Linking Microbial Phylogeny to Metabolic Activity at the Single-Cell Level by Using Enhanced Element Labeling-Catalyzed Reporter Deposition Fluorescence In Situ Hybridization (EL-FISH) and NanoSIMS

Abstract

To examine phylogenetic identity and metabolic activity of individual cells in complex microbial communities, we developed a method which combines rRNA-based in situ hybridization with stable isotope imaging based on nanometer-scale secondary-ion mass spectrometry (NanoSIMS). Fluorine or bromine atoms were introduced into cells via 16S rRNA-targeted probes, which enabled phylogenetic identification of individual cells by NanoSIMS imaging. To overcome the natural fluorine and bromine backgrounds, we modified the current catalyzed reporter deposition fluorescence in situ hybridization (FISH) technique by using halogen-containing fluorescently labeled tyramides as substrates for the enzymatic tyramide deposition. Thereby, we obtained an enhanced element labeling of microbial cells by FISH (EL-FISH). The relative cellular abundance of fluorine or bromine after EL-FISH exceeded natural background concentrations by up to 180-fold and allowed us to distinguish target from non-target cells in NanoSIMS fluorine or bromine images. The method was optimized on single cells of axenic Escherichia coli and Vibrio cholerae cultures. EL-FISH/NanoSIMS was then applied to study interrelationships in a dual-species consortium consisting of a filamentous cyanobacterium and a heterotrophic alphaproteobacterium. We also evaluated the method on complex microbial aggregates obtained from human oral biofilms. In both samples, we found evidence for metabolic interactions by visualizing the fate of substrates labeled with ¹³C-carbon and ¹⁵N-nitrogen, while individual cells were identified simultaneously by halogen labeling via EL-FISH. Our novel approach will facilitate further studies of the ecophysiology of known and uncultured microorganisms in complex environments and communities.

Linking phylogenetic information to function of microorganisms in complex environmental communities is a key challenge in microbial ecology. Isotope-labeling experiments provide a useful means to investigate the ecophysiology of microbial populations and cells in the environment (29, 47). The incorporation of stable or radioisotopes into populations of microbial communities has been shown to reveal information about their metabolic activities (40). Subsequent analysis of biomarkers from cells which assimilated isotope-labeled compounds made it possible to link metabolic activities to the phylogenetic identity of the respective microorganisms. Established biomarkers in stable isotope probing experiments include phospholipid fatty acids (6, 22), DNA (41), and RNA (23).

The combination of fluorescence in situ hybridization (FISH) and microautoradiography (MAR) enables the direct microscopic observation of radioisotope incorporation into individual cells within complex microbial communities (18, 30, 34). However, the application of MAR-FISH is limited to radio-isotopes with a suitable half-life and may be unsuitable for many environmental settings. Also, the biologically relevant elements nitrogen and oxygen cannot be tracked by MAR-FISH. Recently, MAR-FISH and stable isotope probing (28, 31, 49) protocols have been refined, and new, related techniques such as isotope arrays (2), Raman microscopy (13), and high-resolution nanometer-scale secondary-ion mass spectrometry (NanoSIMS) (11) have been developed, thereby expanding the applications of isotope labeling experiments. These techniques have recently been reviewed by Neufeld et al. with respect to spatial resolution, sensitivity, and application to complex microbial communities and yet-to-be cultivated microorganisms (29).

By using secondary-ion mass spectrometry (SIMS), the isotopic/elemental composition of microorganisms can be determined for natural abundance levels (32) or for levels after isotopic enrichment (9). In studies of anaerobic methane-oxidizing archaea, SIMS was coupled with FISH to determine isotopic composition and microbial identity (32, 33, 46). This approach requires the separate determination of microbial identity by epifluorescence microscopy. However, the low spatial resolution of the isotopic analysis (∼5 μm) limits its application. Recent improvements in SIMS design have increased sensitivity and spatial resolution (∼50 nm) (14) so that it is now possible to quantify the isotopic composition of single microbial cells in environmental samples (7, 17). Li et al. combined NanoSIMS and FISH using iodine- and fluorine-labeled oligonucleotide probes and stable isotopes of carbon and nitrogen (19). The approach allowed the simultaneous analysis of microbial identity and stable isotope-labeled cells through the parallel detection of probe-conferred iodine labeling and isotopic cell composition in NanoSIMS measurements. However, the environmental applicability of this approach is limited because elemental labeling by standard FISH techniques is directly dependent on cellular ribosome content. This restricts the application to environments of high microbial activity. Li et al. (19) also reported that signal-to-background ratios for fluorine after hybridization with a ¹⁹F-labeled oligonucleotide probe were low and limited the application of this element for cell labeling.

