria were from 14 genera of ten families, containing eight Gram-positive (Figure 2A–H) and seven Gram-negative species (Figure 3A–G), among which two species, Lachnospiraceae bacterium 28–4 and Firmicutes bacterium ASF500, had not been separately cultivated in vitro. Most of the labeled Gram-positive bacteria divided in binary fission. Lachnospiraceae bacterium 28–4 (Figure 2A), Clostridium clostridioforme (Figure 2B) and an unclassified species of the Fusicatenibacter genus (Figure 2C), all of which belonged to the Lachnospiraceae family, divided by synthesis at the septum. As the most abundant species in this microbiota, Lachnospiraceae bacterium 28–4 was regarded as anti-inflammatory bacteria (Daniel et al., 2017). They were shown to increase the levels of polyamines, a group of compounds that were often associated with colon cancer and regulate inflammation by inhibiting butyrate production (Daniel et al., 2017). Using fluorescence microscopy, we found that this species had a slender spindle shape and elongated from many compact and discrete sections, presenting a unique two-color labeling feature. C. clostridioforme (Figure 2B), Fusicatenibacter (Figure 2C) and Firmicutes bacterium ASF500 (Figure 2D) presented dispersed growth during elongation and zonal synthesis for septation (Brown et al., 2011; Cava et al., 2013). Clostridium difficile (Figure 2E) and Intestinimonas butyriciproducens (Figure 2F) exhibited “medial” (also known as “pre-septal”) growth (Randich and Brun, 2015), where PGN was synthesized near the division plane before full assembly for septation. In Eubacterium plexicaudatum (Figure 2G) and Acetivibrio ethanolgignens (Figure 2H), FDAA-labeling signals indicated their dispersed mode of cell elongation (Cava et al., 2013). It is noteworthy that A. ethanolgignens (Figure 2H) had two highly overlapped FDAA-signals without an apparent septum, suggesting that they might have relatively stable cell wall synthesis rates when exposed to the two FDAA probes without going through any cell dividing.

Compared to Gram-positive bacteria, Gram-negative bacteria typically have a much thinner layer of PGN and exhibit relatively weak FDAA-labeling signals. Bacteroides dorei was labeled by FDAA in a special striped manner and divided in binary fission (Figure 3A). Prevotella copri was presented as short rods (∼1 μm in length) with a clearly growing septum in the midcell (Figure 3B). Several Gram-negative bacteria (Figures 3C–G) that taxonomically belonged to the Clostridiales order of the Firmicutes phylum showed comparable FDAA labeling intensities with Gram-positive bacteria. The thickness of the Gram-negative PGN in the trilaminar outer membrane was ∼8.5 nm (Cheng and Costerton, 1977), and Gram-positive’s PGN was ∼30–50 nm (Higgins and Shockman, 1970; McCready et al., 1976). Butyrivibrio fibrisolvens, which were normally stained as Gram-negative (Cheng and Costerton, 1977), presented a typical Gram-positive PGN labeling (Figure 3C). Their Gram-negative staining was probably because of the relatively thin cell walls (12∼18 nm), which could not retain the Gram-staining complex (Cheng and Costerton, 1977), but still thick enough to have a strong FDAA labeling. Pseudoflavonifractor capillosus (Figure 3D) was originally described as Gram-negative Bacteroides capillosus in the early twenties, and re-categorized as Pseudoflavonifractor that belonged to phylum Firmicutes in 2010, for its high genomic identity (>97%) with the Flavonifractor genus (Ricaboni et al., 2017). Belonging to the class Clostridia, which is usually identified as Gram-positive, Hungatella Hathewayi (Figure 3E) is, however, described as Gram-negative end-pointed bacilli (Steer et al., 2001). These three species (Figures 3C–E) elongated through dispersed lateral growth and divided with a prominent red-labeled septum in the midcell. Interestingly, some Gram-negative species belonging to the Oscillibacter genus (Figures 3F,G) of the phylum Firmicutes showed distinct labeling patterns with relatively strong intensities. Oscillibacter ruminantium, which could cause bacteremia (Sydenham et al., 2014), had asymmetric PGN synthesis (Figure 3F). Oscillibacter valericigenes (Broutin et al., 2020) exhibited a typical pre-septal elongation mode (Figure 3G).

Bacteria in various intestinal segments and niches were growing at distinct cellular stages during the STAMP labeling, and encountered the probes asynchronously. Therefore, it became possible to infer the growth stages of individual bacteria based on their two FDAA-labeling signals. Here, shown as an example, cells of B. fibrisolvens in different cell cycle stages were presented. As an important butyrate-producing rumen bacterium, B. fibrisolvens has been evaluated as a probiotic to prevent colorectal cancer (Ohkawara et al., 2005).

In the shown images of the two-color B. fibrisolvens (Figure 4), it can be observed that its life cycle involves primarily two classical growth modes: zonal cell wall synthesis and septum formation. Zonal cell wall synthesis means the incorporation of PGN along the sidewall, and septum formation denotes that the cytoskeletal division proteins direct the localization of PGN synthesis to generate nascent poles at the cell center (Brown et al., 2011; Randich and Brun, 2015). In a “first-generation” bacterium (∼3 µm in length), PGN was uniformly constructed along the lengthening cell (Figure 4A). When the cell elongated to ∼5 μm, an annular red-labeled incorporation zone appeared in the midcell, suggesting the beginning of septal formation (Figure 4B). With the continuous synthesis of PGN, B. fibrisolvens cell kept elongating until it reached twice the average length of a cell (Figures 4C,D). During division, the zonal growth site is modified to promote inwards growth of a septum and hydrolysis to form the new poles of the offspring cells (Figure 4E). Meanwhile, new growth regions were launched at the centers of the two daughter cells (Figures 4D,E). Then the daughter cells, which had only one red-labeled pole, continued to start the next cell cycle (Figure 4F). Through fine analysis of the STAMP labeling of individual cells, our method provides a unique angle to examine the basic microbial activities of specific bacterial species in the rat gut.

