INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Fluorescence in situ hybridization (FISH) using rRNA-targeted oligonucleotide probes is frequently applied to quantify the composition of microbial communities in different environments (1, 17, 21, 23, 33, 35). In such studies cell numbers are generally obtained by manual counting in an epifluorescence microscope. Usually the relative abundance of a probe target population is determined by comparison of the obtained numbers (i) with counts of all bacterial cells detectable by FISH via simultaneous hybridization with a bacterial probe (10, 11, 30, 34) or probe set (9), or (ii) with counts of all organisms containing DNA by simultaneous application of nucleic acid staining dyes (16, 29, 35, 38).

Although quantitative FISH has provided novel insights into the structure and dynamics of microbial communities, it suffers from tediousness and limited accuracy for samples containing densely aggregated cells like activated sludge flocs or biofilms. The latter problem can in part be ameliorated by the use of confocal laser scanning microscopy (CLSM) for the detection of probe-labeled cells (36). However, even if optical CLSM sections are recorded, it is not feasible to manually count a sufficient number of cells in each hybridization experiment in a reasonable time period to obtain statistically reliable results. This limitation has two reasons. First, manual counting itself is very time-consuming, and thus generally not more than a few thousand cells per hybridization experiment were counted in previous studies. Second, manual counting requires high-magnification CLSM sections, which allow single-cell resolution within clusters. However, such images contain relatively few cells, and therefore many images need to be recorded, rendering the procedure even more time-consuming. Therefore, more precise methods are required to quantify the composition of the microflora in samples containing clustered cells.

In principle, flow cytometry is a more efficient and accurate alternative for quantification of fluorescently labeled bacterial cells (39). However, for the analysis of microbial flocs and biofilms, flow cytometry is of limited use because it necessitates efficient dispersion of clustered bacteria prior to the measurement, a requirement which frequently cannot be fulfilled (38, 39).

To overcome the limitations of manual cell-counting procedures, semiautomated digital image analysis tools were recently developed which quantify fluorescently labeled bacteria in environmental samples (6, 29). But such solutions are not able to efficiently count cells in dense clusters or biofilms because single-cell recognition within these structures cannot be automated. This problem can be circumvented by measuring the areas of specifically stained bacteria in randomly acquired optical CLSM sections. This approach only requires the software to differentiate between labeled biomass (including cell clusters) and unlabeled background but does not rely on single-cell recognition within clusters. The abundance of a particular population is then expressed as fraction of the area occupied by all bacteria (8, 32).

For this purpose, an environmental sample is hybridized simultaneously with different rRNA-targeted oligonucleotide probes: one specific probe that targets the population which is to be quantified, and one domain-specific probe set that detects most bacteria. The population-specific and the domain-specific probes are labeled with different fluorochromes. Following FISH, the fluorescence conferred by the different probes is recorded in separate CLSM images. The areas of the labeled cells in these images are measured by digital image analysis. Since this approach analyses low-magnification images and can be partly automated, it allows rapid quantification of large numbers of bacteria, thereby significantly improving the statistical accuracy of the measurement (8, 32).

Ecological studies of complex microbial communities may attempt to determine not only relative abundances of probe-defined bacterial populations, but also the respective cell concentrations in a sample. This is particularly important if different samples which differ in their prokaryotic biomass content are to be compared. Furthermore, cell concentrations per volume or weight unit of an environmental sample are needed to calculate key functional attributes of bacterial populations, like in situ growth rates and in situ substrate turnover rates per cell. Despite their importance, cell concentrations of FISH-stained bacterial populations have rarely been determined for biofilms or activated sludge flocs by manual counting because these measurements required additional time-consuming and bias-introducing homogenization and membrane filtration steps (17, 24, 26, 37). In addition, it is impossible to directly apply the above-mentioned area-based quantification methods (8, 32) to semiautomatically determine absolute cell numbers in a sample after membrane filtration because these methods cannot accurately measure the entire biovolume of all cells of a probe-labeled population in the filtered biomass on top of defined filter areas.

Recently, CLSM-based methods to semiautomatically measure the biovolume of fluorescently labeled bacteria (15, 19) were published. These methods could theoretically be applied to determine absolute cell numbers of probe-defined bacterial populations on membrane filters. However, biovolume-based quantification is only accurate if serial optical sections are recorded using small vertical step intervals and subsequently combined to image stacks. This procedure is extremely time-consuming and leads to significant bleaching of FISH-labeled bacterial cells.