Here, we report a new combination of FISH and NanoSIMS that expands the applicability of the approach for environmental studies. We used in situ hybridization with horseradish peroxidase (HRP)-conjugated oligonucleotide probes and tyramide signal amplification for elemental labeling of probe-targeted cells to achieve microbial identification by NanoSIMS. Labeled tyramide substrates were custom-synthesized using fluorescent dyes that contained halogen atoms. The use of tyramides labeled with both a fluorophore and with halogen atoms enabled simultaneous fluorescence and element labeling. We named this new method EL-FISH, for element labeling-FISH. Using EL-FISH, we increased the sensitivity of cell detection and extended the range of elements for cell labeling to bromine and fluorine. We demonstrate the methodological improvement of EL-FISH/NanoSIMS with pure cultures of Escherichia coli and Vibrio cholerae. We used the new protocol to study the metabolic interactions of a dual-species microbial consortium consisting of a heterotrophic alphaproteobacterium and a filamentous cyanobacterium. To show that EL-FISH/NanoSIMS can be applied to study single cells in complex microbial communities, we also applied the new method to microbial aggregates obtained from the human gingival sulcus.

MATERIALS AND METHODS

Culture preparation.

E. coli strain K12 was grown until mid-log phase at 37°C in Luria-Bertani (LB) broth. Cells were washed twice in 1× phosphate-buffered saline (PBS) and fixed with 1% paraformaldehyde at room temperature for 30 min. After fixation, cells were washed twice in 1× PBS and stored in 50% ethanol-1× PBS (vol/vol) at −20°C until use. V. cholerae strain 92A1552 (50) was grown at 30°C in LB broth to an optical density at 600 nm of 3.0. For isotopic labeling, cells were harvested by centrifugation at 6,000 × g for 2 min, washed twice in 1× PBS, and then resuspended in M9 mineral medium (26), which was amended with 500 μM ¹³C-labeled complex algal amino acid mixture (98 atom% ¹³C; Sigma-Aldrich, MO). Cells were grown at 30°C and fixed as described above after a 6-h incubation. Pseudomonas aeruginosa strain PAO1 and Bacillus subtilis strain JH642 were grown at 30°C in LB broth to an optical density at 600 nm of 1.0. Cells were fixed as described above.

A coculture of Anabaena sp. strain SSM-00 and Rhizobium sp. strain WH2K was grown without agitation at 22°C in 0.5× SO medium (45) under 21 to 30 μmol of photons/m²/s of cool, white fluorescent light. Prior to isotopic labeling, cells were harvested by centrifugation at 6,000 × g for 2 min, washed twice in 0.5× SO medium without sodium bicarbonate, and then resuspendend in the same medium. Fifty microliters of 500 mM NaH¹³CO₃ (99 atom% ¹³C) was added to a clear sample vial before the addition of 1 ml of the cell suspension. Afterwards, the vial was sealed with a Teflon/silicone septum. One milliliter of ¹⁵N₂ (98 atom% ¹⁵N) was added to each vial with a syringe. The vials were incubated upside down under a cool, white fluorescent light for 24 h at room temperature. Following incubation, cells were fixed as described above.

Microbial aggregates from oral biofilms were sampled from the gingival sulcus of a healthy individual (male, 32 years of age) after 2 days without tooth brushing. These oral specimens were dispersed and suspended in dental transport medium (Anaerobe Systems, CA). ¹³C-labeled complex algal amino acid mixture (98 atom% ¹³C; Sigma-Aldrich, MO) was added to a final concentration of 500 μM. Samples were incubated for 3 h at 37°C under anoxic conditions and fixed as described above.

FISH with HRP-conjugated oligonucleotides and tyramide signal amplification.