The FDAA probes were synthesized by Chinese Peptide Company (Hangzhou, China). Stock solutions were prepared at concentrations of 10 mM in distilled H₂O and stored at −20°C before use. FISH probes and paraformaldehyde fix solution were purchased from Sangon Biotech (Shanghai, China). Other chemicals, not mentioned above, were obtained from Sigma-Aldrich (St. Louis, MO, United states).

Male Sprague-Dawley (SD) rats (weighing 220–250 g) were obtained from SLAC Laboratory Animal Corp (Shanghai, China). Rats were bred in the Department of Laboratory Animal Science of Fudan University under a 12 h day-night cycle at 25°C and a relative humidity of 50% for 7 days, receiving a standard chow diet and free access to clean water. All animal experiments were carried out in accordance with guidelines approved by the Department of Laboratory Animal Science and Use Committee of the Fudan University Institutes of Biomedical Sciences.

The SD rats were sequentially administered with two different FDAA probes (1 ml, 1 mM TADA or Cy5ADA in distilled H₂O) through oral gavage with an interval of 3 h. Four hours after the second gavage, their cecal microbiotas were collected using a previously reported protocol (Lin L. et al., 2020). Briefly, rats were sacrificed after anesthetized excessively by intraperitoneal injections with 10% chloral hydrate, and the ceca were dissected separately and finely minced with a pair of 4.5-inch iris scissors in 6 ml of cold phosphate-buffered saline (PBS). The tissues and digesta were then filtered with a 40 μm cell strainer to remove most of the nonbacterial tissue debris. The filtrates were then centrifuged. The bacterial pellets (whitish-colored) were washed three times with 1.5 ml of PBS by centrifugation (15,000 g, 3 min) and then resuspended in PBS to reach an appropriate concentration for subsequent analyses by flow cytometry and confocal fluorescence microscopy.

DNA from the SD rat cecum’s bacterial samples was extracted using the Omega Bacterial DNA Kit (Omega Bio-Tek, Norcross, GA, United States) according to the manufacturer’s protocol. The V3–V4 hypervariable regions of the bacteria 16S rDNA were amplified by PCR and subsequently paired-end sequenced (2 × 300) on an Illumina MiSeq platform (Illumina, San Diego, United States) according to the standard protocols. The taxonomy of each 16S rRNA gene sequence was analyzed by RDP Classifier (http://rdp.cme.msu.edu/) against the SILVA (SSU123) 16S rDNA database with a confidence threshold of 80%.

Covaris M220 (Gene Company Limited, China) was used to construct the paired library with an average DNA fragment size of about 400 bp, and the paired library was sequenced on Illumina HiSeq4000 platform. BLASTP (Version 2.2.28+, http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used for taxonomic annotations by aligning nonredundant gene catalogs against the integrated NR (non-redundant protein sequence) database with an e-value cutoff of 1 × 10^–5.

Candidate FISH probes were designed using a k-mer–based algorithm similar to KASpOD (Parisot et al., 2012). The FDAA labeled microbiota was washed and resuspended in 50% paraformaldehyde-PBS (v/v) and incubated at room temperature for 1.5 h to fix the bacteria. After washed twice with PBS, an equal volume of EtOH was then added into the suspension and stored at −20°Cfor at least 48 h. The bacteria were spun down and resuspended in a hybridization buffer [0.9 M NaCl, 20 mM tris (pH, 7.5), 0.01% SDS, and formamide, if required]. FAM-labeled FISH probes were added to the sample with a final concentration of 5 ng/µl and incubated at 46°C for 4 h, using a ThermoMixer (Eppendorf, Hamburg, Germany). After hybridization, bacteria were washed two times (15 min) with washing buffer [0.9 M NaCl, 20 mM tris (pH 7.5), and 0.01% SDS], and then resuspended in Tris buffer [20 mM Tris and 25 mM NaCl (pH, 7.5)] before analysis with flow cytometry and fluorescence microscopy.

FDAA-labeled microbiota samples were analyzed using a CytoFLex flow cytometer (Beckman Coulter Life Sciences, Indianapolis, IN, United States). Data analyses were performed using FlowJo (V 10.0.8R1). The tagged microbiota was identified by flow cytometry of logFSC and logSSC, and then fluorescence gated. For each sample, 15,000 events were collected for analysis, with debris and dual particles excluded.

The bacterial suspension was added to an agarose pad (1.5% in PBS, ∼1 mm thick), and the slides were covered with a glass cover slip. Confocal fluorescence imaging was performed using a TCS SP8 laser confocal microscope (Leica, Solms, Germany). Samples were excited for FAM (carboxyl fluorescein) at 488 nm, TAMRA at 555 nm, and Cy5 at 639 nm. The emission is detected by the corresponding emission filters. Image deconvolution was performed with Huygens Essential Deconvolution software (Scientific Volume Imaging, Hilversum, Netherlands), using a theoretical point expansion function.