In this study we thus developed a semiautomated procedure for determining cell concentrations of bacterial populations in complex samples by FISH and CLSM using the area-based quantification method (8, 32). Spiking of the samples with known amounts of Escherichia coli cells, which were used as internal standards for the subsequent FISH analysis, allowed us to infer the absolute cell numbers of probe-target bacteria from their measured areas by digital analysis of CLSM images.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Test strains, culture conditions, cell fixation, and activated sludge sampling. Type strains of Comamonas testosteroni (DSM 1622) and Gluconobacter asaii (DSM 7148) were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ) (Braunschweig, Germany). Cells of C. testosteroni and G. asaii were grown overnight under agitation at 30°C in medium DSM M1 (0.5% [wt/vol] peptone, 0.3% [wt/vol] meat extract, pH 7.0) and DSM M626 (5% [wt/vol] D-sorbitol, 1% [wt/vol] yeast extract, 1% [wt/vol] peptone, pH 6.0), respectively. Pure cultures of Nitrosomonas europaea were maintained as described by Koops et al. (18). E. coli TOP10F' cells (Invitrogen, San Diego, Calif.) were grown overnight in Luria-Bertani (LB) medium (31) at 37°C under agitation. For fixation, cells of all species were washed in phosphate-buffered saline (PBS) and then incubated for 3 h in 3% paraformaldehyde (Sigma, Deisenhofen, Germany) as described by Amann (3). Fixed cells were washed again in PBS and stored at 4°C in PBS until they were used (normally within 3 days after fixation). For long-term storage, the cells were resuspended in a 1:1 mixture of PBS and 96% (vol/vol) ethanol and kept at 20°C.

Cell concentration of pure cultures. The numbers of cells per milliliter in the fixed pure cultures of C. testosteroni, G. asaii, and N. europaea were determined with a Neubauer cell counting chamber (Paul Marienfeld GmbH, Bad Mergentheim, Germany) following the instructions of the manufacturer.

5

7

1

Spiking of pure culture mixtures and activated sludge with E. coli cells. Overnight cultures of C. testosteroni and G. asaii (100 ml) were harvested by centrifugation for 10 min at 4,550 × g and fixed with paraformaldehyde as described above. The cell densities in these concentrated stock solutions were determined using the Neubauer chamber. Cell densities of E. coli overnight cultures were determined photometrically, and E. coli cells were concentrated and fixed with paraformaldehyde as described above. Subsequently, different 1-ml cell mixtures containing 6.3 × 10⁷ C. testosteroni and 3.7 × 10⁷ G. asaii cells as well as 10⁶, 10⁷, 10⁸, or 10⁹ E. coli cells were prepared in 50% PBS-ethanol (EtOH) (vol/vol). In addition, a 1-ml cell mixture containing 6.3 × 10⁷ C. testosteroni and 3.7 × 10⁷ G. asaii cells but without E. coli cells was prepared in 50% PBS-EtOH (vol/vol).

FISH. The defined mixtures of pure cultures were spotted onto microscope slides (Paul Marienfeld GmbH, Bad Mergentheim, Germany) and allowed to dry at 46°C. Afterwards, the slides were immersed for 2 to 3 s in molten 0.5% agarose (Gibco-BRL ultrapure agarose; Life Technologies, Paisley, Scotland) at 37°C. The slides were then placed on ice until the agarose had solidified. Excess agarose on the back side of the slides was removed, and the samples were dehydrated in 50, 80, and 96% (vol/vol) ethanol for 5 min each. The agarose coating was applied to minimize cell loss during the following hybridization and washing steps. After an additional drying step at room temperature, whole-cell hybridization was performed as described by Manz et al. (22).

Microscopy and digital image analysis. After in situ hybridization, the microscope slides were embedded in Citifluor AF1 (Citifluor, Canterbury, United Kingdom). Pictures of fluorescent cells were recorded using a CLSM (LSM 510, Zeiss, Oberkochen, Germany). For detection of Cy3- and Cy5-labeled cells, two helium-neon lasers (543 nm and 633 nm, respectively) and, for Fluos-labeled cells, an argon laser (450 to 514 nm) was used. For each microscope field, fluorescence conferred by the different probes was recorded in separate images. For each hybridization experiment, 30 microscope fields at random positions and in random focal planes were recorded using a Zeiss Plan-Neofluar 40×/1.3 oil objective. This procedure (30 images at low magnification) allows us to record a high number of probe-target cells and thus to accurately determine the relative abundance of heterogeneously distributed probe-target cells in activated sludge samples (8). All pictures acquired corresponded to optical sections of 1-µm thickness obtained by adjusting the pinhole diameter of the CLSM accordingly. They were recorded as 8-bit images of 512 by 512 pixels with a resolution of 1.6 by 1.6 pixels per µm.