Catalyzed reporter deposition (CARD)-FISH was performed as previously described by Pernthaler et al. (38). Samples were spotted on silicon wafers (Ted Pella, CA), air dried, and processed according to the protocol of Pernthaler et al. (38). Hybridized samples were imaged with an epifluorescence microscope (Zeiss Axiophot; Germany) to map wafers and localize hybridized cells for subsequent NanoSIMS measurements. Element labeling was achieved by using custom-labeled tyramides. Tyramides were labeled as described in Pernthaler et al. (39). The following fluorescent dyes containing bromine and fluorine atoms were used: 5-carboxy-2′,4′,5′,7′-tetrabromosulfonefluorescein, succinimidyl ester (544Br; emission wavelength/element label); Oregon Green 488-X, succinimidyl ester (517F); and 6-((4,4-difluoro-1,3-dimethyl-5-(4-methoxphenyl)-4-bora-3a,4a-diaza-s-indacene-2-propionyl)amino) hexanoic acid, succinimidyl ester (BODIPY TMR-X; 570F). All dyes were obtained from Molecular Probes, Inc. (Eugene, OR). HRP-conjugated oligonucleotide probes used in this study were as follows: EUB338-I (5′-GCTGCCTCCCGTAGGAGT-3′) (3), a domain-specific probe targeting most Bacteria; NON338 (5′-ACTCCTACGGGAGGCAGC-3′) (48), a background control probe; ALF968 (5′-GGTAAGGTTCTGCGCGTT-3′) (27), a probe specific to Alphaproteobacteria, except Rickettsiales; and CF319a (5′-TGGTCCGTGTCTCAGTAC-3′) (24), a probe targeting the Cytophaga-Flavobacterium cluster of the Bacteroidetes, including some Sphingobacteria (20).

Hybridization with a covalently halogen-Cy3-labeled probe was performed as described previously (43). The EUB338-I oligonucleotide containing three 5-fluoro-2′deoxyuridine (5Fl-dU) nucleotides was obtained from Operon Biotechnologies, Inc. (Table 1; footnote b gives the label positions).

TABLE 1.

Relative fluorine and bromine abundance of E. coli and V. cholerae cells after FISH

Method, probe, and/or dye (n)a	No. of single cells analyzed	Relative halogen abundance of a set of single cellsf				Signal/background ratio
		19F/12C	19F/14N	81Br/12C	81Br/14N	Halogen/ C	Halogen/ N
Standard FISH with halogen-containing probe EUB338-I with Cy3 and 5Fl-dUb	16	4.5 × 10−3 ± 1.3 × 10−4	5.0 × 10−3 ± 1.2 × 10−4			2	6
EL-FISH with different tyramides on E. colic
517F	7	4.1 × 10−1 ± 1.4 × 10−2	2.3 × 10−1 ± 4.2 × 10−3			181	265
570F	6	1.8 × 10−2 ± 4.6 × 10−4	1.1 × 10−2 ± 2.1 × 10−4			8	13
544Br	11			1.8 × 10−2 ± 1.4 × 10−4	6.5 × 10−3 ± 8.7 × 10−5	22	21
Background controld		2.3 × 10−3 ± 7.0 × 10−5	8.8 × 10−4 ± 4.2 × 10−5	8.1 × 10−4 ± 2.1 × 10−5	3.1 × 10−4 ± 1.3 × 10−5	1	1
EL-FISH on Vibrio sp.e
EUB338-I	22	8.4 × 10−1 ± 1.8 × 10−2	2.2 × 10−1 ± 6.0 × 10−3			42	37
NON338	9	2.0 × 10−2 ± 9.0 × 10−4	6.0 × 10−3 ± 2.0 × 10−3			1	1

^aStandard FISH, fluorine-labeling of E. coli cells with directly fluorine-labeled oligonucleotide probes; EL-FISH on E. coli, comparison of different halogen-containing tyramides after EL-FISH on E. coli cells with probe EUB338-I; EL-FISH on Vibrio sp., comparison of positive (probe EUB338-I) and negative (probe NON338) control hybridizations on V. cholerae cells.

^b5Fl-dU nucleotides were substituted for deoxythymidine nucleotides. The EUB338-I probe sequence was [Cy3]GC[5Fl-dU]GCC[5Fl-dU]CCCG[5Fl-dU]AGGAGT. The probe was Cy3 labeled for hybridization control by epifluorescence microscopy.

^cAll hybridizations performed with HRP-conjugated EUB338-I probe. See Materials and Methods section for tyramide abbreviations. Tyramide signal amplification was for 10 min at 46°C with a 1:500 dilution of tyramide stock.

^dNegative control. Analysis was performed on fixed cells without hybridization and tyramide signal amplification.

^eHybridizations performed with HRP-conjugated EUB338-I or NON338 probe. Tyramide signal amplification using tyramide Oregon Green 488-X (517 F). Mass images provided in supplemental data.

^fNumbers given in the table are average values with standard errors for the number of single cells analyzed.

NanoSIMS.

Analyses were performed using the Lawrence Livermore National Laboratory NanoSIMS 50 instrument (Gennevilliers, France). Samples were prepared by spotting cells onto a 7- by 7-mm silica wafer. Images were generated with a 2.6 pA Cs⁺ primary beam, focused to a nominal spot size of ∼150 nm, and stepped over the sample in a 256- by 256-pixel raster to generate secondary ions. Dwell time was 1 ms/pixel, and raster size ranged between 5 by 5 and 30 by 30 μm. Samples were presputtered at high beam currents (∼1 nA) to a depth of ∼100 nm before measurements to achieve sputtering equilibrium. Secondary ions were detected in the simultaneous collection mode by pulse counting to generate 20 to 60 serial quantitative secondary-ion images (i.e., layers); the number of layers was adjusted according to the scanning frame to ensure comparable coverage between analyses with different raster sizes.

For analyses where cells were targeted with fluorine (F)-labeled tyramides, electron multiplier detectors were positioned to collect ¹²C⁻, ¹³C⁻, ¹⁹F⁻, ¹²C¹⁴N⁻, and ¹²C¹⁵N⁻ ions. For bromine (Br)-targeted cells, electron multiplier settings were ¹²C¹²C⁻, ¹²C¹³C⁻, ¹²C¹⁴N⁻, and ⁸¹Br⁻. Nitrogen in the sample was detected as the CN⁻ cluster ion. Samples were also imaged simultaneously by secondary electrons. The secondary mass spectrometer was tuned for ∼6,800 mass resolving power to resolve isobaric interferences. The depth of analysis during a measurement ranged from 200 to 900 nm. For pure cultures, 5 to 20 cells were analyzed individually, and measurements were repeated on selected cells to ensure measurement accuracy. For the samples with mixed cell types, analyses were conducted in 5 to 10 different locales.

Data were processed as quantitative isotopic ratio images using custom software and were corrected for effects of quasi-simultaneous arrival, detector dead time, and image shift from layer to layer. Each cell was defined as a region of interest by encircling pixels with ¹²C¹⁴N⁻ counts of >30% of the maximum counts in the image. The isotopic composition and/or relative abundance of halogen ions present in each region of interest was calculated by averaging over all replicate layers where both C and N isotopes were at sputtering equilibrium. Halogen ion (¹⁹F⁻ and ⁸¹Br⁻) counts are normalized to ¹²C.

A B. subtilis spore preparation was used as a reference standard for the C and N isotopic measurements (¹³C/¹²C = 0.0110; ¹⁵N/¹⁴N = 0.00370). Isotopic enrichment of standards was independently determined at the University of Utah. Measurement precision, σ_internal, was 0.4 to 1.4% (2σ) for individual ¹³C/¹²C and ¹⁵N/¹⁵N measurements, and replicate analyses of the standard yielded an analytical precision, σ_std, of 2.1% (2σ) for an individual measurement. Isotope enrichment data are presented as atom percent excess (APE); calculations of APE and associated precision for replicate analyses follow the example in Popa et al. (40). APE is calculated based on the initial isotopic ratios of the organism at time zero (R_i) and the isotopic ratio of the sampled organism (R_f): APE = [R_f/(R_f + 1) − R_i/(R_i + 1)] × 100%.

RESULTS

CARD-FISH is an rRNA-based technique for fluorescently staining microorganisms that has broad applicability in microbial ecology. We demonstrate that CARD-FISH can be used to label microbial cells with halogen atoms, such as bromine and fluorine, in order to identify single cells phylogenetically in a sensitive manner by NanoSIMS analysis.

Element labeling of microorganisms with CARD-FISH.