View larger version (55K): [in this window] [in a new window]   FIG. 1.   Principle of the cell quantification method developed in this study. See text for an explanation of steps 1 to 7. (A) Microscopic picture of activated sludge from the Munich II wastewater treatment plant. (B and C) Aliquots of the same activated sludge after spiking with 108 (B) and 109 (C) E. coli cells per ml. FISH was performed with probe GAM42a (red) and the EUB338 probe mix (green). The images containing the fluorescence conferred by the probes were superimposed, and E. coli cells appear yellow due to color blending. (D) Calibration curve generated from corrected cell area fractions of E. coli in spiked sludge aliquots. (E) The same microscopic field as in A, showing simultaneous FISH using probes NEU and Nso1225 (red) and the EUB338 probe mix (green). Cells of ammonia oxidizers appear yellow due to color blending. (F) Graph showing the use of the calibration curve (D) for converting the measured area fraction of an autochthonous population to the equivalent E. coli concentration. (G) Highly magnified optical section through a cell aggregate of ammonia-oxidizing bacteria in activated sludge stained by FISH using probes NEU and Nso1225 (red). (H) Highly magnified optical section through E. coli cells from a pure culture stained by FISH using probe GAM42a (red).

Determination of cell density. The cell concentrations of probe-defined indigenous bacterial populations in a sample were calculated using a seven-step procedure. First, aliquots of the sample (e.g., activated sludge, Fig. 1A) were spiked with E. coli (10⁶ to 10⁹ cells ml¹) as described above (Fig. 1, step 1), and the spiked aliquots were stained by FISH using probe GAM42a and the EUB338 probe mix (Fig. 1B and C). The area fraction of the E. coli cells in each spiked aliquot was determined by digital image analysis (Fig. 1, step 2). For activated sludge samples, the area fraction of the inherent -Proteobacteria was measured in sludge aliquots without addition of E. coli cells. The area fraction of these indigenous cells was subsequently subtracted from the area fraction measured with probe GAM42a in the spiked aliquots to obtain the area occupied by E. coli. Since spiking the samples with E. coli increased not only the area fraction of the cells stained by probe GAM42a but also the total area of all bacteria, the measured E. coli cell area fraction must be corrected to remain directly proportional to the number of added E. coli cells. The corrected area fraction is calculated by the formula

(1)

where A_ec is the measured and A is the corrected E. coli area fraction (in percent). The corrected area fractions were then plotted in a double-logarithmic graph against the E. coli concentration, and a regression line was calculated based on these data points (Fig. 1D). The double-logarithmic transformation was necessary to meet a requirement for linear regression, i.e., an equal variance for all measurements. The regression line was used to calculate the "equivalent E. coli concentrations" from the area fractions of specifically labeled bacterial populations (e.g., ammonia oxidizers stained within activated sludge by FISH; Fig. 1, step 3, and Fig. 1E) in unspiked aliquots of the samples by applying the following equation (Fig. 1, step 4, and Fig. 1F):

(2)

where C_eq is the equivalent E. coli concentration of the bacterial population (in cells per milliliter), A is the measured area fraction of this population, m is the slope, and b is the ordinate intercept of the regression line.

(3)

(4)

Estimation of in situ substrate turnover rates. Average substrate turnover rates of ammonia-oxidizing bacteria were estimated based on the measured cell concentrations. The total number of ammonia-oxidizing cells in the aerated nitrifying basin of the municipal Munich II wastewater treatment plant was calculated by multiplying the number of ammonia-oxidizing cells per milliliter of activated sludge by the reactor volume. The amount of ammonia-nitrogen converted to nitrite (in milligrams per hour) was estimated according to the formula

(5)

where NH₄⁺_t is the transformed ammonia-nitrogen (in milligrams per hour), NH_4i⁺ is the ammonia-nitrogen concentration in the influent (in milligrams per cubic meter), NH_4e⁺ is the ammonia-nitrogen concentration in the effluent (in milligrams per cubic meter), r is the reactor influent rate (7,858 m³ h¹), and 0.9 is a correction factor.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Preparation of the samples and spiking with E. coli A relatively homogeneous distribution of the added E. coli cells within a spiked sample is critical for obtaining an accurate calibration curve. This is particularly important for the measurements of sludge samples which were amended with relatively small numbers of E. coli. Therefore, the area fractions of E. coli cells added to activated sludge were compared after vigorously vortexing the sludge for 1 min or after homogenizing it with an Ultra-Turrax blender (IKA Labortechnik, Staufen, Germany) treatment (1 min). The sludge was spiked with E. coli either before or after these pretreatments.