In order to identify microorganisms by NanoSIMS, we increased the intracellular abundance of fluorine and bromine, which have a relatively low natural abundance in biological samples (relative fluorine and bromine abundances in tissue are 0.05 and 7 ppm, respectively [10]). We chose the halogens fluorine and bromine, but the application of other elements/isotopes is also feasible (e.g., the relative iodine abundance in tissue is 0.05 to 0.7 ppm) (10, 19). We used rRNA-targeted FISH with HRP-conjugated oligonucleotide probes to introduce the element label selectively into probe-targeted cell populations. Element labeling was achieved by the use of tyramides that contained halogen atoms. Halogen-labeled tyramides were custom synthesized using fluorescent dyes, such as Oregon Green (Molecular Probes, Inc.), which contains two fluorine atoms. The fluorescent tyramides were deposited inside cells by the catalytic activity of the probe-bound HRP. In this way cells that hybridized with the selected probe were labeled with both stable halogen isotopes and fluorescent dye molecules, allowing identification of these cells in mixed communities by both NanoSIMS and epifluorescence microscopy.

We tested the CARD-FISH/NanoSIMS protocol by performing hybridizations with E. coli and V. cholerae cells. To demonstrate the applicability of different elemental labels, we hybridized E. coli cells with a general probe for the domain Bacteria (EUB338-I) and performed signal amplification with different tyramides. Three fluorescent dyes with different optical properties and halogen content were chosen for tyramide labeling. Figure 1 and Table 1 summarize the results of the labeling of E. coli with fluorine- and bromine-containing tyramides. All tyramides were applied under standard conditions as described by Pernthaler et al. (39). Cellular abundance of the respective halogen was expressed relative to ¹²C-carbon. We found that hybridized cells in comparison with nonhybridized background control cells had higher relative ratios of halogen to carbon (Fig. 1). The signal amplification with tyramide 517F was most efficient. The probe-conferred fluorine signal was 180-fold above the natural background of fluorine to carbon (Table 1). Use of tyramide 570F resulted in an eightfold increase over the fluorine-to-carbon background ratio. A 20-fold increase in relative bromine abundance was achieved in E. coli cells labeled with tyramide 544Br (Table 1).

FIG. 1.

EL-FISH/NanoSIMS images of E. coli cells hybridized with probe EUB338-I. Cells were labeled with custom-synthesized fluorescently labeled tyramides containing either fluorine or bromine atoms. Secondary-electron image and corresponding image showing the relative abundance of ¹⁹F-fluorine derived from tyramide 517F (A) and tyramide 570F (B). (C) Secondary-electron image and corresponding image showing the relative abundance of ⁸¹Br derived from tyramide 544Br. (D) Secondary-electron image and ¹⁹F-fluorine distribution in negative control with no hybridization or tyramide labeling. The color scale bars indicate the relative halogen-to-carbon abundance and were adjusted to achieve optimal cell visualization while ensuring image comparability.

In contrast, EL-FISH with rRNA-targeted oligonucleotide probes carrying a single fluorescent dye molecule at the 5′ end and a number of halogen-containing nucleotides, such as 5Fl-dU, increased the cellular fluorine-to-carbon ratios only twofold. Hence, the signal-to-noise ratio of the EL-FISH approach over the standard FISH technique is increased 10- to 100-fold, depending on the fluorescent tyramide and element label used (Table 1).

To confirm that the increased relative elemental halogen abundance was due solely to probe hybridization, we hybridized a background control probe (NON338) to ¹³C-enriched V. cholerae cells. After hybridization with the background control probe and incubation with the tyramide 517F, V. cholerae cells did not show elevated fluorine abundance (Table 1; see also Fig. S1 in the supplemental material). Hybridization with a Bacteria-specific probe (EUB338), on the other hand, resulted in a 42-fold increase of intracellular fluorine abundance (relative to ¹²C-carbon) compared with NON338 background control levels (see Fig. S1 in the supplemental material). We also tested whether our technique allowed for specific discrimination of microbes with mismatches at the probe target site. We hybridized cultures of the gram-positive B. subtilis (1.3 weighted mismatches) and the gammaproteobacterium P. aeruginosa (1.5 weighted mismatches) with the oligonucleotide probe CF319a (21). Nonspecific FISH staining or fluorine enrichment of these strains by probe CF319a was not observed (see Fig. S2 in the supplemental material).

Dual-species microbial consortium (Anabaena and epibiont).

Having shown that EL-FISH can be used to increase relative cellular halogen abundance in probe-targeted cells for cell identification by NanoSIMS, we combined the phylogenetic isotope labeling with visualization of metabolic activity. We chose a dual-species microbial consortium consisting of a filamentous cyanobacterium and a heterotrophic alphaproteobacterium (45) to examine whether the combination of EL-FISH/NanoSIMS can be used both to identify microbial partners phylogenetically and to monitor the intra- and intercellular fates of carbon and nitrogen. The distinct morphology of the two microbial species in this sample served as a visual control of label distribution.

Associations of heterotrophic bacteria and photosynthetic cyanobacteria occur in a variety of aquatic habitats (36). These interactions have been described as mutualistic, with fitness benefits for both associated species (35). The heterotrophic bacterium (epibiont) attaches to the cyanobacterial heterocyst, which the supplies the epibiont with organic carbon and nitrogen compounds (35). The respiratory activity of the epibiont is believed to decrease local oxygen concentrations and produce CO₂, thereby stimulating photosynthetic growth of the cyanobacterium (42).

We used pure cultures and cocultures of Anabaena sp. strain SSM-00 and Rhizobium sp. strain WH2K. To investigate whether the physical interaction between the Anabaena and the heterotrophic epibiont involves the transfer of metabolite(s), we incubated the pure cultures and the coculture with ¹³C-bicarbonate (25 mM; 99 atom% ¹³C) and ¹⁵N-dinitrogen (70% ¹⁵N₂ of total N₂; 98 atom% ¹⁵N). Anabaena sp. strain SSM-00 fixed carbon and nitrogen in the absence of the Rhizobium species, while the Rhizobium sp. strain WH2K did not show any ¹³C-bicarbonate and ¹⁵N-dinitrogen incorporation in pure culture (data not shown). PCRs using epibiont DNA and degenerate primers targeting the nifH gene, which encodes a nitrogenase subunit, did not result in a product (45).

When the Rhizobium species was grown in coculture, isotopically labeled carbon and nitrogen were detected, suggesting that carbon and nitrogen compounds fixed by the cyanobacterium were assimilated by the epibiont (Fig. 2D and E). In Fig. 2D newly fixed ¹⁵N-nitrogen is apparent in all three cell types, the heterocyst, vegetative cells, and epibionts. As previously shown, the export of newly fixed nitrogen from the heterocyst to vegetative cells occurs rapidly (12, 25, 40), resulting in a higher relative enrichment of newly fixed ¹⁵N-nitrogen in the vegetative cells than in the heterocyst and epibionts (Fig. 2D). Carbon dioxide fixation through oxygenic photosynthesis enriched the vegetative cells in ¹³C-carbon (Fig. 2E). Enrichment of ¹³C-carbon was also visible in epibiont cells attached to the heterocyst (Fig. 2E). ¹³C-carbon enrichment in the heterocysts was relatively low, consistent with heterocysts' being nongrowing and devoid of photosystem II and of autotrophy (Fig. 2E).

FIG. 2.

Fluorescence and NanoSIMS images of a microbial consortium consisting of filamentous cyanobacteria (Anabaena sp. strain SSM-00) and alphaproteobacteria (Rhizobium sp. strain WH2K) attached to heterocysts. Images taken after a 24-h incubation with ¹³C-bicarbonate and ¹⁵N-dinitrogen. (A) Fluorescence image of the microbial consortium after EL-FISH with probe ALF968. (B) NanoSIMS secondary-electron image corresponding to panels C to E. (C) Localization of fluorine relative to carbon after EL-FISH with ALF968. (D) Distribution of ¹⁵N-nitrogen enrichment. (E) Distribution of ¹³C-carbon enrichment. Color bars indicate relative fluorine abundance (C) and isotope enrichment (D and E) in the image. Het, heterocyst; Veg, vegetative cell; Epi, epibiont; unatt Epi, Epibiont cells not attached to heterocysts.