Evaluation of the quantification protocol using artificial mixtures of pure cultures. The developed quantification approach was first tested with bacterial pure cultures. For this purpose, cultures of G. asaii (-subclass of Proteobacteria) and C. testosteroni (-subclass of Proteobacteria) were mixed after their cell concentrations had been determined with a Neubauer cell counting chamber. The final cell concentrations in the mixture were 3.7 × 10⁷ ± 0.5 × 10⁷ cells ml¹ for G. asaii and 6.3 × 10⁷ ± 0.2 × 10⁷ cells ml¹ for C. testosteroni (all the confidence limits indicate 95% confidence intervals).

View larger version (13K): [in this window] [in a new window]   FIG. 2.   Calibration curves used to convert cell area fractions to equivalent E. coli concentrations for the determination of cell concentrations in pure culture mixtures of C. testosteroni and G. asaii (A), in activated sludge from the Munich II wastewater treatment plant (B), and in the same activated sludge after addition of 1.7 × 108 N. europaea cells per ml (C). The general linear equation of these curves is log Ceq = m × log A + b, where Ceq is the equivalent E. coli concentration, A is the cell area fraction, m is the slope, and b is the ordinate intercept of the line. The values of m are 1.253317 (A), 1.105818 (B), and 1.004006 (C). The values of b are 5.27404 (A), 6.429176 (B), and 6.83411 (C). Error bars which are smaller than the marker symbols are not shown.

1

1

Quantification of ammonia-oxidizing bacteria in activated sludge. The number of autochthonous ammonia-oxidizing bacteria was determined in a nitrifying activated sludge from the Munich II wastewater treatment plant. An equimolar mixture of probes Nso1225 and NEU was used for the in situ detection of ammonia oxidizers of the -subclass of Proteobacteria. Probe Nso1225 targets all recognized ammonia oxidizers of the -subclass with the exception of Nitrosococcus mobilis (28). The single central mismatch of N. mobilis is discriminative under stringent conditions, so probe NEU was used in addition. This probe targets N. mobilis and other not yet described ammonia oxidizers from activated sludge which are also not detected by probe Nso1225 (unpublished results).

1

1

1

Estimated activity of ammonia-oxidizing bacteria in activated sludge. The average rate of ammonia oxidation per autochthonous ammonia-oxidizer cell within the activated sludge was calculated based on the determined concentration of autochthonous ammonia oxidizers. As described above, 9.8 × 10⁷ ± 1.9 × 10⁷ ammonia-oxidizer cells ml¹ were found in the activated sludge. Thus, with a total reactor volume of 27,144 m³ the total amount of ammonia oxidizers in the reactor was 2.7 × 10¹⁸ ± 0.5 × 10¹⁸ cells.

1

1

1

1

1

1

1

1

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Concentrations of ammonia nitrogen in the influent () and the effluent () of the nitrification basin of the Munich II wastewater treatment plant measured during the last 2 weeks before sampling.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Cell quantification methods suitable for microbial ecology should provide precise results for environmental samples containing bacteria which are not homogeneously distributed. Furthermore, they should be independent of cultivation, as most bacteria in natural or engineered systems have hitherto not been isolated (5, 34). A generally applicable method should allow specific enumeration of both single species and higher phylogenetic taxa (e.g., genera or subclasses). New protocols must be tested with regard to these requirements by use of suitable model systems.

In this study, mixtures of pure cultures and an activated sludge sample were used, because these model systems have different advantages when evaluating a new quantification method. The selected pure cultures consisted of uniformly shaped cells, which can easily be counted in a counting chamber to verify the results obtained with the new quantification protocol. On the contrary, activated sludge contains numerous different cultivated and uncultivated prokaryotic species (2, 7, 33) with different morphologies and abundance and thus represents a challenge for any quantification method.

Applicability and accuracy of the newly developed quantification method. The new quantification method was successfully applied to measure cell concentrations of specific populations in bacterial pure culture mixtures as well as in activated sludge. The results obtained were not expressed as relative abundance based on a reference value such as total cell counts, but as absolute cell numbers per volumetric unit. Since the developed quantification procedure is based on FISH with rRNA-directed oligonucleotide probes, it can be used in any environment that is amenable to FISH analysis. It makes no difference whether the quantified organisms grow as single cells or in dense aggregates as long as individual cells can be resolved microscopically. However, the accuracy of the results obtained with this method depends on a uniform cell size of the target population (see below). Size variations within a target population can be caused by polymorphism of a single species or if probes with a broader specificity were applied for detection of morphologically different bacterial taxa. It should also be noted that the accuracy of this method depends on a relatively homogeneous distribution of reference cells (used for spiking) within the environmental sample. For the analyzed activated sludge sample, this was easily achieved by adding the E. coli cells to the sample followed by a short mixing step. Since composition and density of aggregates or flocs might vary between different environments, special pretreatment (e.g., homogenization) of other samples may be necessary to ensure optimal dispersal of the cells used for spiking.

1

1

1

1