We used EL-FISH to label one partner of the dual species consortium with fluorine to enable visualization by NanoSIMS (Fig. 2A). The power of elemental and isotopic imaging by NanoSIMS was revealed when images corresponding to different isotope masses were compared. Figure 2C shows the selective labeling of the epibiont cell with fluorine after in situ hybridization of the consortium with an alphaproteobacteria probe (ALF968). Anabaena cells, which were not targeted by the alphaproteobacteria probe, contained only background levels of fluorine. Rhizobium cells not attached to a heterocyst did not show ¹³C or ¹⁵N enrichment (Fig. 2D and E). However, these cells were visible in secondary electron images and detected with EL-FISH because of their increased fluorine abundance (Fig. 2B and C). These findings show that EL-FISH enables cell detection and phylogenetic identification independent of the cell's metabolic state.

Multispecies microbial aggregates obtained from oral biofilms.

To demonstrate the applicability of our method for the identification of metabolically active microbial cells in complex microbial communities, we examined aggregates obtained from an oral biofilm from the human gingival sulcus. Oral biofilms encompass more than 750 bacterial species, of which approximately 40% can be cultivated while the remaining species have only been identified by sequencing of their 16S rRNA genes (1). Microorganisms belonging to the Cytophaga-Flavobacterium cluster of the Bacteroidetes are present in oral biofilms (8). The predominant genera Capnocytophaga, Prevotella, and Porphyromonas have been implicated in the pathogenesis of periodontal disease (37, 44).

Microbial aggregates from oral biofilms were incubated in a mineral medium with 500 μM ¹³C-labeled amino acids for 3 h, fixed, and hybridized with a group-specific Cytophaga-Flavobacterium probe (CF319a) (24). Following CARD-FISH, carbon and fluorine enrichment was visualized by NanoSIMS (Fig. 3). The probe-targeted cell population was identified by elevated intracellular fluorine levels. By comparing the ¹³C APE image (Fig. 3B) with the fluorine image (Fig. 3C), Cytophaga-Flavobacterium species were identified as metabolizing the ¹³C-labeled amino acid because of their enrichment in both ¹³C and ¹⁹F (mean of ¹⁹F/¹²C ratio for ¹³C-enriched cells, 0.0891 ± 0.0059; mean of background except ¹³C-enriched cells, 0.0698 ± 0.0007) (Fig. 3, arrows). Figure 3 also shows that microorganisms other than the probe-targeted Cytophaga-Flavobacterium species metabolized ¹³C-labeled amino acids. Notably, not all cells with elevated fluorine were ¹³C enriched. The mostly slender, long filamentous cells were Cytophaga-Flavobacterium species that did not metabolize ¹³C-labeled amino acids. Cells that were both enriched in ¹³C-carbon and exhibited a relative increase in fluorine abundance mostly had a rod shape or coccoid morphology.

FIG. 3.

NanoSIMS images of microbial aggregates obtained from an oral biofilm from the human gingival sulcus after incubation with ¹³C-labeled amino acids and EL-FISH. (A) Secondary-electron image. (B) Distribution of ¹³C-carbon enrichment. (C) Relative abundance of ¹⁹F-fluorine to ¹²C-carbon after hybridization with probe CF319a. Arrows in panels B and C indicate probe-identified microorganisms that have incorporated the labeled substrate. Color bars indicate the relative isotopic/elemental enrichment/abundance in the image.

DISCUSSION

We developed a FISH protocol (EL-FISH) to label probe-targeted cell populations in a specific manner with the halogens bromine and fluorine to enable rRNA-based cell identification by NanoSIMS. In combination with stable isotope tracer experiments, we obtained functional and phylogenetic information simultaneously from individual cells in microbial communities in a single NanoSIMS analysis.

Tyramides containing a fluorophore and halogen atoms proved to be very useful during sample preparation because they enabled direct correlation of fluorescence and halogen signals. First, samples were imaged by epifluorescence or confocal laser scanning microscopy for specific probe hybridization signal. Fluorescence images obtained from microbial populations and single cells helped guide subsequent NanoSIMS analyses to spots of interest on the silicon wafer. By analyzing fluorescence first, optimization of in situ hybridization was achieved simultaneously with sample preparation for NanoSIMS and obviated the need for additional experimental efforts.

EL-FISH offers advantages in the sensitivity of detection over standard FISH techniques. In many environmental samples, cell detection by rRNA-directed in situ hybridization is hampered by the low ribosome content of target microorganisms. When directly labeled oligonucleotide probes are used, fluorescence signal intensity and element-labeling efficiency depend on the cellular ribosome content. Using EL-FISH, deposition of fluorescent dye and element label is mediated by the catalytic activity of the probe-conjugated HRP. The enzymatic signal amplification reaction allows the adjustment of probe-conferred signal intensities by variations in tyramide concentration and incubation times. Thus, cells from oligotrophic environments, which have low ribosome contents, can effectively be enriched with halogens to allow identification by NanoSIMS. We showed that elements of relatively low abundance in biological tissue, such as bromine and fluorine (10), can be effectively used for EL-FISH cell labeling. The EL-FISH halogen labeling takes advantage of the very high sensitivity of NanoSIMS for fluorine and bromine.

The rRNA-based element labeling of individual microbial cells for cell identification by NanoSIMS does not require the modification of existing CARD-FISH protocols. The only alteration is the use of halogen-containing tyramides. Many halogen-containing fluorescent dyes are commercially available for custom tyramide conjugate synthesis and can be used for NanoSIMS analysis. In contrast, oligonucleotide probes, which are covalently modified through the addition of halogen atoms, might exhibit an altered melting behavior, depending on the number and physiochemical properties of the added element. While hybridizations with these probes might be less challenging, their altered elemental composition will require additional controls and optimization of the hybridization conditions and is associated with lower detection sensitivity.

We conducted in situ hybridizations with pure cultures of E. coli to characterize fluorine and bromine labeling of cells by EL-FISH (Table 1). The relative halogen abundance obtained after EL-FISH was high compared to the natural background of the respective halogen in nonhybridized cells (Fig. 1). We also compared the elemental labeling efficiency of EL-FISH with standard FISH and showed that standard FISH resulted in lower cellular halogen abundance than EL-FISH when oligonucleotide probes with three 5Fl-dU nucleotides were used (Table 1).

Successful binding of the HRP-oligonucleotide conjugate to its probe sequence-defined ribosomal target site is a prerequisite for enzyme-mediated tyramide deposition. Control hybridization experiments with ¹³C-enriched V. cholerae cells showed that the use of a background control probe (NON338) resulted in about 40-fold lower intracellular fluorine abundance and no detected halogen background (Table 1; see also Fig. S1 in the supplemental material).

We also demonstrated that combined EL-FISH/NanoSIMS provides information on metabolic activity of single cells and offers insights into the distribution of microbial activities in and among individual cells of probe-identified populations. Our method will facilitate studies of the phenotypic response of individual cells to environmental perturbations and enable the quantitative description of cellular behavior.

EL-FISH has the potential to be extended to mRNA and protein detection. This would enable the simultaneous visualization and quantification of multiple cellular characteristics when information on transcriptional and translational activity is integrated with carbon/nitrogen distribution and cellular identification. Key enzymes of metabolic pathways can reveal a cell's principle mode of energy conservation (e.g., sulfate reduction and methanogenesis) or allow identification of the assimilatory pathway that is used for the synthesis of new biomass.

Given the high spatial resolution of NanoSIMS (∼50 nm), it should be possible to study intracellular distribution and localization of cellular activity (cytoplasmic, membrane bound, and periplasmic) by isotope tracer experiments or labeling of cellular features through EL-FISH or immunostaining. The detection of multiple cell features or different 16S rRNA phylotypes at the same time may require additional elemental labels. We have successfully established the use of bromine and fluorine. Li et al. report the application of iodine (19). Selenium, boron, gold, and silver might be other suitable elements for cellular labeling because of their relatively low natural abundance in biomass (10). Subsequent application of different probes and substrates has been demonstrated for CARD-FISH, and a combination of the technique with standard FISH is feasible (4, 5, 46). Further experiments will evaluate the use of multiple elements for simultaneous identification of different microbial populations or labeling of different intracellular features, such as mRNA and proteins, based on EL-FISH/NanoSIMS.

EL-FISH/NanoSIMS combines features of high-resolution microbial imaging techniques, stable isotope probing, and microbial identification methods that rely on molecular biomarkers. This combination of techniques will facilitate the study of hitherto uncultivated microorganisms in diverse environmental communities and will reveal insights into the interrelationships of individual microbial community members (15, 16).